Introspection and consciousness 9780199744794, 0199744793

Table of contents :
Introspection and consciousness: an overview / Declan Smithies and Daniel Stoljar --
1. Introspection, what? / Eric Schwitzgebel --
2. Awareness and authority: skeptical doubts about self-knowledge / Fred Dretske --
3. Knowledge of perception / Daniel Stoljar --
4. Renewed acquaintance / Brie Gertler --
5. On the phenomenology of introspection / Charles Stewart --
6. The epistemology of introspection / Ernest Sosa --
7. Knowing what I see / Alex Byrne --
8. Self-knowledge, 'transparency', and the forms of activity / Richard Moran --
9. Self-intimation and second-order belief / Sydney Shoemaker --
10. A simple theory of introspection / Declan Smithies --
11. Judgment as a guide to belief / Nicolas Silins --
12. Discrimination and self-knowledge / Patrick Greenough --
13. Introspection, explanation, and perceptual experience: resisting metaphysical disjunctivism / Aaron Zimmerman --
14. Mind-independence and visual phenomenology / Maja Spener --
15. Introspection about phenomenal consciousness: running the gamut from infallibility to impotence / Terry Horgan.

Citation preview

Introspection and Consciousness

Philosophy of Mind Series Editor: David J. Chalmers, Australian National University Self Expression Owen Flanagan

Gut Reactions Jesse J. Prinz

Deconstructing the Mind Stephen Stich

Phenomenal Concepts and Phenomenal Knowledge Torin Alter, Sven Walter (editors)

The Conscious Mind David J. Chalmers Minds and Bodies Colin McGinn What’s Within? Fiona Cowie The Human Animal Eric T. Olson Dreaming Souls Owen Flanagan Consciousness and Cognition Michael Thau Thinking Without Words José Luis Bermúdez Identifying the Mind U. T. Place (author), George Graham, Elizabeth R.Valentine (editors)

Beyond Reduction Steven Horst What Are We? Eric T. Olson Supersizing the Mind Andy Clark Perception, Hallucination, and Illusion William Fish Cognitive Systems and the Extended Mind Robert D. Rupert The Character of Consciousness David J. Chalmers Perceiving the World Bence Nanay (editor) The Senses Fiona Macpherson (editor)

Purple Haze Joseph Levine

The Contents of Visual Experience Susanna Siegel

Three Faces of Desire Timothy Schroeder

Attention is Cognitive Unison Christopher Mole

A Place for Consciousness Gregg Rosenberg

Consciousness and the Prospects of Physicalism Derk Pereboom

Ignorance and Imagination Daniel Stoljar Simulating Minds Alvin I. Goldman

Introspection and Consciousness Declan Smithies and Daniel Stoljar (editors)

Introspection and Consciousness EDITED BY

Declan Smithies and Daniel Stoljar

Oxford University Press, Inc., publishes works that further Oxford University’s objective of excellence in research, scholarship, and education. Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam

Copyright © 2012 Oxford University Press Published by Oxford University Press, Inc. 198 Madison Avenue, New York, New York 10016 www.oup.com Oxford is a registered trademark of Oxford University Press All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Library of Congress Cataloging-in-Publication Data Introspection and consciousness / edited by Declan Smithies and Daniel Stoljar. p. cm. — (Philosophy of mind) Includes bibliographical references. ISBN 978-0-19-974479-4 (alk. paper) 1. Self-knowledge, Theory of. 2. Consciousness. I. Smithies, Declan. II. Stoljar, Daniel. BD438.5.I58 2012 126—dc23 2011028177

9 8 7 6 5 4 3 2 1 Printed in the United States of America on acid-free paper

CONTENTS

Acknowledgments Contributors Introspection and Consciousness: An Overview

vii ix 3

Declan Smithies and Daniel Stoljar

PART I.

SKEPTICISM ABOUT INTROSPECTION

1. Introspection, What?

29

Eric Schwitzgebel

2. Awareness and Authority: Skeptical Doubts about Self-Knowledge

49

Fred Dretske

3. Knowledge of Perception

65

Daniel Stoljar

PART II.

THEORIES OF INTROSPECTION

4. Renewed Acquaintance

93

Brie Gertler

5. On the Phenomenology of Introspection

129

Charles Siewert

6. The Epistemology of Introspection

169

Ernest Sosa

7. Knowing What I See

183

Alex Byrne

8. Self-Knowledge, ‘Transparency’, and the Forms of Activity Richard Moran

211

vi

Contents

PART III.

CONSTITUTIVISM

9. Self-Intimation and Second-Order Belief

239

Sydney Shoemaker

10. A Simple Theory of Introspection

259

Declan Smithies

11. Judgment as a Guide to Belief

295

Nicholas Silins

12. Discrimination and Self-Knowledge

329

Patrick Greenough

PART IV.

INTROSPECTION AND THE NATURE OF EXPERIENCE

13. Introspection, Explanation, and Perceptual Experience: Resisting Metaphysical Disjunctivism

353

Aaron Zimmerman

14. Mind-Independence and Visual Phenomenology

381

Maja Spener

15. Introspection about Phenomenal Consciousness: Running the Gamut from Infallibility to Impotence

405

Terry Horgan

Index

423

AC K N OW L E D G M E N T S

We are grateful to Peter Ohlin at Oxford University Press and his team for their assistance in all stages of this project. We are also indebted to David Chalmers, the editor of Oxford’s Philosophy of Mind Series, for advice and guidance throughout. Two anonymous referees for Oxford University Press provided extremely detailed and incisive comments on the whole volume. All of the papers in this collection were specially commissioned with one exception, “Self-Intimation and Second Order Belief ” published in Erkenntnis 71 (2009): 35–51. We thank the author, Sydney Shoemaker, and the publisher, Springer, for permission to reprint this paper. A number of papers in this volume were presented at the Introspection and Consciousness Workshop at the Australian National University in October 2008. We thank all the speakers and participants at this event. We are grateful too for the support both this event and the volume itself received from the School of Philosophy, Research School of Social Sciences, Australian National University, and from the Australian Research Council. We are much obliged to Jordan Busse who prepared the manuscript and the index for us. His work was funded by a grant-in-aid from the College of Arts and Humanities at the Ohio State University and we gratefully acknowledge their support. The cover art is a portrait by John Brack of the Australian painter, Fred Williams. The portrait depicts a particular person, but we chose it not for that reason but because we see in it an expression of some of the themes of introspection and consciousness that this volume is dedicated to exploring. Declan Smithies and Daniel Stoljar

This page intentionally left blank

CONTRIBUTORS

Alex Byrne is Professor of Philosophy at the Massachusetts Institute of Technology. Fred Dretske is Senior Research Scholar in Philosophy at Duke University. Brie Gertler is Associate Professor of Philosophy at the University of Virginia. Patrick Greenough is Senior Lecturer in Philosophy at the University of St Andrews and Research Fellow in the Center for Time at the University of Sydney. Terry Horgan is Professor of Philosophy at the University of Arizona. Richard Moran is Brian D. Young Professor of Philosophy at Harvard University. Eric Schwitzgebel is Professor of Philosophy at the University of California, Riverside. Sydney Shoemaker is Emeritus Professor of Philosophy at Cornell University. Charles Siewert is Robert Alan and Kathryn Dunlevie Hayes Professor of Humanities at Rice University. Nicholas Silins is Assistant Professor of Philosophy at Cornell University. Declan Smithies is Assistant Professor of Philosophy at the Ohio State University. Ernest Sosa is Board of Governors Professor of Philosophy at Rutgers University. Maja Spener is a Lecturer at St Catherine’s College, University of Oxford. Daniel Stoljar is Professor of Philosophy at the Australian National University and Future Fellow of the Australian Research Council. Aaron Zimmerman is Associate Professor of Philosophy at the University of California, Santa Barbara.

This page intentionally left blank

Introspection and Consciousness

This page intentionally left blank

Introspection and Consciousness: An Overview Declan Smithies and Daniel Stoljar

Introspection stands at the interface between two major currents in philosophy and related areas of science: on the one hand, there are metaphysical and scientific questions about the nature of consciousness; and on the other hand, there are normative and epistemological questions about the nature of self-knowledge. Introspection seems tied up with consciousness, to the point that some writers define consciousness in terms of introspection; and it is also tied up with self-knowledge, since introspection is the distinctive way in which we come to know about ourselves and, in particular, about our own conscious mental states, processes, and events. Each of these topics—consciousness and self-knowledge—has generated an extensive philosophical literature in its own right. But despite some notable exceptions,1 the relationship between consciousness and self-knowledge has been curiously neglected and remains poorly understood. Indeed, until quite recently, the subfields of philosophy of mind and epistemology were pursued largely in isolation from one another. Recent philosophy of mind has been dominated by metaphysical questions about the nature of consciousness and its place in the physical world, while much less attention has been devoted to questions about the epistemic role of consciousness as a source of knowledge and justified belief. Similarly, recent epistemology has been organized around questions about the nature of knowledge and justified belief, but much of this discussion has developed independently of recent work in philosophy of mind about the nature of consciousness. 1

One important exception is Peacocke (1998).

4

Introspection and Consciousness

The impetus behind this volume is to bring together these two lines of research by exploring the nature of introspection, which lies at the intersection between consciousness and self-knowledge. This volume collects fourteen new essays and one reprinted essay in which the interplay between concerns in epistemology and the philosophy of mind is a major focus. In turn, our goal in this introduction is to set out the main themes of these essays and to map out the overall geography of the terrain.

1. The Difference Thesis Much recent work on introspection is organized around the idea that we have a distinctive way of knowing about ourselves and, in particular, about our own conscious mental states. This idea is motivated in part by reflection on the apparent asymmetry between first-person and third-person perspectives. It seems that each of us has a way of knowing about our own mental states, which cannot be used in knowing about the mental states of others. A key task for any theory of introspection is to give an account of this distinctive way of knowing about our own mental states. Suppose, for example, you are currently thinking about rhubarb. Moreover, you know this—that is, you know that you are currently thinking about rhubarb. This is a piece of self-knowledge that you have: it is knowledge about yourself and, more specifically, about the contents of your conscious stream of thought. But how did you come to have this self-knowledge? Presumably, not by perception: there is no “inner eye” that literally enables you to see that you are thinking about rhubarb. Nor do you observe anything else—say, a particular course of behavior—from which you can infer that you are thinking about rhubarb. So how do you know that you are thinking about rhubarb? “By introspection,” we say. And when we say this we are at least marking the fact that the way in which you know this, whatever it is, is different from other ways of knowing about the world, including sensory perception. The difference thesis, as we might call it, is the thesis that introspection is different from other ways of knowing about the world. This is not to specify the precise ways in which introspection is different from, say, perception. Indeed, there are infinitely many respects in which introspection and perception are both similar and different from one another and so there are infinitely many ways of unpacking the difference thesis. Some will be more interesting than others—for instance, it is not so interesting to learn that introspection and perception are picked out by different words or discussed by different authors. However, a more

Introspection and Consciousness: An Overview

5

interesting thesis is that introspection is different from perception in either psychological respects, or epistemological respects, or both. Psychological versions of the difference thesis claim that introspection is different from other ways of knowing about the world in psychological respects—in other words, it is a psychological process of a different psychological kind. Epistemological versions of the difference thesis, by contrast, claim that introspection is different from other ways of knowing about the world in epistemological respects. In picturesque terms, proponents of the difference thesis claim that the psychology textbook or the epistemology textbook needs a separate chapter on introspection, as opposed to a subsection in another chapter about perception, or inference, or testimony. What is the relationship between psychological and epistemological versions of the difference thesis? Some chapters in this volume claim that introspection is both psychologically and epistemologically different from perception. Others claim that introspection is psychologically different from perception, but epistemologically similar insofar as it is subsumed by a more general theory of knowledge or justification. It is more difficult to motivate the thesis that introspection is epistemologically different from perception and yet psychologically similar, so one strategy for undermining the epistemological difference thesis is to argue against the psychological difference thesis. This is Eric Schwitzgebel’s strategy in chapter 1, while Fred Dretske’s strategy in chapter 2 is to mount a more direct attack on the epistemological difference thesis. Thus, the chapters in this volume represent a wide range of diverse perspectives on the relationship between the psychology and the epistemology of introspection. Having contrasted psychological and epistemological versions of the difference thesis, we can go on to draw further distinctions. Among psychological versions of the difference thesis, for example, we can ask whether they are located at the personal or the subpersonal level of psychological explanation. It is one thing to claim that there is a distinctive kind of psychological process involved at the personal level in forming beliefs on the basis of introspection, as opposed to perception, inference, and so on. It is another thing to claim that this is underpinned by a distinctive kind of psychological processing at the subpersonal level—say, by a dedicated information-processing mechanism or module. Foundational questions arise here, as elsewhere, about the relationship between personal and subpersonal levels of psychological explanation and their relevance to claims in epistemology. Among epistemological versions of the difference thesis, we can distinguish between the claim that the epistemology of introspection is peculiar and the claim that it is privileged.2 The epistemic peculiarity thesis says that introspection is 2

This distinction is drawn by Byrne (2005); see also Alston (1971).

6

Introspection and Consciousness

different from other ways of knowing about the world in certain epistemological respects, whereas the epistemic privilege thesis says that introspection is better than other ways of knowing about the world in certain epistemological respects—for instance, it is more reliable, or it is immune from certain kinds of ignorance and error. Logically speaking, these claims are orthogonal, since two ways of knowing might be different, but equally reliable, while a single way of knowing might be reliable in some domains and unreliable in others. So, what is the relationship between the epistemic peculiarity thesis and the epistemic privilege thesis? Once again, this volume represents a broad spectrum of views. Some claim that introspection is both peculiar and privileged, while others claim that it is peculiar but not privileged or privileged but not peculiar. Despite the absence of any emerging consensus, the variety of perspectives represented in this volume makes an important contribution toward deepening our understanding of the full range of possible options and their respective strengths and weaknesses. In what follows, we will survey these options, beginning with a position that we call “skepticism about introspection.”

2. Skepticism about Introspection Skepticism about introspection is the view that there is no subject matter for a theory of introspection because there is no distinctive way in which we know about our own minds. Skepticism about introspection is a denial of the difference thesis in all its forms: on this view, we do not have an introspective way of knowing about our own minds that is either different from or better than other ways of knowing about the world. For instance, Gilbert Ryle (1949) famously argued that there is no relevant epistemic or psychological difference between first-person and third-person perspectives, since the way in which we know about our own minds is just the same as the way in which we know about the minds of others—that is, by inference from observation of behavior. Thus, he writes, “Knowledge of what there is to be known about other people is restored to approximate parity with self-knowledge. The sorts of things that I can find out about myself are the same as the sorts of things I can find out about other people, and the methods of finding them out are much the same” (1949, 155). Ryle’s theory is usually rejected on the grounds that it cannot explain the self-knowledge that we ordinarily take ourselves to have. For instance, you know that you are currently thinking about rhubarb, but you cannot know this by inference from observation of your behavior, since your thought exerts no causal impact on your behavior—or so we may suppose. You may know about your behavioral

Introspection and Consciousness: An Overview

7

dispositions—say, your disposition to accept a bet that pays you if you are thinking about rhubarb—but you know these facts on the basis of your knowledge of what you are thinking, and not vice versa. There is a useful contrast to be drawn here between modest and extreme versions of skepticism about introspection. The extreme skeptical position denies that we have the self-knowledge that we ordinarily take ourselves to have. The modest skeptical position, by contrast, concedes that we have this self-knowledge, but denies that we have it in the way that we ordinarily take ourselves to have it. The chapters by Eric Schwitzgebel and Fred Dretske represent two quite different ways of developing the modest skeptical position by appealing, respectively, to considerations in psychology and epistemology.

2.1. Psychology Schwitzgebel’s chapter aims to undermine the assumption that there is a unique and distinctive way of knowing about our own minds, which is picked out by the term ‘introspection’. He argues on the basis of detailed consideration of examples that—both between cases and within a single case—there is a range of different processes involved in the generation of what we would ordinarily classify as introspective judgments. Thus, he argues for a kind of introspective pluralism, according to which, “Introspection is not a single process but a plurality of processes.” It is useful to consider Schwitzgebel’s proposal in light of Jerry Fodor’s (1983) distinction between modules and the central system. On some views, introspection involves the operation of a dedicated module that is domain specific and informationally encapsulated—in other words, it processes a limited domain of information in a way that draws on a limited subset of the background information represented in the total system in which it is embedded.3 On Schwitzgebel’s view, by contrast, introspection is an aspect of the central system—that is to say, the processing that yields introspective judgment is not limited to certain kinds of information, but can draw on any relevant information that is represented in the system. Schwitzgebel makes this point vivid by proposing that the boxology of introspection is “a cognitive confluence of crazy spaghetti.” If there is no dedicated module for introspection, then introspection is no different at the level of its subpersonal underpinnings from other ways of knowing about the world, such as inference to the best explanation. Moreover, Schwitzgebel’s pluralism is meant to undermine the idea that introspection is fundamentally different from other ways of knowing about the world at the level of personal

3 For instance, on some views, introspection involves the operation of a relatively isolated mechanism of self-scanning (Armstrong 1968) or self-monitoring (Nichols and Stich 2003).

8

Introspection and Consciousness

psychology. He does offer a personal level characterization of introspection as, “the dedication of central cognitive resources, or attention, to the task of arriving at a judgment about one’s current, or very recently past, conscious experience, using or attempting to use some capacities that are unique to the first-person case . . . with the aim or intention that one’s judgment reflect some relatively direct sensitivity to the target state.” And yet his pluralism consists in the claim that this abstract characterization is multiply realized by the operation of many different psychological processes at the personal level, including perception, inference, and so on. On this view, there is no sharp contrast to be drawn between the way in which we know our own minds and other ways of knowing about the world. Schwitzgebel’s pluralism appears to conflict with the difference thesis in a number of its forms, including the claim that introspection is epistemologically different from, and better than, other ways of knowing about the world.Therefore, proponents of the epistemological difference thesis are faced with an important challenge: either to defend the psychological difference thesis against Schwitzgebel’s pluralism or else to explain why it is consistent with the epistemological difference thesis.

2.2. Epistemology In chapter 2, Dretske outlines a position that he calls conciliatory skepticism. Dretske’s skepticism is conciliatory in two directions. First, he concedes that we know our own minds.What he is skeptical about is an epistemological version of the difference thesis, according to which the way in which we know our own minds is different from the way in which we know the minds of others. What is in question, then, is not our possession of self-knowledge, but rather the source of our self-knowledge. Dretske’s skepticism is conciliatory in another way too, since he allows that one has a distinctive way of knowing what one thinks, but not the fact that one thinks it. Dretske claims that one has a distinctive way of knowing what one thinks by virtue of one’s acquaintance with the propositions that are the contents of one’s thoughts. Moreover, he concedes that if one knows what one thinks, then one knows that one thinks it. However, he denies that if one knows what one thinks in a certain way, then one knows that one thinks it in the very same way.4 The driving thought behind Dretske’s skepticism is that there is nothing in my acquaintance with the propositions that are the contents of my thoughts that tells me that I am thinking them. As he writes elsewhere, “There is nothing you are aware of, external

4 What is at issue here is not the closure of knowledge under known entailment, but rather the transmission of warrant across known entailment.

Introspection and Consciousness: An Overview

9

or internal, that tells you that, unlike a zombie, you are aware of it. Or, indeed, of anything at all.” In chapter 3, Daniel Stoljar presents a detailed reconstruction of Dretske’s skeptical argument—he calls it ‘the evidence argument’—for the special case of seeing.5 In Stoljar’s version, the argument begins with the premise that if I know that I see my son, then there is an answer to the question, “How do I know that I see my son?” Moreover, an answer to this question cites the evidence on the basis of which I know that I see my son, where my evidence comprises facts or propositions that I am aware of and that count in favor of believing that I see my son.6 The key premise is that when I see my son, there is nothing that I am aware of which counts in favor of believing that I see my son. I am visually aware of facts about my son—say, the fact that he is doing somersaults—but these facts do not count in favor of believing that I see my son. After all, it is obviously invalid to argue from the premise that my son is doing somersaults to the conclusion that I see my son. The conclusion drawn from these premises is that I do not know that I see my son. But this seems intolerable: isn’t it a Moorean fact that I know that I see my son? It is perhaps not immediately obvious how Dretske’s argument is meant to undermine the epistemological difference thesis. After all, what it seems to challenge is my knowledge that I see my son, rather than the source of my knowledge. However, the difference thesis is assumed in motivating the key premise of the argument. If I know by perception that I see my son, then there are facts of which I am perceptually aware that count in favor of believing that I see my son. Similarly, if I know by inference that I see my son, then there are facts of which I am aware from which I can infer that I see my son. But if the difference thesis is true, then it is not by perception or by inference that I know that I see my son. In that case, there are no facts of which I am aware that count in favor of believing that I see my son. In effect, the evidence argument provides a reductio of the difference thesis by showing how it leads to the denial of Moorean facts. In response to this argument, Stoljar defends the difference thesis by appealing to a version of what he calls ‘the foot-stamping response’. If I know that I see my son, then the fact that I see my son is a fact that I am aware of, which entails that I see my son and so counts in favor of believing it. In other words, there is an answer to the question, “How do I know that I see my son?” that satisfies Dretske’s conditions. Dretske anticipates this response and rejects it on the grounds that we need a nontrivial

5

Dretske (2003) provides his own statement of the argument. Compare Pryor’s (2005) discussion of the premise principle, which states: “The only things that can justify a belief that p are other states that assertively represent propositions, and those propositions have to be ones that could be used as premises in an argument for p.” 6

10 Introspection and Consciousness answer if we are to avoid begging the question. However, Stoljar distinguishes evidenceseeking and explanation-seeking answers to the question, “How do I know that I see my son?” He accepts that the explanation-seeking question must have a nontrivial answer but denies that it must satisfy Dretske’s conditions on evidence; on the other hand, he accepts that an answer to the evidence-seeking question must satisfy Dretske’s conditions on evidence but denies that it must be nontrivial. Either way, he concludes, Dretske’s argument fails to undermine the epistemological version of the difference thesis.

3. Theories of Introspection It is a Moorean fact that we know our own minds. However, the challenge is to explain how we know our own minds. Moreover, it explains nothing to say that we know our own minds by introspection, since the nature of introspection is precisely what is at issue. As we have seen, the usual starting point is to say that introspection is a distinctive way of knowing about one’s own mental states, which is different from other ways of knowing about the world. But this leaves open the question: How exactly is introspection different from other ways of knowing about the world? While some of the chapters in this volume develop skeptical arguments against the difference thesis, others attempt to develop theories of introspection that maintain the difference thesis in one form or another. In this section, we survey the various different theories of introspection that are defended in the chapters of this volume.

3.1. Acquaintance and Inner Sense According to inner sense theories, we know our own minds on the basis of a form of perception—namely, inner perception, rather than outer perception. But what exactly does it mean to say introspection is a form of inner perception? Presumably, it just means that there are certain similarities (and also perhaps certain differences) between introspection and ordinary cases of sensory perception. Thus, there are potentially as many different versions of the inner sense theory as there are similarities and differences between introspection and perception. Epistemological versions of the inner sense theory claim that the epistemology of introspection is to be explained on the model of perception. In perception, one’s beliefs about the environment are caused and justified by a conscious experience, which presents or represents the environment. Similarly, according to at least some versions of the inner sense theory, one’s beliefs about one’s own mental states are caused and justified by a conscious experience, which presents or represents one’s

Introspection and Consciousness: An Overview

11

mental states. Theories of this kind are endorsed by Brie Gertler in chapter 4 and Terry Horgan in chapter 15. Horgan, for instance, claims that phenomenal character is self-presenting in the sense that the way it seems to be is the way it is. On this view, there is not only a way that one’s environment perceptually seems to be, but also there is a way that one’s perceptual seemings introspectively seem to be. Horgan insists that these secondorder introspective seemings are not to be identified with second-order beliefs about one’s seemings; rather, they are conscious states that cause and justify one’s secondorder beliefs. According to Horgan, second-order seemings involve “direct acquaintance . . . the state’s appearing a certain way, acquaintance-wise, is constitutive of the state’s actually being that way.” The notion of acquaintance also plays an important role in Gertler’s theory of our introspective knowledge of sensations. The core of her proposal is that subjects can refer to their sensations by using introspective demonstratives. Moreover, these introspective demonstratives, like perceptual demonstratives, are grounded in the way things seem to the subject (in a phenomenal sense, rather than an epistemic sense). However, there is also an important difference between perception and introspection on her view, since the referent of a perceptual demonstrative makes a causal contribution to how things seem, whereas the referent of an introspective demonstrative partly constitutes how things seem. Thus, for Gertler, as for Horgan, it is the constitutive connection between the way things seem and the way things are that is characteristic of introspective acquaintance. In chapter 5, Charles Siewert mounts a case against inner sense theories by arguing that there is no phenomenological basis for a distinction between first-order and second-order sensings or seemings. He concedes that one senses one’s own sensings, just as one laughs one’s laughs and dances one’s dances. However, he argues that one’s second-order sensing is coincident with one’s first-order sensing in the sense that there is no distinction to be drawn between second-order sensing and what is sensed. By contrast, there is a distinction between first-order sensing and what is sensed, which is revealed by the fact that the way in which I sense a tilted penny can change, although I do not sense the penny to change. And yet there is no way to hold fixed my first-order sensing, while changing how I sense my sensing—for instance, by perceptually attending to an object in a certain way, while shifting my introspective attention to my sensing in a way that affects how it is sensed.7 7 Compare Shoemaker (1996, ch.10) for a related discussion of the differences between perceptual attention and introspective attention.

12 Introspection and Consciousness Siewert concludes that the way in which we attend to our own experience is quite different from the way in which we attend to the objects of perception. Still, he maintains, we can attend to our own experience by thinking phenomenaldemonstrative thoughts of the form, the way this object perceptually seems to me. Siewert argues that we cannot think thoughts of this kind unless we have the experiences in question. But he denies that such thoughts are explained by second-order sensings. According to Siewert, my first-order sensing plays a dual role: “It can comprise an act of attending to a visible object, even as it helps constitute a cognitively attentive phenomenal-demonstrative thought about itself.”

3.2. Reliabilism and Virtue Theory Reliabilism, unlike the inner sense theory, does not allocate an essential epistemic role to inner seemings or sensings. Rather, such states play an epistemic role only insofar as they are components of a reliable doxastic mechanism. According to reliabilism, a belief is justified in virtue of the fact that it is formed on the basis of a doxastic mechanism that is reliable in the sense that it tends to yield true beliefs. In particular, a belief is introspectively justified in virtue of the fact that it is formed on the basis of a reliable introspective mechanism. David Armstrong (1968) proposes a reliabilist theory on which “introspection is a self-scanning process in the brain.” He also claims that the process of introspective self-scanning can be understood as a form of inner sense by analogy with sense perception. However, he has an unorthodox view of the nature of sense perception. On his view, perception is simply the acquisition of information, which disposes one to form beliefs about the world. Therefore, inner sense is nothing but the operation of a reliable doxastic mechanism that yields beliefs about one’s own mental states. On Armstrong’s reliabilist theory, unlike the inner sense theory as defined above, there is no essential epistemic role for belief-independent introspective seemings. In chapter 6, Ernest Sosa proposes a virtue theory of introspection, which is usefully compared with Armstrong’s reliabilism. Sosa’s target question is: How do introspective seemings justify introspective beliefs? But despite the appeal to introspective seemings, Sosa is no friend of the inner sense theory. In his terminology, a seeming is an intellectual attraction to assent to a proposition, which may or may not be resisted. As such, seemings are to be contrasted with sensory experiences on the grounds that they are conceptually constituted and can be rationally based in other states—for instance, one’s attraction to assent to the proposition that the Müller-Lyer lines are different lengths is rationally based in one’s sensory experience, which presents or represents them as such.

Introspection and Consciousness: An Overview

13

One key contrast is that Armstrong regards introspection and perception as fundamentally similar, while Sosa claims that there is an important difference—namely, that perceptual seemings are rationally based in sensory experiences with a corresponding content, whereas introspective seemings are not. Nevertheless, Sosa insists that introspective seemings are not without any rational basis. For instance, my current visual experience provides the rational basis for its seeming to me that I am having such an experience. Indeed, Sosa argues that an introspective seeming is justified only if it is based on the very mental state that one seems to be in. Thus, if it seems that I am in pain, but only because I am overly neurotic, then my seeming is not introspectively justified. Sosa also appeals to the problem of the speckled hen in motivating a reliability condition on introspective justification, which states that an introspective seeming is justified only if it manifests a reliable competence to discriminate cases in which one is in a certain kind of mental state from cases in which one is not. There are further differences between Sosa’s virtue theory and reliabilism as it is developed by Armstrong and others. In particular, Sosa’s virtue theory is distinguished from reliabilism by its central appeal to the concept of a competence. This enables a dimension of performance normativity, with categories such as performance success, performance skill or competence, and performance aptness, where the success manifests the competence exercised.8 According to Sosa, then, there is an important psychological difference between perception and introspection in the ways they are rationally based in experience. However, there is also an important epistemological similarity between perception and introspection, since each justifies belief by virtue of the manifestation of a reliable virtue or competence.

3.3. Inferentialism and Transparency In chapter 7, Alex Byrne proposes an inferentialist theory of introspection on which we know our own minds on the basis of inference. He does not endorse Ryle’s claim that we know our own minds by inference to the best explanation from observed data about our own behavior. Nevertheless, the motivation for inferentialism is similar—that is, to demystify our capacity for introspective self-knowledge by explaining it in terms of a more general capacity to make inferences. In Byrne’s terminology, an inferentialist account of introspection is economical, rather than extravagant. Byrne’s inferentialism uses an apparatus of epistemic rules and rule-following. An epistemic rule is a conditional of the form: “If conditions C obtain, believe that p!” 8

For a more detailed discussion of virtue epistemology, see Sosa (2007).

14 Introspection and Consciousness One follows an epistemic rule of this form if and only if one believes that p because one knows (and so believes) that conditions C obtain. In earlier work, Byrne (2005) argued that self-knowledge of one’s beliefs can be acquired by following the rule, bel: bel: If p, believe that you believe that p! However, this prompts a Dretske-style evidential objection: How can following bel yield self-knowledge? After all, the mere fact that p is true neither entails nor makes it probable that I believe that p, so it seems as if bel is a bad rule. And yet Byrne argues that bel is a good rule, since it is self-verifying in the following sense: if one follows it, then one’s resulting second-order belief that one believes that p is true. Byrne’s chapter in this volume extends his inferentialist theory from introspective knowledge of belief to the case of seeing. He argues that self-knowledge of one’s visual perceptions can be acquired by following the rule, see: see: If [. . .x. . .]v and x is an F, then believe that you see an F! This rule licenses inferences from v-propositions—that is, propositions concerning “the sensible qualities of objects in the scene before the eyes”—to propositions about one’s visual perceptions. Again, this prompts a Dretske-style evidential objection: How can following see yield self-knowledge? After all, facts about the sensible qualities of objects do not entail or even make it probable that one sees that those facts obtain. In response, Byrne does not claim that see is self-verifying because one could learn what the v-facts are by reading them in “the—as-yetunwritten—language of vision”; however, he claims that it is practically selfverifying in the sense that, for all ordinary situations, one knows that [. . .x. . .] v only if one sees x.This, he argues, is enough to blunt the force of the evidential objection. Byrne’s inferentialism raises various questions. First, how economical is it? Arguably, it is extravagant insofar as it introduces new inferential rules, like bel and see, as well as new inferential capacities required for following these rules. Indeed, Sydney Shoemaker and Nicholas Silins raise questions about whether following rules like bel and see is properly construed as inference or reasoning at all. In chapter 8, Richard Moran raises a related set of issues about our entitlement to follow these rules. Arguably, the mere fact that a rule is self-verifying is not sufficient to explain our entitlement to follow it: just consider the rule, “If x is composed of H2O, then believe that x is composed of water!” As Moran puts the point, “Following a rule for belief . . . requires from the rule-follower some understanding of, and an endorsement of, the rational connection between the contents mentioned in the rule.”

Introspection and Consciousness: An Overview

15

Moran agrees with Byrne that one can answer the question whether one believes that p by answering the question whether p.9 Thus, he also faces a version of Dretske’s evidential objection in the guise of what he calls ‘the puzzle of transparency’—namely, how can answering the question whether p yield knowledge that one believes that p? Moran argues that we cannot solve this puzzle except by appealing to facts about the nature of rationality—in particular, by appealing to one’s rational entitlement to believe that one believes what one rationally ought to believe. Thus, he writes, “The transition described in Transparency is not an inference from evidence about a particular person, but rather something more like a general presupposition of rational thought, to the effect that, from the first-person point of view, I must take what I believe about something to be the expression of my sense of reasons relating to the content of that belief.” Moran’s solution to puzzle of transparency can be reconstructed as follows. Let us assume that I am rationally entitled to believe (1) that if p is true, then I ought to believe that p. Now let us add Moran’s assumption that I am rationally entitled to believe (2) that if I ought to believe that p, then I believe that p. It follows (by means of a closure step) that I am rationally entitled to believe (3) that if p is true, then I believe that p.Therefore, I am rationally entitled to answer the question whether I believe that p by answering the question whether p, which solves the puzzle of transparency. Despite the elegance of this solution, it raises questions of its own.What explains my rational entitlement to believe that I do in fact believe whatever I ought to believe? It is a highly substantive assumption that I am rationally entitled to believe that my beliefs are as they rationally ought to be. What explains and motivates this assumption about the nature of rationality? Is it a primitive fact about rationality? Or does it follow from more general considerations in the theory of rationality?

3.4. Primitivism One of the issues raised in the debate between Byrne and Moran concerns the extent to which a theory of introspection should be economical, rather than extravagant. It is sometimes claimed that if we know our own minds, then our knowledge is based on either observation, inference, or nothing at all. For instance, Paul Boghossian (1989) appeals to this trilemma in arguing for skepticism about introspective self-knowledge. Indeed, Dretske’s skeptical argument in chapter 2 relies on similar assumptions, since he argues that if we do not know our own minds by observation or inference, then none of the facts that we are aware of provide evidence about our own mental states. However, Christopher Peacocke (1998) claims 9

For classic discussions of the transparency of belief, see Evans (1982) and Moran (2001).

16 Introspection and Consciousness that this trilemma is “spurious.” Instead, Peacocke suggests that mental states provide reasons for their own self-ascription. What is at issue here is whether a theory of introspection should be economical, rather than extravagant, in Byrne’s sense. In chapter 10, Declan Smithies draws a related contrast between reductionism and primitivism about the epistemology of introspection. Reductionist theories attempt to explain the epistemology of introspection by assimilating it to a more general model that applies either to observation or inference or both; reductionism is therefore committed to denying the difference thesis at least in some versions. Inner sense theories, reliabilism, and inferentialism are all, in their different ways, versions of reductionism. Primitivist theories, by contrast, take the view that the epistemology of introspection is primitive, or sui generis, and so cannot be explained by appealing to a more general model that applies elsewhere. In chapter 10, Declan Smithies proposes a version of primitivism, which he calls the simple theory of introspection. According to the simple theory, introspection is a way of knowing that one is in a certain mental state, which one has just by virtue of being in that mental state. What sets the simple theory apart from inner sense theories, reliabilism, and inferentialism is the claim that the mere fact that one is in a certain mental state is sufficient to put one in a position to know that one is in that mental state.There is no further requirement to the effect that one has an inner representation of that mental state, or a reliable introspective mechanism, or a selfverifying inferential capacity for believing that one is in that mental state.

4. Constitutivism The main question at issue in the debate between primitivism and reductionism is the extent to which introspection is epistemologically different from other ways of knowing about the world. A further question is whether introspection is not only different, but also better than other ways of knowing about the world in important epistemological respects. In other words, the question is whether introspection is epistemically privileged, as well as epistemically peculiar. A historically influential idea is that introspection is epistemically privileged in the sense that it is immune from certain kinds of ignorance and error. According to constitutivism, there are limits on the possibility of ignorance and error about one’s own mental states because there is a constitutive, internal, or necessary connection between being in a certain mental state and believing that one is in that mental state. In its simplest version, constitutivism is the following thesis:

Introspection and Consciousness: An Overview

17

There are some mental states, M, such that, necessarily, one is in M if and only if one believes that one is in M. As it stands, this psychological thesis is subject to compelling objections.10 For almost any mental state, there are possible cases in which one’s mental states come apart from one’s beliefs about one’s mental states. Therefore, proponents of constitutivism tend to qualify the thesis in one or the other of two ways. One strategy is to qualify the antecedent by adding further psychological or epistemic conditions that must obtain if one’s mental states are to covary with one’s beliefs about one’s mental states—for instance, one must have the relevant concepts, attention, powers of discrimination, and rationality. Another strategy is to qualify the consequent by replacing the concept of believing that one is in a certain mental state with some other psychological or epistemic notion, such as being disposed to believe, or having justification to believe, that one is in a certain mental state.These and other strategies are discussed in the chapters by Stoljar, Shoemaker, Smithies, and Silins. Any constitutivist thesis of this general form prompts the following questions. First, there is a generalization question: Which mental states are such that one is in those mental states if and only if one believes, or is disposed to believe, or has justification to believe, that one is in those mental states? And second, there is an explanatory question: What explains why some mental states meet this condition, rather than others, or none at all? Let us begin by considering the explanatory question, returning to the generalization question in section 5. The obvious strategies for answering the explanatory question appeal to the nature of psychology, or epistemology, or both. For instance, one strategy is to argue that it is in the essential nature of certain mental states that one is in those mental states if and only if one believes that one is in those mental states. Another strategy is to argue that it is in the essential nature of beliefs, or their contents, that one believes that one is in a certain mental state if and only if one is that mental state. However, the prospects seem dim for an explanation of constitutivism in purely psychological terms. This is because we need to use epistemological concepts in specifying the conditions under which mental states covary with beliefs about those mental states. In particular, we need to impose a rationality condition in order to exclude various forms of irrationality, including self-deception, which consist in a mismatch between one’s mental states and one’s beliefs about one’s mental states. This suggests that the epistemological concept of rationality must bear a significant burden in explaining the necessary connection between mental states and our beliefs about those mental states.

10

See Smithies (this volume) for discussion of specific examples.

18 Introspection and Consciousness In chapter 9, Sydney Shoemaker gives three different arguments for the claim that there is a constitutive connection between believing a proposition and believing that one believes it. Each of these arguments makes essential appeal to the concept of rationality. First, there is an argument from Moore’s paradox: it is rationally incoherent to believe a proposition while disbelieving or withholding belief that one believes it. Second, there is (what Shoemaker calls) the zany argument: if one believes that p, then one has reason to believe a certain normative proposition—namely, that one ought to be guided by the assumption that p in one’s practical and theoretical reasoning— but if one believes this normative proposition, then one believes that one believes that p, since these beliefs are identical in their dispositions. Third, there is the argument that rational belief revision, at least when it is motivated by intentional activity on the part of the subject, requires higher-order beliefs about the contents of one’s beliefs. On this version of constitutivism, it is a requirement of rationality that, for certain mental states, one is in a mental state if and only if one believes that one is in that mental state. Thus, Shoemaker writes, “It is part of being a rational subject that belief that p, together with the possession of the concept of belief and the concept of oneself, brings with it the belief that one believes that p.” Similarly, Stoljar claims that it is a requirement of rationality that one is a self-expert: for many first-person, present-tense mental states, one will know that one is in them if certain background conditions are met. However, these claims about the connection between rationality and self-knowledge raise an obvious concern.Why is it a requirement of rationality that one has self-knowledge of any sort? After all, it is not a requirement of rationality that one has knowledge of many different topics, so why is the topic of the self any different? In chapter 10, Declan Smithies argues that the rational ideal of introspective self-knowledge is best explained as a consequence of access internalism in the theory of justification. If justification is accessible, in the sense that one has justification to believe a proposition if and only if one has justification to believe that one does, then ideal rationality involves omniscience and infallibility about which propositions one has justification to believe. Smithies argues that justification is accessible in this sense only if the mental states that determine which propositions one has justification to believe are introspectively accessible in the sense that one has introspective justification to believe that one is in a mental state if and only if one is in that mental state. Therefore, if justification is accessible, then ideal rationality involves omniscience and infallibility about the mental states that determine which propositions one has justification to believe. Opposing viewpoints are presented by Nicholas Silins and Patrick Greenough. Silins argues in chapter 11 that introspective justification is fallible and hence that introspectively justified beliefs can be false. His topic is the phenomenon of transparency, which is also discussed in the chapters by Byrne, Moran, and Shoemaker.

Introspection and Consciousness: An Overview

19

Silins argues that the question whether one believes that p is transparent to the question whether p in the following sense: If you judge that p, then your judgment that p gives you immediate justification to believe that you believe that p. According to Silins, however, there are cases in which one judges that p, but one’s judgment does not express what one really believes. In such a case, Silins argues, one has immediate introspective justification to believe that one believes that p, although one does not in fact believe it. On this view, there is a strong analogy between the epistemology of introspection and perception, since in each case there is a possibility akin to illusion in which one has immediate justification to believe something false. Greenough’s aim in chapter 12 is to argue that there is no coherent conception of rationality that requires omniscience or infallibility about one’s mental states. In this, he builds on Timothy Williamson’s (2000, ch. 4) argument that no mental states are luminous in the sense that if one is in M, then one is in a position to know that one is in M. Greenough’s arguments are inspired by Williamson’s, but they differ in several important respects. First, Greenough’s arguments rely on weaker assumptions about the limits on our powers of discrimination. And second, they establish stronger conclusions. Greenough argues that there is no core class of mental states, M, which satisfies any of the following conditions: If one is in M, then one is in a position to form a justified belief that one is in M. If one is in M, then one believes that one is in M. If one believes on the basis of introspection that one is in M, then one knows that one is in M. If one believes on the basis of introspection that one is in M, then one is in M. Greenough’s arguments therefore pose an important challenge to any view that endorses a constitutive connection between self-knowledge and rationality.11

5. Introspection and Consciousness So far, we have been discussing questions about the nature of introspection: What is the distinctive way, if any, in which we know about our own minds? Next, we consider a question about the subject matter of introspection: What can we know about our own minds in this distinctive way? 11 For two very different attempts to meet this challenge on behalf of constitutivism, see Greenough (forthcoming) and Smithies (forthcoming).

20

Introspection and Consciousness

On a classical Cartesian conception of the mind, the subject matter of introspection is exactly coextensive with the domain of the mental: A state is mental if and only if one knows by introspection that one is in that state. By common consensus, however, this Cartesian conception of the mind has been refuted by the empirical discovery of unconscious mental states, such as the unconscious beliefs and desires that were central to Freud’s psychoanalytic revolution and the unconscious mental representations that have figured in computational psychology since Chomsky’s cognitive revolution. Since Freudian mental states and Chomskian mental states are neither conscious nor known by introspection, there is an obvious amendment to the classical Cartesian conception of the mind. On a neo-Cartesian conception of the mind, there is a core class of mental states that is known by introspection—namely, one’s conscious mental states: A mental state is conscious if and only if one knows by introspection that one is in that mental state. This neo-Cartesian conception of mind is left intact by the Freudian and Chomskian revolutions. Even if it should be rejected on other grounds, it provides a useful starting point for reflecting on the relationship between introspection and consciousness. If there is a necessary connection between introspection and consciousness, then what explains the existence of this connection? Is the necessary connection to be explained as a consequence of the nature of introspection or the nature of consciousness? In other words, which side of the biconditional has explanatory priority? Can we explain consciousness in terms of introspection? If there are multiple concepts of consciousness, as Ned Block (1997) has argued, then this may depend upon which concept of consciousness is in question. For instance, there seems to be a notion of consciousness on which a mental state is conscious if and only if one is conscious of that mental state in the sense that one knows that one is in that mental state and, moreover, one knows this by introspection, rather than in some other way. So, there is perhaps at least one concept of consciousness—Block calls it ‘monitoring consciousness’—that can be explained in terms of introspection. However, Block insists that this is not the only concept of consciousness. For example, he argues that there is another concept of consciousness—which he calls ‘phenomenal consciousness’—for which a mental state is conscious if and only if there is something it is like for the subject to be in that mental state. Can we explain phenomenal consciousness in terms of introspection? According to higher-order thought theories, at least on one development, a mental state is phenomenally conscious if and only if one knows by introspection that one is in

Introspection and Consciousness: An Overview

21

that mental state.12 Arguably, however, there are possible counterexamples in which one is in a phenomenally conscious mental state, but one does not know that one is in that mental state because one does not have the concepts, or one is not in paying attention, or one misclassifies one’s mental state. One reaction to examples of this kind is to modify the proposal by drawing a distinction between conceptual and nonconceptual forms of introspective awareness.13 The claim is that one can have a nonconceptual form of introspective awareness of one’s mental state, even if one lacks any conceptual form of introspective awareness of the fact that one is in that mental state. According to this proposal, a mental state is phenomenally conscious if and only if one has a nonconceptual form of introspective awareness of that mental state. On some views, one’s introspective awareness of a conscious mental state is constituted by a distinct mental state, whereas on other views, it is constituted by the conscious mental state in question. Either way, one has introspective awareness of one’s conscious mental states, which explains and justifies one’s introspective knowledge that one is in those conscious mental states. On this view, then, introspective knowledge is explained in terms of phenomenal consciousness, rather than vice versa. What are the prospects for explaining introspection in terms of phenomenal consciousness? As we have seen, there are cases in which a mental state is phenomenally conscious and yet the subject lacks introspective knowledge that he is in that mental state. And yet this is consistent with the claim that phenomenal consciousness is necessary but not sufficient to explain introspective knowledge. For instance, one might claim that if certain conditions are met, including conditions on concept possession, attention, and rationality, then a mental state is known by introspection if and only if it is phenomenally conscious. However, this proposal faces counterexamples to the necessity of phenomenal consciousness, since we have introspective knowledge of standing attitudes, such as beliefs and desires, which are not phenomenally conscious states. One option here is a qualified version of Ryle’s view that our ways of knowing our own beliefs and desires, unlike our own phenomenally conscious states, are much the same as our ways of knowing those of others. In this vein, Schwitzgebel writes,“If an attitude is not consciously experienced, then it seems—just as a matter of empirical fact?—that we can learn about it only relatively indirectly, using roughly the same variety of tools we use to learn about other people’s attitudes.”

12

See Rosenthal (1997) for a classic version of the higher-order order thought theory of consciousness. 13 See Horgan and Kriegel (2007) for a version of this maneuver. See also Zimmerman (this volume) for a detailed discussion of the distinction between conceptual and nonconceptual forms of introspective awareness.

22

Introspection and Consciousness

On the other hand, many chapters in this volume, including those by Moran, Shoemaker, Silins, and Smithies, develop the opposing view that we do have a distinctive way of knowing what we ourselves believe, which is different from the way in which we know what others believe. If we have introspective knowledge of what we believe, then this dampens the prospects for explaining introspective knowledge in terms of phenomenal consciousness. After all, the question arises: What do beliefs and phenomenally conscious states have in common in virtue of which they can be known by introspection? “Phenomenal consciousness” is not a plausible answer. But perhaps we can explain introspective knowledge in terms of a different notion of consciousness, which applies to beliefs and phenomenally conscious states alike.Thus, Shoemaker explains our introspective knowledge of beliefs by appealing to the functional property of availability, which is a close relative of Block’s notion of access consciousness, which is distinct both from phenomenal consciousness and from monitoring consciousness. According to Shoemaker’s definition, “a belief is ‘available’ if the subject is ‘poised’ to assent to its content if the question of whether it is true arises, to use it as a premise in her reasoning, and to be guided by it in her behavior.” Shoemaker argues that if one believes that p in a way that is available, then one has reason to believe that one believes that p. Moreover, one might extend Shoemaker’s proposal by arguing that if one is in an available phenomenally conscious state, then one has a reason to believe that one is in that phenomenally conscious state. On this view, it is availability, or access consciousness, rather than phenomenal consciousness, which explains why a mental state is known by introspection. According to Silins, by contrast, phenomenal consciousness plays a crucial role in explaining our introspective knowledge of belief. On Silins’s version of the transparency thesis, if you judge that p, then your judgment provides you with a source of immediate, but fallible justification to believe that you believe that p. Judgments, in Silins’s sense, are conscious in the phenomenal sense that there is something it is like for the subject to make a judgment. Moreover, the fact that judgments are phenomenally conscious states is crucial in explaining why they provide a source of introspective justification for second-order beliefs about one’s own beliefs. Phenomenal consciousness also plays an important role in Smithies’s account of our introspective knowledge of belief. Smithies argues that beliefs are phenomenally individuated in the sense that they are individuated by their disposition to cause phenomenally conscious states of judgment. Moreover, he argues that if a mental state is phenomenally individuated, then one has introspective justification to believe that one is in that mental state if and only if one is in that mental state. On this view, phenomenal individuation is what beliefs have in common with phenomenally conscious states in virtue of which they are known by introspection.

Introspection and Consciousness: An Overview

23

Thus, phenomenal consciousness plays a crucial, but indirect role in explaining our introspective knowledge of belief. By contrast, on Sosa’s virtue epistemology, there is no such essential epistemic role for phenomenal consciousness. On his view, beliefs formed on the basis of reliable forms of blindsight can be justified so long as they manifest a reliable competence. Similarly, beliefs formed on the basis of an inner analogue of blindsight can be justified so long as they manifest a reliable competence. On this view, it is a contingent feature of our psychology, rather than a necessary feature of epistemology, that we are unable to form justified beliefs about our own deeply unconscious mental states. It is widely regarded as a desideratum for a theory of introspection to provide a uniform account of our introspective knowledge of beliefs and phenomenally conscious states. For instance, the aim of Byrne’s chapter is to extend his inferentialist theory of introspection from the case of belief to the case of perceptual experience. Moran, by contrast, resists this kind of extension. His explanation of the transparency of belief is tied to the claim that beliefs are active in the sense that their presence in normal human adults is sensitive to reflective appreciation of the reasons that count in favor of having those beliefs. By contrast, perceptual experiences and bodily sensations are passive in the sense that they are insensitive to rational considerations. According to Moran’s account, then, we should not expect a uniform account of our introspective knowledge of beliefs and other phenomenally conscious mental states.

6. Introspection and the Nature of Experience The nature of introspection is an important philosophical issue in its own right, but theories of introspection also have consequences for a wider range of issues in the philosophy of mind. Several chapters in this volume consider how claims about introspection impose constraints on theories of the nature of phenomenal consciousness, including the nature of visual experience. In chapter 13, Aaron Zimmerman explores the role of introspection in adjudicating between competing theories of the nature of visual experience. According to representationalism, visual experience is a relation to a representational content, which obtains whether or not one’s experience is veridical. According to naïve realism, by contrast, visual experience (when it is veridical) is a relation that has mind-independent objects as constituents. This has the consequence that there is no nondisjunctive kind in common between veridical and hallucinatory instances of visual experience—they are not “species of a single genus.”

24

Introspection and Consciousness

Michael Martin (2006) argues for this disjunctivist conclusion on the grounds that naïve realism is the theory that best articulates how visual experience is revealed to us by introspection. In response, Zimmerman distinguishes three different notions of introspection: a-introspection is a matter of forming beliefs about one’s experience while using any experiments, observations, or inferences that seem relevant; b-introspection is a matter of forming beliefs about one’s experience in a way that does not rely on any further reasoning; and c-introspection is a matter of attending to one’s experience without conceptualizing it in one way or another. He proceeds to argue that representationalism is preferable to naïve realism on each of these different notions of introspection. Zimmerman goes on to argue that the best explanation of the introspective indiscernibility between veridical and hallucinatory instances of visual experience is inconsistent with disjunctivism. Martin (2006) anticipates this argument and claims that there is no need to explain introspective indiscernibility, since we can fully characterize the subject’s perspective in terms of the negative epistemic claim that hallucinations are indiscriminable from veridical visual experiences. However, Zimmerman’s response is that this reveals an internal tension in Martin’s view. On the one hand, he needs a substantial characterization of introspection in order to motivate naïve realism in the first place, but on the other hand, he needs a minimalistic characterization in order to block the argument from introspective indiscernibility. Thus, Zimmerman’s chapter raises doubts about whether there is any conception of introspection that can underwrite a stable and well-motivated form of naïve realism. In chapter 14, Maja Spener targets an assumption shared by proponents of representationalism and naïve realism alike—namely, that the phenomenal character of visual experience is exhausted by its presentation or representation of mind-independent objects and properties.This assumption is usually motivated by the thesis that experience is transparent in the sense that if one introspects the phenomenal character of one’s visual experience, then one is aware of nothing but what one’s experience presents or represents. The transparency thesis is widely regarded as an introspective datum about the phenomenal character of visual experience and hence as a constraint on the adequacy of any theory of visual experience. Spener argues that introspective claims about the phenomenal character of experience, including the transparency thesis, cannot serve as pre-theoretical constraints on a theory of the nature of visual experience unless they are “plain” in the sense that they do not rely on independent theoretical commitments. Moreover, she argues that it is not plain that the phenomenal character of visual experience is exhausted by its presentation or representation of mind-independent objects and properties. Instead, she argues, it is a thesis about which theorists may disagree in a way that is driven by their independently held theoretical commitments.Therefore,

Introspection and Consciousness: An Overview

25

she concludes that this introspective claim about the phenomenal character of experience cannot function as a pre-theoretical constraint on the adequacy of any theory of the nature of visual experience. The existence of intractable disagreement about the phenomenal character of experience also sets the agenda for Terry Horgan in chapter 15. His goal is to explain how the existence of such disagreement is compatible with his claim that phenomenal character is self-presenting. Horgan claims that there are some questions whose answers are fully determined by the nature of phenomenal character, although we cannot ascertain these answers simply by attending to the phenomenal character of experience while exercising our competence with phenomenal concepts. The question is: Why not? Horgan argues that answering such questions requires a greater degree of skill in the application of phenomenal concepts than is required for competence with those concepts. In particular, competence with phenomenal concepts is a matter of being able to apply them in specific cases, whereas answering questions about the satisfaction conditions of phenomenal states requires more general cognitive skills, including the ability to evaluate general hypotheses on the basis of extrapolation from particular cases.Thus, Horgan concludes that the existence of intractable disagreement about phenomenal character fails to undermine his claim that phenomenal character is self-presenting. The chapters in this section illustrate how debates in the philosophy of mind about the nature of phenomenal experience are essentially tied up with questions about the nature of introspection. More generally, the chapters in this volume considered as a whole illustrate the way in which questions about the nature of introspection are central to a wide range of fundamental questions in epistemology and the philosophy of mind.

References Alston, William. 1971. “Varieties of Privileged Access.” American Philosophical Quarterly 8: 223–241. Armstrong, David. 1968. A Materialist Theory of the Mind. London: Routledge. Block, Ned. 1997. “On a Confusion about a Function of Consciousness.” In The Nature of Consciousness, edited by N. Block, O. Flanagan, and G. Guzeldere. Cambridge, Mass.: MIT Press. Boghossian, Paul. 1989. “Content and Self-Knowledge.” Philosophical Topics 17: 5–26. Byrne, Alex. 2005. “Introspection.” Philosophical Topics 33: 79–104. Dretske, Fred. 2003. “How Do You Know You Are Not a Zombie?” In Privileged Access: Philosophical Accounts of Self-Knowledge, edited by B. Gertler. Burlington, Vt.: Ashgate Publishing.

26

Introspection and Consciousness

Evans, Gareth. 1982. The Varieties of Reference. Oxford: Oxford University Press. Fodor, Jerry. 1983. The Modularity of Mind: An Essay on Faculty Psychology. Cambridge, Mass.: MIT Press. Greenough, Patrick. Forthcoming. “How to Be a Neo-Cartesian.” Horgan, Terry, and Uriah Kriegel. 2007. “Phenomenal Epistemology: What Is Consciousness That We May Know It So Well?” Philosophical Issues 17: 123–144. Martin, Michael. 2006. “On Being Alienated.” In Perceptual Experience, edited by T. Gendler and J. Hawthorne. Oxford: Oxford University Press. Moran, Richard. 2001. Authority and Estrangement: An Essay on Self-Knowledge. Cambridge, Mass.: Harvard University Press. Nichols, Shaun, and Stephen Stich. 2003. Mindreading: An Integrated Account of Pretense, Self-Awareness, and Understanding of Other Minds. Oxford: Oxford University Press. Peacocke, Christopher. 1998. “Conscious Attitudes, Attention, and Self-Knowledge.” In Knowing Our Own Minds, edited by C. Wright, B. Smith, and C. Macdonald. Oxford: Oxford University Press. Pryor, James. 2005. “There Is Immediate Justification.” In Contemporary Debates in Epistemology, edited by M. Steup and E. Sosa. Oxford: Blackwell. Rosenthal, David. 1997. “A Theory of Consciousness.” In The Nature of Consciousness, edited by N. Block, O. Flanagan, and G. Guzeldere. Cambridge, Mass.: MIT Press. Ryle, Gilbert. 1949. The Concept of Mind. Chicago: University of Chicago Press. Shoemaker, Sydney. 1996. The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press. Smithies, Declan. Forthcoming. “Mentalism and Epistemic Transparency.” Australasian Journal of Philosophy. Sosa, Ernest. 2007. A Virtue Epistemology: Apt Belief and Reflective Knowledge. Oxford: Oxford University Press. Williamson, Timothy. 2000. Knowledge and Its Limits. Oxford: Oxford University Press.

Part I Skepticism about Introspection

This page intentionally left blank

1 Introspection, What? Eric Schwitzgebel

1. Thesis and Alternative Views My thesis is: introspection is not a single process but a plurality of processes. It’s a plurality both within and between cases: most individual introspective judgments arise from a plurality of processes (that’s the within-case claim), and the collection of processes issuing in introspective judgments differs from case to case (that’s the between-case claim). Introspection is not the operation of a single cognitive mechanism or small collection of mechanisms. Introspective judgments arise from a shifting confluence of many processes, recruited opportunistically. The following analogy might be helpful. Suppose you’re at a psychology conference or a high school science fair and you’re trying to quickly take in a poster. You are not equipped with a dedicated faculty of poster-taking-in. Rather, you opportunistically deploy a variety of processes with the aim of getting the gist of the poster: you look at the poster—or perhaps only listen to a recital of portions of it, if you’re in the mood or visually impaired—you attend to what the poster’s author is saying about it; you follow out implications, charitably rejecting some interpretations of the poster’s content as too obviously foolish; you think about what it makes sense to claim given the social and scientific context and other work by the author or the author’s advisor, if you know any; you pose questions and assess the author’s responses both for overt content and for emotional flavor. Although the cognitive systems involved range widely and are not dedicated just to taking in posters, not just any activity counts as taking in a poster—one’s judgments about the poster

30

Skepticism about Introspection

must aim to reflect a certain kind of sensitivity to its contents. Likewise for introspection, I will suggest: the cognitive activities range widely and vary between cases—that is the main claim I will defend—and yet, as I will suggest near the end of this essay, it wouldn’t be natural to call a judgment introspective if it weren’t formed with the aim or intention of reflecting a certain kind of sensitivity to the target mental state. As far as I can tell, no previous philosopher or psychologist has defended both within-case and between-case pluralism about introspection. Although defenders of the view that introspection (or “reflection” or “inner sense”) resembles sensory perception of the outside world could have developed this view in a pluralist direction, historically they have not done so (e.g., Locke [1690] 1975; Kant [1781, 1787] 1997; Wundt 1888, 1896–1897). Some philosophers and psychologists have distinguished between two types of processes that can lead to introspective or quasiintrospective judgments, but two is not many, and often the point of such distinctions is to isolate a unitary target process of interest (e.g., Brentano [1874] 1973; Wundt 1888; Russell 1912; Nichols and Stich 2003). Contemporary philosophers tend to adopt one of two perspectives on introspection and self-knowledge. One approach characterizes introspection (or “self-awareness”) as the operation of a mental self-scanning or self-monitoring process (e.g., Armstrong 1968; Lycan 1996; Nichols and Stich 2003; Goldman 2006). Typically, this process is characterized as involving a single fairly simple detection or monitoring mechanism or family of closely related, simple monitoring mechanisms. Small complexities can plausibly be added: Nichols and Stich (2003) couple the monitoring mechanism with a second, sometimes-competing, sometimes-cooperating mechanism, involving the application of a general “theory of mind”; Goldman (2006) couples monitoring with a capacity to “redeploy” representational contents and to translate representations from one type of mental code into another. But the addition of such complexities doesn’t constitute broad pluralism. A second approach emphasizes mechanisms or procedures other than selfmonitoring as the primary ground of self-knowledge or of the privileged selfascription of mental states. Broadly speaking, these approaches fall into five classes: • Self-fulfillment: Self-ascriptions might involve embedding a target content within a self-ascriptive content in a self-fulfilling way. For example, “I am thinking of a banana” might be automatically true because the thought that I am thinking of a banana contains within it the thought of a banana (see Burge 1988, 1996; Heil 1988; Gertler 2001, this volume; Papineau 2002; Chalmers 2003; Horgan and Kriegel 2007; Horgan, this volume; and for a somewhat different version,

Introspection,What?

31

Shoemaker 1996, this volume; arguably this maneuver goes back to Descartes [1641] 1984). • Self-shaping: Self-ascriptions might tend to be true because in making a selfascription one is committing to a certain way of thinking or acting, presently or in the future, in accord with one’s self-ascription. For example, the selfascriptive thought “I hate churros” might be casually influential in creating or sustaining hatred of churros (see Moran 2001; McGeer and Pettit 2002; McGeer 2008). • Self-expression: Self-ascriptions like “that hurts!” or “I don’t wanna” might essentially be complicated ways of wincing or frowning, or of saying “ow!” or “that stinks!”—that is, they might be self-ascriptive linguistic variants of ordinary spontaneous expressions, which require no prior self-scanning (see Wittgenstein [1953] 1968; Bar-On 2004; Gordon 2007). • Direct inference: Self-ascriptions might be derived inferentially, or quasi-inferentially, directly from judgments about the outside world, requiring no more introspective self-scanning than does ordinary inference. For example, from p, just as one might straightaway conclude p or q, without introspecting one’s judgment that p (presumably not all inference requires introspection), similarly one might straightaway conclude I believe that p (see Dretske 1995;Tye 2000; Byrne 2005, this volume). • Theoretical inference: Self-ascriptions might be largely grounded in observations of one’s own behavior, combined with the theories of folk psychology. For example, from the fact that you often drive across town to get Thai food, you might infer that you like Thai food (see Bem 1967; Nisbett and Wilson 1977; Gopnik 1993a, b). For a fuller review of these approaches to self-knowledge, see Schwitzgebel 2010. Advocates of such views tend not to embrace broad pluralism, though they may emphasize a single competing process. Prinz (2004, 2007) and Hill (2009, 2011) both defend explicitly pluralist views of introspection, but their pluralism appears to be only between-case: although they suggest that many different kinds of cognitive mechanism can yield introspective knowledge, they tend to portray individual introspective judgments as issuing each from the operation of a single mechanism. (Prinz also seems to regard as “introspective” many processes that most philosophers and psychologists would not ordinarily regard as introspective, such as, apparently, ordinary simple recall; see also my discussion of Hill in Schwitzgebel 2011b.) The blind men have each, it seems to me, nicely described a piece or two of the elephant. But none have adequately displayed the pieces’ integration into a messy, moving organism.

32

Skepticism about Introspection

2. Examples 2.1. Visual Experience I look out the window and reach the judgment not only that there’s a tree outside but also that I’m having a visual experience of that tree. I have greenish visual experience of the leaves, and the tree’s spreading branches seem to dwarf the mountain in the background. It has just rained, and in the reemerging sun, the tree sparkles strikingly. Focusing my gaze on the rightmost branches, I notice a fluttering indistinctness in my experience of the left side of the tree. I cross my eyes, thinking it might make the tree double, but instead the tree only swims around my visual field, blurring and fl attening. So, how do I know all this about my visual experience? Let’s begin here: you, standing next to me, seeing me look attentively out the window, might reach some of the same conclusions about me. Minimally, you can safely guess that I’m having greenish visual experience of some sort. After all, you know (a) that I am looking at a green thing in good conditions, and (b) (let’s suppose) that I’m not colorblind. Now of course I, too, know (a) and (b) about myself. Might (a) and (b) be part of my grounds in thinking that I’m having a greenish visual experience? They seem unlikely to be the sole grounds of my judgment—presumably, if I were to have gone suddenly and unexpectedly colorblind (through, say, the secret action of a prankster neuroscientist) I would notice that I’m not in fact having greenish visual experience, despite my looking attentively at what I know to be a green object in plain view. But even if (a) and (b) aren’t the sole grounds of my judgment, it seems reasonable for me to allow my knowledge of outward objects and my own capacities to play some causal and justificatory role in my knowledge of my visual experience. If I know that I’m looking at an evenly painted white surface, I might more naturally reach the judgment that I’m having a visual experience of even whiteness than if I know that I’m looking at a surface with a gradual shift in color. If I know that the burrito I’m biting into has cheese in it, I might be naturally and justifi ably primed to judge that it tastes “cheesy.” If I see you move behind me with a redhot poker and then suddenly I feel a startling touch on my neck, I might swiftly and readily judge that I’m feeling heat and pain, not coolness, even if you have actually touched me with an ice cube. (Maybe I do, in fact, for a split second feel heat and pain, or maybe not; what’s at issue right now isn’t that, but only the contribution my expectations do and should make to my judgment about my experience.) I know the tree has leaves; I know it has just rained; I know what trees in general are like, and what the scene from that window is normally like.

Introspection,What?

33

All this knowledge influences, it seems plausible to suppose, not only my experience but also my expectations about my experience, my readiness to make certain judgments about my experience, and thus those experiential judgments themselves. If my experience is other than expected, I am called up short; I hesitate; it takes a little time, perhaps, and some reconfirmation before I come around, if indeed I do come around. When I try crossing my eyes and don’t receive, or don’t think I receive, the expected double vision of the tree, I react in part by wondering whether I have really succeeded in crossing my eyes; so I try again, wiggling my eyes in various ways, using some combination of motor intentions, proprioceptive feedback, and visual feedback to assess the state of my eyes. I wonder, too, whether there might be a double image of the tree that I’m failing, at least momentarily, to notice (Helmholtz [1856, 1909] 1962; Titchener 1910; Schwitzgebel 2011a, ch. 2). When I hold my eyes fixed on the tree’s rightmost branches and fail to discern the details of the leftmost, my failure in that seemingly outward visual task is part of the basis of my judgment that my visual experience is indistinct away from the point of fixation. My sense of my visual experience is probably shaped, too, by culturally available metaphors, especially painting, photography, and movies, which might draw me toward thinking of my experience as in some way fl at like a painting or as possessing distinctness of shape and color well into the periphery (Noë 2004; Schwitzgebel 2011a, ch. 2). Why am I inclined to think of the tree as dwarfing the mountain? Does this have to do with the projective size each would have on fl at media, or the visual angle subtended? Is there also a sense in which the mountain looks much bigger than the tree? How stable and well grounded and culturally variable are such judgments about the experience of size? The tree sparkles in the sunlight in a way I find striking, and my judgment that this aspect of the scene is striking is partly phenomenological, partly cognitive or aesthetic—a judgment that probably interacts loopingly with my knowledge of the environment, my knowledge of my visual experience of the environment, and my knowledge of other aspects of my reaction to the scene. I hear myself speak, inwardly or outwardly, I shift my gaze, I shift my attention without shifting my gaze, and those processes, too, influence both my visual experience and my apprehension of my visual experience. My judgments about my experience, then, are influenced by at least: my expectations about my experience, my knowledge of the outward environment, my knowledge of what I can and cannot discern, culturally available metaphors and general theories about visual experience, and my knowledge of other aspects of my psychology, in temporally entangled loops. Is there, embedded within this tangle, a distinct, genuinely introspective process, separable at least

34

Skepticism about Introspection

in principle from any nonintrospective influences upon the various emerging judgments? I feel the pull of that idea. The arguments in some of the subsequent essays in this volume appear to turn on the possibility of isolating, in principle, a purely introspective process from amid such noise (especially Gertler, this volume; Sosa, this volume; Zimmerman, this volume). My suggestion in this essay, however, is that it is best to resist treating introspection as distinctive and isolatable. There is no important, cognitively distinct process that is the process of introspection, pure. The view that introspection of visual experience is a process distinct from the processes of visual perception, when that view is combined with a broadly selfmonitoring approach to introspection, appears to invite the following cognitive model: first, there is a process of visual perception and then, afterwards, begins the process of introspecting one’s perceptual experience. Maybe the first process, the perceptual process, continues while the introspective process works at a delay upon its results, always a stage or more behind. However, I doubt this is the best way to conceive of the cognitive processes involved in my example of looking at the tree. It’s more useful, I suspect, to treat the ordinary perceptual processes of vision in that case as part of, or as overlapping with, the introspective processes that shape my judgments about my experience. My visually obtained and constantly updated knowledge of the objects around me is a crucial part of the cause and grounds of my judgments about my visual experience of those same objects. So is the process of trying and failing to visually discern properties of the world. If perception is a complex looping process involving activities of the body such as the movement of fingers and eyes (e.g., Hurley 1998; Noë 2004), so too, I suggest, is introspection in the example above: My activity of holding my eyes still and attempting to discern the shape of the leaves in the periphery, my activity of trying to determine if I have successfully crossed my eyes, my looking around, my recruitment of general knowledge and knowledge specific to the situation, are all part of a multifaceted project that it is artificial to try to divide into introspective and nonintrospective pieces. Introspection can be a bodily activity. Here is another phenomenon that strains against the idea that introspection is a cognitively distinct process sharply separable from the processes of outward perception: judgments about sensory experience can easily collapse into judgments about the outside world with no crisp border between; and the two sorts of judgments, in such cases, are often seemingly driven by virtually identical cognitive processes. So, for example, if asked, for each of a series of stimuli, to report on one’s visual experience of the color of the stimuli, one might first say “green,” then “red,” then “green again,” with the explicit intention of reporting

Introspection,What?

35

only on one’s visual experience, that is, on a piece of phenomenology rather than on properties of the outward stimuli. But after settling into the monotony of the task, it is quite natural to slip absentmindedly into expressing, instead, one’s judgment about the outward stimuli themselves—the colors of the material objects— especially if there’s no reason to doubt that one’s perception is veridical. Such slipping was my frequent experience in reading Titchener’s famous manual of introspective training and attempting to replicate some of his exercises; and Titchener felt it necessary to repeatedly warn aspiring introspective experts against such “stimulus error” or “R-error” (Titchener 1901–1905; see also Boring 1921; Schwitzgebel 2005, 2011a, ch. 5). Where one suspects illusion, “is green” and “looks green” express very different judgments; where one does not suspect illusion, they can blur into each other. Despite their different truth conditions, the two sorts of judgment—one about the stimulus object, one about the experience of that object—are often difficult to pull apart psychologically. We generally use the same terms to express both the objectual and the introspective judgment (e.g., “green” for both the property of the object and the property of the experience the object produces in me), and often, it seems, there is no discrete fact of the matter which of the two judgments I am making or whether I am making both simultaneously. We gradually, insensibly traverse the distinction between introspective and nonintrospective judgment. In such cases, introspection might be best regarded as perception with a twist or with a slightly different aim that can be half forgotten. The processes of perception, then, would be part of the process of introspection.

2.2. Emotion I think about what, if anything, I am emotionally experiencing right now. I notice, first, that my lips are pursed, and I relax them; I notice some tension in my chest. But then I think to myself that emotional experience is not, or might not be, entirely bodily. In fact, it seems a little odd that I should leap straight to bodily self-apprehension in thinking about my emotion. Do I usually do this when I reflect on my emotional experience? Some kind of negative affect is present—perhaps I’m tense about writing this essay? A visual image of a blank word-processing screen has come before my mind. But I had been looking forward all day to finally having a chance to write! As I think about the little remaining time to write this essay, I seem to become more unsettled. I am tense, I decide, about the looming deadline. I find my lips pursed again and rub them with my left hand. There is a bit of an odd feeling in my cheeks, but I don’t know if it is associated with the emotion. Being tense about the deadline doesn’t

36 Skepticism about Introspection seem like the only thing that is going on with me emotionally right now—but what more there is I can’t quite put a finger on. I find myself listening to the freeway traffic in the distance, calming myself a bit with eyes closed and head in hands, and then I imagine how I would look to someone viewing me from behind. I think that thinking about this particular introspective task is worsening my mood, making me tenser and maybe almost angry. I would have liked a happier example, one that better displays the sunny disposition and amiable character I believe myself to have. Perhaps, partly, I am distressed at the negativity of this example, and that distress is further reinforcing the negativity. Maybe, too, there is some self-shaping involved: maybe what I was really experiencing was a relatively undifferentiated negativity or tension, induced partly by over-caffeination, and it became concretized as deadline anxiety partly as a result of my entertaining that hypothesis. My introspective—or seemingly introspective—assessment of my emotional state seems to flow, again, from multiple sources, including at least my knowledge of my social environment (I am writing an essay under deadline pressure), proprioceptive knowledge of my body (my lips are pursed, my throat tense), and knowledge of my thoughts and imagery (knowledge which is, in its turn, presumably also at least partly introspective); perhaps, too, there was some relatively direct causal influence from my emotional phenomenology to my judgment about it. (See section 4 for a discussion of whether such direct causal influences might be what introspection really is.) The process was temporally extended, and I noted how things seemed to shift as my mind moved from topic to topic. It’s almost like I was pulling together pieces of a story. Jakob Hohwy (2011) has argued that introspective knowledge, like sensory knowledge, tends to be exploratory: we have initial expectations or hypotheses, and on the basis of them our cognitive systems make predictions about how things will change (or remain the same) over time, across various conditions. To test these hypotheses, we often act to alter those conditions, and we adjust our assessment of the probabilities according to whether our implicit or explicit expectations of change are confirmed or violated. In sensation, we might move the object or our heads, or we might tap on the object. In the present example, I fix my attention upon the hypothesized source of anxiety, the looming deadline, and note what seems to change in me as a result. More subtly, perhaps, I introduce and track proprioceptive changes when I rub my lips. On this view, the motion of my hand against my lips and the act of bringing the deadline vividly to mind are parts of, and not simply preconditions of, the exploratory introspective process. If I pursue the wrong contingencies or draw the wrong conclusions from them, then I have failed not only to set up the preconditions of introspection in the way I would have liked, but I have failed in the

Introspection,What?

37

introspective task itself. For example, I err introspectively if sinusitis is the cause of my cheeks’ discomfort and yet the discomfort’s failure to recede when I turn my mind away from the looming deadline contributes to a false impression that I’m anxious about more than just the deadline. As in the visual case, bodily, perceptual, and cognitive processes that are not intrinsically introspective can become part of or overlap with the introspective process. On the approach advocated in this paragraph, we cannot say that here, on the one hand, are the outward bodily and perceptual processes and here, on the other hand, is the purely introspective process. Shifting the example, my knowledge that I’m afraid of the rattlesnake or (perhaps differently) that I’m feeling afraid of it might derive in part from my general knowledge that rattlesnakes are dangerous and my visual knowledge that one is only three feet away; it might derive in part from my proprioceptive and visual knowledge that I’ve just flinched, from my knowledge that I felt a tingling surge of what I would call adrenaline, from my sense that I have the impulse to run; it might derive in part from my knowledge that I just uttered an expletive, either in inner or outer speech, from an awareness that I’m imagining the snake biting me, from a kind of numb paralysis I feel; I might have an impulse to say to my hiking partner “I’m terrified of that snake,” which I do or do not disinhibit. Uttering those words, in fact, may help make them true—or at least congeal my emotion and give it specific shape. Probably too, there are some low-level neural or cognitive connections that operate in none of these ways but work more directly from my fear to enhance the likelihood that I will judge myself afraid. Note my appeal here to several of the processes described in section 1: self-detection, self-shaping, self-expression, and theoretical inference.

2.3. Imagery In the morning on the way to work, I blasted a tune on the car stereo, Sonic Youth’s “Kool Thing.” Now it’s 2:00 in the afternoon and I notice that that tune has been running through my head. It has been running for at least three seconds, I think, maybe much longer, and not for the first time today. As I reflect, the tune seems to sharpen or become more vivid. It seems that I can choose to emphasize the vocals or the guitars, and I think about the extent to which I can imagine both the vocals and guitars simultaneously. I conclude that I can do so, especially if I nod my head in rhythm and do something that feels like using my mouth and voice to track the lead guitar line (though no noticeable sound issues from my mouth). Imagery is so much under our immediate control that concurrent introspective judgments about it seem bound, in most cases, to be supported to some

38 Skepticism about Introspection extent by self-shaping, that is, by the process of controlling my mental life in such a way that it conforms to my judgments about it (rather than simply the other way around). Imagery judgments might even be self-fulfilling, if the target image can be a part of the self-attributive judgment about it. When I judge that I’m visually imagining my mother’s face or hearing the chorus of “Kool Thing” in my head, I am partly working to make these self-attributions true as I reach them. However, it seems unlikely that I can make just any judgment about my imagistic phenomenology true simply by willing it to be so: if I judge that I am visually imagining the Taj Mahal with every arch and spire simultaneously well defined, or that I am imagining, simultaneously, the vocals, bass line, drums, and both guitars, or that I am visually imagining a triangle that is neither equilateral, isosceles, nor scalene but somehow all and none of these at once (Locke [1690] 1975; contra Berkeley [1710] 1965), I might be wrong. And hopefully if I am wrong, something in me—some influence, direct or indirect, from the imagery experience itself—will lead me to refrain from the attribution, or cancel it, or at least hesitate and feel uncertain. When I’m trying to determine if I can imagine the lead guitar and vocals at the same time, it seems that I am not only creating or sustaining the imagery but also checking to see if I have successfully created it as intended. Self-regulative feedback is integral to bodily action, except in the swiftest ballistic movements; we might think of imagery creation similarly. Of course, self-shaping and self-fulfillment can’t explain knowledge of very recently past imagery, since self-shaping and self-fulfillment are necessarily present or future oriented. And often, too, there is little environmental or inferential basis for judgments about the contents of one’s recently past imagery; nor does it seem that we can directly, and nonintrospectively, self-express past and gone mental states in the way, perhaps, that we can directly and nonintrospectively burst out with a “that hurts!” or “I don’t wanna.” Thus, recently past imagery is a case where relatively direct causal influences from conscious experience to one’s judgments about it are most evident—thus revealing the incompleteness of any account of self-knowledge limited to the five non-self-monitoring procedures mentioned in section 1: self-fulfillment, self-shaping, self-expression, direct inference, and theoretical inference. (I am assuming here that the operation of those five methods is not continuous or very frequent; otherwise, some combination of those five methods plus memory of their outputs might explain self-knowledge of recently past imagery.) These relatively direct influences from immediately past imagery might take a variety of possible forms: the influences might be mediated by short-term memory, or iconic memory, or a looping process; or they might involve fading activation or the normal temporal course of a feed-forward causal

Introspection,What?

39

brain process; or they might reflect a partial temporal overlap between cognitive processes. The empirical question is open, but here as elsewhere I’ll bet on multiplicity. In noticing “Kool Thing” running through my head, it seems likely, then, that I’m partly shaping it as it transpires, to conform with my judgments about it, and partly exhibiting some relatively direct sensitivity to the experience that is thereby created. Plausibly, too, as in the visual experience and emotional cases described above, my judgment about my experience draws upon general knowledge that makes various experiences or features of experiences seem more or less likely. That knowledge might include: what would be a plausible memory image, given what I know about the band’s usual instrumentation, about that style of music, and about that particular song; my opinions about imagery in general (which are likely to be partly culturally conditioned); and my possibly accurate or possibly distorted opinions about my own imagery capacities (see Schwitzgebel 2011a, ch. 3). Perhaps, too, I am apt to burst into song as a way of expressing, and simultaneously concretizing, my knowledge of my imagery experience.

3. The Boxology of Introspection It’s often helpful for cognitive scientists modeling psychological processes to describe the mind’s functional architecture using boxes and arrows, with the boxes indicating various functionally discrete processes or systems and the arrows indicating the causal or functional relationships among those discrete processes or systems. Figure 1.1 on the next page expresses my view of introspection, using the “boxology” of cognitive science. The model in that figure may be contrasted, for example, with the boxological models on pages 162 and 165 of Nichols and Stich 2003, which feature tidy arrows in and out of the Belief Box, through a Monitoring Mechanism, a Percept-to-Belief Mediator, and a Theory of Mind Information store. You might also notice a resemblance between my model in Figure 1.1 and recent boxological models of visual processing, if the latter are squinted at. Three broad considerations favor this boxology of introspection. First, each of the methods of arriving at self-knowledge described in section 1 seems appropriate to some cases, and the various methods appear to have considerable potential to compete or cooperate in individual instances; and, furthermore, as Prinz (2004) and Hill (2009) argue, it seems unlikely even on a pure scanning view that there would be a single type of scanning mechanism for all possible target states. These considerations

40

Skepticism about Introspection

Figure 1.1. The boxology of introspection

suggest substantial between-case pluralism, at least. Second, as I hope the examples of section 2 illustrated, it seems plausible that in many cases of apparently introspective self-knowledge a wide variety of resources and capacities are brought to bear on the judgment, varying both within and between cases. And third, the more neuroscientists discover about the massive interconnection of the brain, the more it seems architecturally likely that, generally speaking, people’s conscious judgments will draw upon a large variety of influences, from the short and direct to the loaded and circuitous. It’s worth noting, perhaps, that similar considerations recommend a similar boxology for other broad, person-level cognitive processes, like memory, visual perception, and decision. I support that generalization of the diagram (and I briefly discuss the complex influences on memory judgments in Hurlburt and Schwitzgebel 2011), with three qualifications. First, I don’t intend to deny entirely the existence of simple or functionally isolated cognitive processes—perhaps some of the processes operating early in the visual stream are approximately functionally isolated, for example. But such processes, if they exist, are unlike introspection, memory, or visual perception; they are not the broad types of cognitive processes recognized by the terms of folk psychology and capable by themselves of

Introspection,What?

41

generating conscious judgments. Second, empirical investigation can weave a bit of order out of the chaos, allowing us to reify features of the swirl into a large variety of interacting subprocesses, partially isolatable, at least as an approximation. This has already occurred to some extent for memory and visual perception, and it may start to occur for introspection as psychology starts more seriously to contemplate its mechanisms. However—and this is my third qualification—there are wellestablished partial deficits of memory and vision that suggest a certain degree of functional separability among subprocesses; there is currently no parallel taxonomy of partial introspective deficits—no clear pattern, for example, of functional double dissociations among introspective sub-processes (pace Nichols and Stich 2003; see Carruthers 2011).

4. Introspection, What? I doubt that we can draw sharp lines through this snarl, cleanly isolating some genuinely introspective process from related, adjoining, and overlapping processes. What we have, or seem to have, is a cognitive confluence of crazy spaghetti, with aspects of self-detection, self-shaping, self-fulfillment, spontaneous expression, priming and association, categorical assumptions, outward perception, memory, inference, hypothesis testing, bodily activity, and who only knows what else, all feeding into our judgments about current states of mind. To attempt to isolate a piece of this confluence as the introspective process—the one true introspective process, though influenced by, interfered with, supported by, launched or halted by, all the others—is, I suggest, like trying to find the one way in which a person makes her parenting decisions, the one cognitive process behind writing a philosophical essay, or (to return to the example from the beginning of the chapter) the one cognitive process of taking in a science poster. The causes, the influences, the considerations, are too rich within most cases and too variable between cases for any but a radically pluralist account to do justice to the phenomena. One might try to go subpersonal: If there is a cognitive subsystem with the task of keeping a bead on happenings in other parts of the mind, then perhaps that is the introspective system, even if it alone is not responsible for the judgments we arrive at? But surely there are many such systems, if there are any: the mind is thoroughly entangled and the different parts, and subparts of those parts, are designed to track and respond to goings-on elsewhere, at the micro-level as well as the macro-level, from relatively early to relatively late stages of processing, often beyond what we would normally consider to be our introspective ken. This

42

Skepticism about Introspection

kind of intersystem tracking seems hardly sufficient for introspection, at least in any ordinary sense of the term. What must be added to such processes to render their operation the operation of introspection proper? I suggest that they must get tangled up with the whole variety of processes that drive person-level conscious judgment. One might attempt some sort of self-fulfilling content embedding story (as briefly described in section 1, a view that seems recently to have gained momentum): introspection involves loading target mental states into judgments about those very states —“I’m thinking about a hedgehog,” “I’m experiencing [this]” (see, e.g., Gertler’s and Horgan’s contributions to this volume). We can reach infallible judgments in this way, perhaps (just as when we say “this sentence refers to itself ” we necessarily speak the truth)—but that very infallibility shows that we have missed our target, or at least the target that I and probably most other people have in mind when we think about introspection: for introspection in practical use is not infallible; we don’t always get it right in our introspective judgments about our emotional states, about the level of detail in our imagery, about the various features of our visual experiences, our pains, our inner speech. Elsewhere, I have argued that we err very often (Schwitzgebel 2011a, ch. 7), but the frequency of error isn’t as much the present issue as the possibility of error. The kind of introspection that matters to human aff airs and to the methods of psychology and consciousness science is the kind of introspection that involves the fallible application of categories in explicit, conscious judgments—and thus involves the wide variety of resources we flexibly bring to bear in reaching such judgments. But it would be a wacky sort of pluralism that counted every cognitive process as introspective, so let me conclude by suggesting some boundaries. The kinds of examples I have offered, and the kinds of cases I think most philosophers have in mind when they discuss introspection, are cases in which we arrive at explicit, conscious judgments about our own current or very recently past mental states. So perhaps we can say that we ought not regard as introspective any confluence of processes that fails to issue in that sort of judgment, or fails at least to be headed in the direction of issuing in that sort of judgment (though it may be cut short or collapse for some reason). And if the proper product of introspection is a conscious judgment, then we can also say, I think, that introspection consumes attention, on the assumption that forming a conscious judgment necessarily consumes attention. I am inclined to recommend the following view: introspection is the dedication of central cognitive resources, or attention, to the task of arriving at a judgment about one’s current, or very recently past, conscious experience, using or

Introspection,What?

43

attempting to use some capacities that are unique to the first-person case (like most of the capacities emphasized in the accounts in section 1), with the aim or intention that one’s judgment reflect some relatively direct sensitivity to the target state. It by no means follows from this characterization that introspection is a single or coherent process or the same set of processes every time. Now of course I can arrive at conscious, explicit judgments about my current or very recently past conscious experience without doing anything like what we would normally consider introspection: for example, I might read the outputs of a neuroimaging machine, apply a general theory about how those outputs relate to consciousness, and (hypothetically at least) arrive at a judgment about my current conscious experience on that basis alone. Thus, the characterization above requires that introspection involves the attempt to use capacities unique to the first-person case and that reflect a relatively direct sensitivity to the target state. Likewise, there seems something odd about calling a judgment introspective if it is entirely a matter of creating the target state in the course of self-ascribing it with no aim or intention that one’s judgment reflect sensitivity to that state. One further consequence of that last condition is that self-attributions that pop to mind unbidden are not introspective—or rather, they are not introspective unless we are liberal about what counts as having the relevant aim or intention. Maybe we should be liberal. I prefer to leave the matter vague, allowing for in-between cases and stronger and weaker senses of “introspection.” I also leave it vague what counts as “relatively direct.” Attempting to specify too precisely the boundaries of introspection would require, I suspect, knifing more sharply through the spaghetti than the phenomena warrant. Is this, then, really a “multiple realization” view of introspection? And if so, is it consequently just a variety of ordinary functionalism? In a way. But here’s the twist: just as the functionalist about pain denies that pain is a single type of physical process, because pain can be variously realized at the physical level, so also would I deny that introspection is a single type of cognitive process, since introspection can be variously realized at the cognitive level. Despite our ability to gesture at a class of cognitive activities we might call “introspective,” no common cognitive core is shared by all and only introspective processes. To make that last point is, I think, just to restate (a modest form of) between-case pluralism about introspection. Within-case pluralism—at least as developed in section 2—adds the further thought that the processes constituting any single introspective event will normally be, in large part, a combination of processes that exist primarily to serve nonintrospective cognitive functions. If this view is functionalism, it isn’t the type of cognitivist functionalism that treats diverse physical processes as nonetheless cognitively unified. Whatever unification there is exists at a higher or

44 Skepticism about Introspection different level of abstraction—perhaps only amid the rather vague abstractions of folk psychology. A final clarification: my characterization of introspection limits the targets of introspection to conscious experiences. Now while the most central and uncontroversial examples of introspection—and all the examples I have used in this essay—take conscious experiences as their targets, philosophers often suggest that introspection can also take another important class of targets, to wit, attitudes, like belief and desire. Can’t we also introspect those? I propose the following: if an attitude is consciously experienced, we can introspect it, and its availability as a target of introspection is already permitted by the characterization above as its stands. On the other hand, if an attitude is not consciously experienced, then it seems—just as a matter of empirical fact?—that we can learn about it only relatively indirectly, using roughly the same variety of tools we use to learn about other people’s attitudes (though supplemented with a more direct knowledge of potentially related conscious states like inner speech, imagery, or emotional experience; see Carruthers 2011; also Ryle 1949; Goldman 1993, 2006; Hill 2009). Thus, it would be misleading to say that we introspect nonconscious attitudes— misleading because it would suggest that we can discover them in part by deploying a capacity or process, or a certain range of capacities and processes, unique to the first-person case. Philosophers have often characterized introspection as fundamentally epistemically superior to perception, cognitively or structurally simpler than perception, and perhaps also prior to perception and more foundational. If the picture I have sketched in this essay is correct, such claims are all false.1

1

For helpful comments and discussion special thanks to Dorit Bar-On,Tim Bayne, Bill Bechtel, Peter Carruthers, Dave Chalmers, Kevin Falvey, Brie Gertler, Chris Hill, Jakob Hohwy, Linus Huang, John Jacobson, Tori McGeer, Rafe McGregor, Russell Pierce, Cati Porter, Pauline Price, Charles Siewert, Declan Smithies, Maja Spener, Daniel Stoljar, John Sutton, and Aaron Zimmerman. Many others, I’m sure, are unjustly forgotten. I also thank audiences at U.C. San Diego, University of Southern California,Toward a Science of Consciousness, U.C. Riverside, Macquarie, Australian National University, University of Bristol, Oxford, and York University England, as well as the various people who have posted comments on my blog.

Introspection,What?

45

References Armstrong, David M. 1968. A Materialist Theory of the Mind. London: Routledge. Bar-On, Dorit. 2004. Speaking My Mind. Oxford: Oxford University Press. Bem, Daryl J. 1967. “Self-Perception: An Alternative Interpretation of Cognitive Dissonance Phenomena.” Psychological Review 74: 183–200. Berkeley, George. [1710] 1965. “A Treatise Concerning the Principles of Human Knowledge.” In Principles, Dialogues, and Philosophical Correspondence, edited by C. M.Turbayne. New York: Macmillan. Boring, E. G. 1921. “The Stimulus Error.” American Journal of Psychology 32: 449–471. Brentano, Franz. [1874] 1973. Psychology from an Empirical Standpoint. Edited by O. Kraus and L. McAlister, translated by A. C. Rancurello et al. London: Routledge. Burge, Tyler. 1988. “Individualism and Self-Knowledge.” Journal of Philosophy 85: 649–663. ———. 1996. “Our Entitlement to Self-Knowledge.” Proceedings of the Aristotelian Society 96: 91–116. Byrne, Alex. 2005. “Introspection.” Philosophical Topics 33: 79–104. Carruthers, Peter. 2011. The Opacity of Mind. Oxford: Oxford University Press. Chalmers, David J. 2003. “The Content and Epistemology of Phenomenal Belief.” In Consciousness: New Philosophical Perspectives, edited by Q. Smith and A. Jokic. Oxford: Oxford University Press. Descartes, René. [1641] 1984. “Meditations on First Philosophy.” In The Philosophical Writings of Descartes, Vol. 2, edited by J. Cottingham, R. Stoothoff , and D. Murdoch. Cambridge: Cambridge University Press. Dretske, Fred. 1995. Naturalizing the Mind. Cambridge, Mass.: MIT Press. Gertler, Brie. 2001. “Introspecting Phenomenal States.” Philosophy and Phenomenological Research 63: 305–328. Goldman, Alvin I. 1993. “The Psychology of Folk Psychology.” Behavioral and Brain Sciences 16: 15–28. ———. 2006. Simulating Minds. Oxford: Oxford University Press. Gopnik, Alison. 1993a. “How We Know Our Minds: The Illusion of First-Person Knowledge of Intentionality.” Behavioral and Brain Sciences 16: 1–14. ———. 1993b. “Psychopsychology.” Consciousness and Cognition 2: 264–280. Gordon, Robert M. 2007. “Ascent Routines for Propositional Attitudes.” Synthese 159: 151–165. Heil, John. 1988. “Privileged Access.” Mind 97: 238–251. Helmholtz, H. [1856, 1909] 1962. Helmholtz’s Treatise on Physiological Optics. Edited by J. P. C. Southall. New York: Dover. Hill, Christopher S. 2009. Consciousness. Cambridge: Cambridge University Press. ———. 2011.“How to Study Introspection.” Journal of Consciousness Studies 18(1): 21–43.

46

Skepticism about Introspection

Hohwy, Jakob. 2011. “Phenomenal Variability and Introspective Reliability.” Mind and Language 26: 261–286. Horgan, Terence, and Uriah Kriegel. 2007. “Phenomenal Epistemology: What Is Consciousness That We May Know It So Well?” Philosophical Issues 17: 123–144. Hurlburt, Russell T., and Eric Schwitzgebel. 2011. “Presuppositions and Background Assumptions.” Journal of Consciousness Studies 18(1): 206–233. Hurley, Susan. 1998. Consciousness in Action. Cambridge, Mass.: Harvard University Press. Kant, Immanuel. [1781, 1787] 1997. Critique of Pure Reason. Edited and translated by P. Guyer and A. W. Wood. Cambridge: Cambridge University Press. Locke, John. [1690] 1975. An Essay Concerning Human Understanding. Oxford: Oxford University Press. Lycan, William G. 1996. Consciousness and Experience. Cambridge, Mass.: MIT Press. McGeer,Victoria. 2008. “The Moral Development of First-Person Authority.” European Journal of Philosophy 16: 81–108. McGeer, Victoria, and Philip Pettit. 2002. “The Self-Regulating Mind.” Language and Communication 22: 281–299. Moran, Richard. 2001. Authority and Estrangement. Princeton: Princeton University Press. Nichols, Shaun, and Stephen P. Stich. 2003. Mindreading. Oxford: Oxford University Press. Nisbett, Richard E., and Timothy D. Wilson. 1977. “Telling More Than We Can Know: Verbal Reports on Mental Processes.” Psychological Review 84: 231–259. Noë, Alva. 2004. Action in Perception. Cambridge, Mass.: MIT Press. Papineau, David. 2002. Thinking about Consciousness. Oxford: Oxford University Press. Prinz, Jesse J. 2004. “The Fractionation of Introspection.” Journal of Consciousness Studies 11(7–8): 40–57. ———. 2007. “Mental Pointing: Phenomenal Knowledge Without Concepts.” Journal of Consciousness Studies 14(9–10): 184–211. Russell, Bertrand. 1912. The Problems of Philosophy. London: Oxford University Press. Ryle, Gilbert. 1949. The Concept of Mind. New York: Barnes and Noble. Schwitzgebel, Eric. 2005. “Difference Tone Training: A Demonstration Adapted from Titchener’s Experimental Psychology (1901–1905), vol. I, part 1, 39–46.” Psyche 11(6). (A link to this web-based demonstration: .) ———. 2010. “Introspection.” Stanford Encyclopedia of Philosophy. Edited by E. M. Zalta. ———. 2011a. Perplexities of Consciousness. Cambridge, Mass.: MIT Press. ———. 2011b. “The Philosophical and Psychological Context of Descriptive Experience Sampling.” Journal of Consciousness Studies 18(1): 288–294. Shoemaker, Sydney. 1996. The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press. Titchener, E. B. 1901–1905. Experimental Psychology. New York: Macmillan.

Introspection,What?

47

———. 1910. A Text-Book of Psychology. New York: Macmillan. Tye, Michael. 2000. Consciousness, Color, and Content. Cambridge, Mass.: MIT Press. Wittgenstein, Ludwig. [1953] 1968. Philosophical Investigations, 3d ed. Translated by G. E. M. Anscombe. New York: Macmillan. Wundt, Wilhelm. 1888. “Selbstbeobachtung und innere Wahrnehmung.” Philosophische Studien 4: 292–309. ———. 1896–1897. Outlines of Psychology. Translated by C. H. Judd. New York: Stechert.

This page intentionally left blank

2 Awareness and Authority: Skeptical Doubts about Self-Knowledge Fred Dretske

I have lately (Dretske 2003a, b, c) been advancing a view about self-knowledge that forty years ago I would have thought implausible. Nonetheless, here I am, once again, promoting it. Why? Not because of a sudden revelation or deep insight into the nature of knowledge or character of mind. No, I defend it because over these years I have developed commitments—all of which seem quite reasonable to me— that lead, irresistibly, to the conclusion that although we have direct awareness of and, therefore, a unique authority about what is in our mind and about what we think, feel, and experience, we enjoy no special access to or authority about the fact that we have a mind—to the fact that we actually have thoughts, feelings, and experiences. So I thought it might be worth describing how I got here, what led me to this bizarre view. Let me be clear, though, about the view I mean to defend. It is not the denial that we know we have thoughts and feelings. It is the denial that we have a way of knowing this not available to others, a kind of awareness of or access to our thoughts and feelings that gives us a special authority about the fact that we have them. You know you have thoughts and feelings, yes, but you know that your spouse, children, friends, and colleagues have them too. And they know you have them. So knowledge is not the issue. It is the source of this knowledge.

50

Skepticism about Introspection

Although I spoke above of thoughts, feelings, and experiences, I will here focus on current conscious thoughts, where a thought (that p) is construed broadly enough to include any propositional attitude having p as its content in which the truth of p is accepted by, but not required for, those having this attitude toward p. So believing and judging (as well as thinking in one of its everyday senses) that p are ways of thinking that p in this enlarged sense. Knowing, remembering, and seeing that p—since requiring the truth of p—are not (though they may involve believing that p). Neither are wanting, hoping, and doubting that p, since the truth of p is not thereby accepted. I believe the argument can be generalized, but the effort to make the general case would entangle me in too many troublesome details. For an attempt to deal with some of these details, see Dretske 2003a.

1. Self-Knowledge The view I am promoting is a modest form of skepticism. Perhaps it should not be called skepticism at all, since it does not deny the existence of self-knowledge. It only challenges a widely accepted view about the source of this knowledge. I, nonetheless, call it conciliatory skepticism. It is conciliatory in its willingness to accept the existence of self-knowledge, but skeptical about a common picture of where this knowledge comes from. To set the problem, then, I begin by conceding that: (1) Those who understand what thinking is, and are thus able to think they think, know they think. This is not a claim I expect debate about. There may be those who (for Wittgensteinian reasons) object to it, but most people, I assume, will find it laughably obvious. So I will not waste time defending it. I merely point out that the claim to self-knowledge is, as it must be, restricted to those who are conceptually developed enough to believe they think. There are those—young children, for instance—who think but who do not yet think they think.1 They do not, therefore, know they think. But those of us who understand what thinking is (well enough anyway to

1 Or so developmental psychologists tell us: see Astington 1993; Baron-Cohen 1995; Bartsch and Wellman 1995; Carpendale and Lewis 2006; Flavell, Miller, and Miller 1993; Flavell 2003; and Gopnik 1993. The developmental transition (to understanding what thinking is) is gradual and indeterminate, but for the sake of brevity I speak of it as occurring somewhere between the ages of three and four years. All that is important for my purposes is that children think before they (are able to) think they think. That much seems incontestable.

Awareness and Authority

51

think we think) know we do. What is at issue, once again, is not that we know it, but how we know it. A word of clarification about (1): it concedes knowledge to those able to think they think. The thought that one thinks is, of course, a very special kind of thought. It is self-verifying. If I think I think I have to be right. My thinking it is so makes it so. But if you think I think, you can be wrong. Your thinking it is so does not make it so. So my thought that I think is infallible in the way that your thought that I think is not. This is clearly true, but the truth of (1) is not to be interpreted as deriving from the self-verifying character of these higher-order thoughts. That is not the explanation of why (1) is true. Thinking you think would be a way of demonstrating (and thus coming to know) you think if you knew you had this self-verifying thought, yes, but this clearly does not represent a way of coming to know you think. It assumes you already know it.2 There may, of course, be those (reliability theorists?) who will argue that it is not the knowledge that we have self-verifying thoughts (that we think) that gives us knowledge that we think. That, obviously, would be circular. It is, rather, the mere existence of these self-verifying thoughts. Since they are guaranteed to be true, they are perfectly reliable—in fact, absolutely infallible—guides to truth. This, and this alone, qualifies the belief that one thinks as knowledge, and thus explains the truth of (1). This is too crude a view of knowledge to spend a lot of time arguing about. Aside from obvious objections involving necessary truths—belief in which (however ill-founded) enjoy a rock solid guarantee—there are contingent truths, belief in which are self-verifying for similar reasons. That does not make them knowledge. The fact that carbon is essential for life and life necessary for belief—thus making a belief that carbon exists self-verifying—would surely not, by itself, make a belief that carbon exists knowledge. If, nonetheless, there are readers who subscribe to this austere theory of knowledge, I will simply declare, up front, that I am looking for something different. Something more satisfying than a belief (even if it is selfverifying) that we have no reason to think we have. I am, that is, looking for reasons to think we think whether or not such reasons are deemed necessary for knowing we think according to certain implausible (or so I think) theories of knowledge.3 I am asking whether it is reasonable to think you think and, if it is, what your reasons are.

2 For the same objection to the self-verification of reflexive thoughts as a source of self-knowledge, see Brian Loar (1994, 64) and the discussion in Bernard Kobes (2003, 203–205). 3 I am a reliabilist about knowledge, but the kind of reliability I embrace is that relating to the reasons, grounds, or evidence one has for the proposition that one believes, not simply reliability (with or without supporting reasons) of one’s belief in the proposition.

52

Skepticism about Introspection

2. Awareness of Current Thoughts We are aware of our currently conscious thoughts in a way that gives us authority about what we think. We may not be infallible. We may sometimes be wrong, sometimes uncertain, and at other times we may have to make up our mind about what exactly we think, but the fallibility, uncertainty, and indeterminacy that characterizes so much of our thinking in no way challenges our authority when we are certain about what we think. When the magician, in the midst of a card trick, asks the young lady to think of a card, the lady, surely, is the authority about what card she is thinking of. The trick falls flat if the magician, by way of explaining his failure to produce the right card, accuses the lady of being mistaken about what card she is thinking of. Sorry! That is not an option. The lady knows in a way the magician does not what card she is thinking of. If he wants to know what card that is, he has to ask her. He cannot tell her. Privileged access to one’s thoughts does not mean one has exclusive access to them, but it is obviously an enormous advantage. I can find out what you think, but most of my ways of finding this out depend on you already knowing what you think. You can tell me what you think. I can read your secret diary. Or I can observe your behavior: surely you would not be taking your umbrella unless you thought it might rain. You, though, have a kind of direct access to what you think that I do not have. My second concession, then, is merely an expression of the widely (universally?) accepted idea that we all enjoy a kind of privileged access to, and a consequent authority about, our currently conscious thoughts.4 (2) One has unique access to one’s currently conscious thoughts, and this access gives one authority on (and is the source of one’s knowledge of ) what one is presently thinking. This, though, sounds like a damaging admission for me to make. It appears to be an admission that the self-knowledge described in (1) has its source in the privileged access described in (2). Isn’t this just the familiar introspectionist’s story—exactly what my conciliatory skeptic denies? No, it is not. (1) and (2) are perfectly consistent with the denial that our unique access to our own thoughts, an access that makes us authoritative about their content, 4 I restrict myself to currently conscious thoughts—the only ones we have much authority about. One clearly does not enjoy such authority on one’s past thoughts, and it seems clear that there are a good many dispositional beliefs one is not conscious of having. Those who are surprised to hear that Los Angeles is east of Reno, Nevada, for instance, betray a tacitly held belief that Los Angeles is west of Reno. I found out I believed Los Angeles was west of Reno when I found out (to my surprise!) that it was not.

Awareness and Authority

53

about what we think, is the (or even a) source for the knowledge that we think. The following section explains why this is so.

3. Forms of Awareness The conciliatory skeptic denies neither (1) nor (2). What he denies is the connection between them. He denies that your unique access to your own thoughts, a form of awareness that makes you authoritative about what you think, is the source of your knowledge that you think. Remember: (3) Awareness of X can give one knowledge of X without being the (or even a) source for one’s knowledge (if one has it) that it is X. One can be aware of (see) a counterfeit $100 bill and thereby come to know in this way, by the way it looks, that it has Ben Franklin’s picture on it, and not be able to learn, at least not in this way, that it is counterfeit. When the forgery is good, knowledge that it is counterfeit must come from a source other than perception of the counterfeit—the mode of awareness that gives you knowledge of the counterfeit. Maybe someone (the person who printed it) tells you it is counterfeit. Or you read the analysis of treasury department experts. If so, then knowledge that it is counterfeit has a different source than perceptual knowledge of the counterfeit. The same is true of things (and this is a lot of things) that fail to reveal to those aware of them (in the mode in which they are aware of them) what they are. Seeing spies and widows is not normally a mode of awareness of these people that tells you what they are. Awareness of an X, whether this is a spy or a thought, is one thing. Awareness (knowledge) that it is an X is quite another. So the kind of privileged awareness of our own thoughts acknowledged in (2), must be carefully distinguished from the knowledge (awareness that they are thoughts) conceded in (1). Wait a minute! Concession (2) does not merely acknowledge a unique access to, and direct awareness of, one’s own thoughts. It concedes that this access gives one special authority about the thoughts. It concedes, furthermore, that this privileged awareness is the source of one’s knowledge of what one thinks. Isn’t this exactly what is being denied? No, it is not. A conciliatory skeptic does not deny that your awareness of your own thoughts gives you authority on and knowledge of them. Of course it does. It makes you an authority on their content, on what you think. It is the source of your knowledge (if you have it; more of this in a moment) of what you are thinking. Just as the mode of awareness that provides you with information about whose picture

54

Skepticism about Introspection

is on the counterfeit bill does not give you the information that it is a counterfeit bill, the mode of awareness that tells you what you are thinking does not reveal that you are thinking it.5 Consider the following analogy. It is a bit of a cartoon, but it makes the essential structure clear when one is dealing (as we are with thought) with awareness of an abstract object (a proposition): The Philosophical Gazette, smitten with my deep insights, arranges to publish everything I write. They give me oral assurances of this, and (let us assume) their word is gold. On the basis of their assurances, then, I know they publish everything I write. I now write p. The Gazette promptly publishes it. Given what I have been told, I know they are publishing p. Yet the way I know what they are publishing is not the way I know they are publishing it. They told me they were publishing everything I write, and it is in virtue of that knowledge that I can tell (come to know) what they are publishing by observing what I write. It is vision (together, of course, with what I already know) that tells me what The Gazette is publishing, but this knowledge presupposes a prior knowledge (obtained earlier via audition) that they are publishing what I write. Self-knowledge, I submit, has the same structure. It is a privileged awareness of one’s own thoughts that gives one authority on what one thinks (thus (2)), but it is not the source of one’s knowledge that one thinks it. That knowledge has a different source. I am not here disputing6 that: (4) One has to know (by some means or other) that one thinks in order to know what one thinks.

5

This may be what Davidson (1988, 664) had in mind when he said in his reply to Burge that Burge’s argument “shows only that one cannot go wrong in identifying the content of an attitude, which is not a reason why one cannot go wrong about the existence of the attitude.” For the same point, see also Gallois (1996, 168). 6 Although, I have disputed it (see Dretske 2003b, c). The basis for the dispute involves the distinction between a referential and an attributive sense of knowing-what. Boër and Lycan (1986, 18) distinguish between a referential and an attributive sense of knowing who in which a person could, in the referential (but not attributive) sense, know who the murderer is without knowing she is a murderer. So too, one might argue, a person might (in the referential sense) know what is on the table without knowing that it is on the table. Generalizing, then, one might go on to argue that a person might know what he thinks (in a referential sense of “what I think”) without knowing it is something he thinks. Such an argument, though, would take me into controversial waters. It raises issues about closure (knowing what you think without knowing you think it). Although I am perfectly willing to abandon closure (see Dretske 2005), I here avoid distracting controversy by accepting (4). That is, I grant, for the sake of argument (and so as not to alienate readers committed to closure) that one cannot know what is  without knowing it is .

Awareness and Authority

55

All I am contesting is that these pieces of knowledge—knowledge of an act’s content (what one thinks) and knowledge of the act itself (that one thinks it)—necessarily have the same source. Knowledge that one thinks p may not have its source in introspection any more than knowledge that The Gazette is publishing p has its source in perception of what I write even though it is perception of what I write that (given what I know) tells me what they are publishing. To better understand this form of awareness, a mode of awareness that makes one an authority on what one thinks without disclosing—and thus without making one an authority on—the fact that one thinks it, it is useful to consider people— small children—who are aware of what they think without knowing they think it. This discussion will also reveal the respects in which such awareness can be the source of one’s knowledge of content, of what one thinks, without being the (or even a) source for one’s knowledge that one thinks it.

4. Unwitting Authority Sarah, a normal three-year-old, does not think she has thoughts. She does not think she does not have them either. She does not understand what thinking is. She nonetheless thinks Daddy is home. That is what she tells Mommy. That is why she runs to open the door when she hears a car pull in the driveway. What Sarah tells her mother is (what else?) what she thinks: that Daddy is home. She does not yet understand, though, that what she tells her mother is what she thinks. Despite this ignorance, her thoughts (and desires) explain her behavior in the same way our thoughts and desires explain ours. The only difference is that Sarah does not yet understand that her behavior can be explained by the fact that, whether or not her father is home, she thinks he is home. We do. If Sarah’s behavior is to be explained by what she thinks even when she does not realize she thinks it, and the behavior in question is a deliberate, purposeful act (Sarah has and is prepared to give reasons for what she does, and these reasons are, in part, what she thinks), there must be a sense in which Sarah is aware of what she thinks—that her father is home—without understanding, without knowing, without being aware, that it is something she thinks. This is the epistemically uncommitted or neutral form of awareness we spoke of earlier in our discussion of (3). Just as you can be aware of my cousin or a counterfeit $100 bill (concrete objects) without realizing it is my cousin or counterfeit, Sarah can be aware of what she thinks (an abstract object) without realizing it is something she thinks. If Sarah takes this “object” to be anything at all, she takes it to be (what philosophers would call) a fact: the fact that her father is home. When it is important to distinguish this

56

Skepticism about Introspection

non-epistemic7 form of awareness from knowledge, from awareness of facts, I will use the term acquaintance. Sarah is aware of, acquainted with, the proposition that her father is home. She may also, of course, be aware of the fact (know) that her father is home. Awareness of the proposition is, if you will, the subjective component of fact-awareness. Her awareness of the fact that her father is home is an awareness of the proposition that he is home when this judgment or belief (in which the truth of p is accepted) is brought about (by the fact) in the appropriate (knowledge-making) way. This talk of propositions as the objects of thought, as things we are aware of, may appear quaint, maybe a little old-fashioned. It certainly is not trouble-free.8 If one deems the troubles too great, one can substitute the words “content” or “what is thought” for the word “proposition” in what follows. This will not change anything. We need some way to describe a person’s mental state when (like Sarah) she is doing something because of what she thinks and who would, moreover, give this (what she thinks) as her reason (both as a justification and explanation) for acting that way. How, after all, is one to describe Sarah’s mental state when she tells her mother that her father is home and runs to the door with a mistaken belief that her father is home? Despite our willingness to describe her as aware of something (that her father is home) when her father is home, Sarah is clearly not (when mistaken) aware that her father is home. He is not home. So that cannot be what she is aware of. Nor is she aware that she thinks he is home. She does not realize that she thinks this. We are left, it seems, with what she thinks, the content of her thought, the proposition that he is home, as what she is conscious of that explains why she behaves and talks this way.9 7 It is important to understand that in calling this a “non-epistemic” form of awareness, I do not mean that the awareness of X occurs in the absence of knowledge that it is X. No, I mean that one can be fully aware of X in this (non-epistemic) way and not know it is X. When I see your cousin, I am non-epistemically aware of him not because I am aware of him without knowing he is your cousin (I might know he is your cousin), but because my awareness of him does not depend on this knowledge. I could see, and thus be aware of, him without knowing he is your cousin. 8 The troubles involve singular thoughts (e.g., Freddy is home, Santa lives at the North Pole), de se thoughts (I spilled the sugar), and indexical thoughts (the treasure is buried here). For useful discussion, see Davies 1998; McLaughlin and Tye 1998; Feit 2008; and Aune 2009. I ignore these problems here with the conviction (hope?) that whatever view one has about the objects of thought one will have to find some way of saying what I here say using the language of propositions. I think the epistemology is independent of the metaphysics. 9 I will doubtless be accused of committing a version of the sense-datum fallacy—inferring that if one seems to be aware of a fact, one must be aware of a seeming-fact (a proposition). I confess to inferring this, but I do not think it is a fallacy. Nor do I think that it is a fallacy to conclude that if it seems to you as though you are seeing a cup, you are aware (not see, but nonetheless aware or conscious of) a seeming-cup. One just has to be very careful not to confuse seeming-cups with cups (or perceptual objects of some kind).

Awareness and Authority

57

We use this proposition to specify what it is like for Sarah at that moment. If qualia are understood to be how things seem or appear to be, then among Sarah’s qualia is to be included what she thinks as she runs to the door. Unlike sensory qualia (e.g., the quality of the sound she hears as a car pulls in the driveway), this aspect of Sarah’s mental life takes the form of a proposition: Daddy is home. It now seems to Sarah as though her father is home. It did not seem that way to her a minute ago. So there has been a change in her overall conscious state when, upon hearing a car in the driveway, she came to think her father was home.10 It is, moreover, Sarah’s direct awareness of what she thinks, the proposition that her father is home, that explains her authority on what she thinks. Sarah’s authority is an “unwitting” authority because she does not know she thinks her father is home. So (given our earlier stipulation (4): you cannot know what you think without knowing that you think it) Sarah does not know what she thinks. Nonetheless, although she does not know what she thinks, she is aware of (acquainted with) what she thinks, and it is her awareness of this proposition that enables her, without realizing it, to say what she thinks and thus speak her mind. It is what puts Sarah in a privileged position to explain to others why she ran to the door. Her reasons for going to the door reveal, quite unerringly, what Sarah thinks. They, thereby, confirm her unique authority about what is in her mind—an authority that is quite consistent with her ignorance that she has a mind.11 This is not the kind of authority Andreas Kemmerling (2009) has in mind when he describes absolute authority as what you possess when anyone who assumes you are honest cannot rationally doubt p when you say that p but can coherently doubt it when anyone else (who is assumed to be honest) says it. Kemmerling says, and I agree, that we each enjoy absolute authority on whether we are currently thinking that p. If I say I think p, and you regard me as honest, you cannot rationally doubt I think p, but this is something you might well doubt if someone else you regard as honest told you that I think p. Sarah does not have this kind of authority because she does not say she thinks her father is home. She has not yet learned to play that

10 For a similar view of “cognitive” qualia, see Peacocke (1998, 64), who describes current conscious thoughts as contributing “to what, subjectively, it is like for the person who enjoys them.” Also see Flanagan 1992, ch. 4; and Goldman 1993, section 8; Pitt 2004, 2009; Siewert 1998; and Horgan and Tienson 2002. 11 Recall footnote 7. Although Sarah’s mode of awareness of this proposition is non-epistemic, this is quite consistent with her knowing a lot about this proposition and, in particular, with being able to tell (authoritatively) what proposition it is. You do not have to know (be aware that) p and q are propositions to be aware of, and thereby distinguish, p from q anymore than you have to be aware (know) that X and Y are counterfeit to be visually aware of, and thereby distinguish, X (the $100 counterfeit bill) from Y (the $50 counterfeit bill).

58

Skepticism about Introspection

language game. All she says is that he is home, and she is not an authority on her father’s whereabouts. Nonetheless, if we understand Sarah’s declarations not as statements about where her father is, but instead as expressions of where she believes he is—which is, after all, what we know them to be—she enjoys Kemmerling-like authority. Assuming she is sincere, I cannot rationally doubt she thinks that her father is home when she tells me he is home—something I can certainly doubt if someone else I believe to be sincere tells me Sarah believes it. Sarah has the same authority about what she thinks as I would have about what The Philosophical Gazette is publishing if they neglect to tell me they are publishing everything I write. If they do not inform me of their intentions, I will not know that they are publishing p when I write p. Given our earlier concession (you cannot know what they are publishing if you do not know they are publishing it), I will not know what they are publishing despite my awareness (via my perception of what I write) of what they are publishing. My ignorance that they publish what I write prevents me from knowing what they publish, but it does not prevent me from being aware of what they publish, and this awareness gives me unwitting authority about what they publish. If you (unlike me) know the Gazette’s intentions, I will be the one you consult when you want to know (in advance) what they are publishing. You cannot, of course, ask me what they are publishing. I will not know the answer to that question. Still, my authority on what they are publishing will be evident in my answers to questions about what I am writing. These answers tell you, quite reliably, what they are publishing. They give you knowledge of what The Gazette is publishing despite coming from someone (me) who does not know what they are publishing. It is this kind of authority Sarah has about what she thinks. Once Sarah learns—from whatever source—that the reason she gives for what she does is an expression of what she thinks, she will achieve a mature (a “witting”) authority about what she thinks. She will graduate from mere acquaintance with what she thinks to knowledge of what she thinks. But until she graduates, we have a mildly paradoxical (to Cartesians) state of affairs: by listening to and watching her, Mother knows what Sarah is presently thinking, something that Sarah herself (given her ignorance that she is thinking it) cannot find out by introspective means.

5. Introspection The point we have reached so far is this: for purposes of finding out that you think, awareness of what you think is of absolutely no help. It is of no help for the same reason awareness of what was reported on the evening news—that it is snowing in

Awareness and Authority

59

Miami—is of no help in finding out whether this was reported on the evening news. If you want to find out whether the Miami snowfall was reported on the news you have to consult something other than what was reported on the news because that—that it is snowing in Miami—will not tell you what you want to know. In this respect, propositions are evidentially worthless. They have no probative value. The fact that it is snowing in Miami might give you a reason to believe something—that, for instance, it is abnormally chilly in Miami—but the proposition will not give you a reason to believe it. It will not give you a reason to believe anything—even (if you happen to believe it) that you believe it. This is why awareness of what you think and the resulting authority this awareness gives you about what is going on in your mind does not make you an authority on the fact that you have a mind. It would, for this reason, be a mistake to view introspection as a form of inner sense, as a perception-like awareness of one’s own thoughts. Sarah is aware of her own thoughts, yes, but unlike the (often) revealing aspects of ordinary objects (trees, people, automobiles, houses), the aspect of her thought she is aware of, its content, does not disclose, indicate, or reveal that it is the content of a thought. It does not indicate or reveal anything at all except (see footnote 11) what it is she is thinking. Sarah cannot, therefore, come to know she has thoughts by awareness of her thoughts in the way she might become aware (know) that the dog is on the couch by awareness of the dog on the couch. Unlike perception of ordinary concrete objects (dogs and couches), awareness of one’s own thought contents is not a way of coming to know one has thoughts anymore than is awareness of what The Philosophical Gazette is going to publish my (or even a) way of coming to know they are publishing it.12

6. Conclusion I have conceded in (1) that we know we think, but I have carefully avoided saying, or even speculating on, how we know it. I have spent my time nourishing skeptical suspicions by showing that our access to our own thoughts, a mode of awareness that gives us authority (“unwitting” authority) on, and (once we achieve “witting” authority) knowledge of, what we think is capable of doing so only if we have some other way of knowing, some other way of finding out, that

12 I take this to be merely an additional (probably unnecessary) nail in the coffin Shoemaker (1996) constructed for perceptual models of introspection.

60

Skepticism about Introspection

we think it. This, though, leaves a lacuna in the argument for conciliatory skepticism. Perhaps there is some aspect of our own thought other than its content that we enjoy exclusive awareness of which provides this essential missing knowledge, a feature or aspect of our thought which reveals it to be a thought.13 We seem able to tell by introspective means, for instance, whether we are angry that p or hope that p, whether we regret that p or are glad that p. We are aware of content, yes, but we also seem to enjoy an awareness of the attitudinal (perhaps these are functional) aspects of these states. Perhaps, then, it is these other aspects of our thoughts (fears, regrets, desires) that reveal (to mature subjects) their thought-like nature. I have ignored this possibility. By a careful choice of examples and analogies I have suggested, instead, that the disputed knowledge (that we think)—if we really have it—comes from others. It comes from those (parents, teachers) from whom, or with the help of whom, we came to understand what it means to think. I have suggested this. I have not shown it. My motive for ignoring this possibility was the conviction that if you are going to identify an item you are directly aware of as a thought, something having intentional content, it would surely require awareness of the item’s intentional content, the feature that makes it a thought. That, after all, is the key property, the property that makes it a thought. If what you are aware of lacks that property, it is not a thought. Everything else is window dressing. You cannot, after all, find out whether an apparent $100 bill is a real $100 bill by seeing whose picture is on it. Or what color it is. That is not going to tell you what you want to know. You have to look to whatever feature makes it a genuine $100 bill. In the case of thought, that is content. If the content does not tell you it is a thought, nothing else will. But, as I say, I suggested this. I did not show it. So my argument for conciliatory skepticism falls short of the mark. To complete it I would have to show that there is no other aspect of our own thoughts we are (exclusively) aware of that (unlike the thought’s content) indicates its possessor to be a thought. I doubt that there are such aspects, but I do not know how to show that there are not. Frankly, I do not know how we know we think. I do not even know how we came to think we think. But I am convinced the answer to how we know it (if we know it) lies in the answer to how we came to think it—how, that is, we came to understand what thinking is. There must be something about the process of acquiring this understanding that makes those who undergo the process not only able

13 A suggestion I am grateful to Heather Logue for making in her comments on an earlier draft of this paper.

Awareness and Authority

61

to think they think, but to think it for reasons (or in circumstances) that qualify their thinking it as a form of knowledge. Other philosophers have had similar hunches. In his well-known study of concepts, for instance, Chris Peacocke (1992, 157–58) suggests that to possess the concept of thought and, thus, to be able to think you think, you must be caused to think you think thus-and-so by thinking thus-and-so. If you are not “compelled” to have this 2nd order belief (that you think thus-and-so) by your 1st order thought (that thus-and-so)—thereby converting (according to Peacocke) the 2nd order belief into knowledge—you do not (yet) have the concept of thought. You do not yet fully understand what thinking is. So you do not yet satisfy the condition laid down in (1) for knowing you think. You do not yet think you think. Acquiring the concept of thought transforms causal arrangements in the head so that when a thinker finally gets around to thinking she thinks, she is caused to think it in a knowledgeable way. This, of course, would be a neat explanation of (1). It would explain why those who understand what thinking is know they think. It would not, however, answer the question I am struggling with here. It would merely transform the way this question is asked. It would transform the question “How do we know we think?” into the question “What made us able to think we think?” What made now fouryear-old Sarah’s belief that her father is home cause her to believe she thought he was home (thus giving her knowledge that she believed it) when it did not do so a few months earlier? Did some knowledgeable informant (a teacher? a parent?) just tell her: Look, Sarah, whenever you think your father is home, you can (indeed, you are fully entitled to) believe that you think he is home? This, though, cannot be the way it works. Sarah cannot follow these instructions. She could follow these instructions only if she already had a way of telling whether she thought her father was home. If she already has that talent, she does not need the instructions. My own hunch (and it is only a hunch) is that since one is already aware of what one thinks (see §3 and §4 above) before one knows one thinks it, the key to understanding that one thinks is learning to identify this “object” as something one thinks. Since (see §5) there is nothing about this object of awareness that reveals it to be the content of a thought, identifying it would seem to involve understanding the subtle interplay between explanatory and justifying reasons.14 When Sarah, mistakenly thinking her father is home, runs to the door to greet him and sees only the

14 See Moran’s (2001, §4.5) insightful discussion of an agent’s knowledge of his own intentional actions and the unique stance this involves on the part of the actor toward his own beliefs.

62

Skepticism about Introspection

mailman’s truck, her reason for going to the door vanishes. What she was prepared to give as her reason is no longer available. Her father is not there. So when Mother, observing her disappointment, asks, “Did you think Daddy was home?” Sarah must come to understand that this proffered explanation for her action, although it somehow involves both Daddy and being home does not require Daddy to be home. Although it mentions her father, it is a fact about her, Sarah, not her father, a fact about her that (wherever Daddy happens to be) explains (even if it fails to justify) her excited rush to the door. This, surely, is a critical step in coming to understand what it is to think Daddy is home. It is something that can explain your action—running to the door—even when you have no reason—no justifying reason—for running to the door. How one achieves this understanding is a mystery to me. Maybe that is because I do not spend much time around three-year-olds. I hope psychologists who do will someday explain this to me. When they do, they will also explain why (1) is true: why those who think they think, know they do.15

15 I read earlier versions of this chapter at a number of venues. I, thank, especially, Fred Adams, John Campbell, Andreas Kemmerling, Karen Neander, Mark Phelan, David Sanford, Dennis Stampe, and Jonathan Schaffer for helpful comments. Geert Keil, Katia Saporiti, and Heather Logue gave me useful comments at conferences in Heidelberg and Oslo. The editors of this volume, Declan Smithies and Daniel Stoljar, also provided useful critiques. I am also grateful to the Alexander von Humboldt Foundation for their support of this research.

Awareness and Authority

63

References Astington, J. W. 1993. The Child’s Discovery of the Mind. Cambridge, Mass.: Harvard University Press. Aune, B. 2009. An Empiricist Theory of Knowledge. Montague, Mass.: Bowler Books. Baron-Cohen, S. 1995. Mindblindness: An Essay on Autism and Theory of Mind. Cambridge, Mass.: MIT Press. Bartsch, K., and H. M. Wellman. 1995. Children Talk about the Mind. Oxford: Oxford University Press. Boër, S. E., and W. Lycan. 1986. Knowing Who. Cambridge, Mass.: MIT Press. Burge, T. 1988. “Individualism and Self-Knowledge.” Journal of Philosophy 85: 649–693. ———. 1996. “Our Entitlement to Self-Knowledge.” Proceedings of the Aristotelian Society 96: 91–116. Carpendale, J., and C. Lewis. 2006. How Children Develop Social Understanding. Malden, Mass.: Blackwell Publishing. Davidson, D. 1988. “Reply to Burge.” Journal of Philosophy 85: 664–665. Davies, M. 1998. “Externalism, Architecturalism, and Epistemic Warrant.” In Knowing Our Own Minds, edited by C.Wright, B. C. Smith, and C. Macdonald. Oxford: Oxford University Press. Dretske, F. 2003a. “How Do You Know You Are Not a Zombie?” In Privileged Access and First-Person Authority, edited by B. Gertler. Aldershot: Ashgate Publishing. ———. 2003b. “Knowing What You Think vs. Knowing That You Think.” In The Externalist Challenge: New Studies on Cognition and Intentionality, edited by R. Schantz. Berlin: Walter de Gruyter. ———. 2003c. “Externalism and Self Knowledge.” In Semantic Externalism, Skepticism and Self-Knowledge, edited by Susana Nuccetelli. Cambridge, Mass.: MIT Press. ———. 2005. “The Case Against Closure.” In Contemporary Debates in Epistemology, edited by Ernest Sosa and Matthias Steup. Oxford: Blackwell. Feit, N. 2008. Belief about the Self: A Defense of the Property Theory of Content. Oxford: Oxford University Press. Flanagan, O. 1992. Consciousness Regained. Cambridge, Mass.: MIT Press. Flavell, J. H. 2003. Development of Children’s Knowledge about the Mind: The Heinz Werner Lectures. Worcester, Mass.: Clark University Press. Flavell, J. H., P. Miller, and S. Miller. 1993. Cognitive Development. Englewood Cliffs, N.J.: Prentice Hall. Gallois, A. 1996. The World Without, The Mind Within: An Essay on First-Person Authority. Cambridge: Cambridge University Press. Goldman, A. 1993.“The Psychology of Folk Psychology.” Behavioral and Brain Sciences 16: 15–28. Gopnik, A. 1993. “How We Know Our Minds: The Illusion of First Person Knowledge of Intentionality.” Behavioral and Brain Sciences 16: 1–14.

64

Skepticism about Introspection

Horgan, T., and J. Tienson. 2002. “The Intentionality of Phenomenology and the Phenomenology of Intentionality.” In Philosophy of Mind: Classical and Contemporary Readings, edited by D. Chalmers. Oxford: Oxford University Press. Kemmerling, A. 2009. “On Being Able to Say What One Thinks.” Manuscript. Kobes, B. 2003. “Mental Content and HOT Self-Knowledge.” In Reflections and Replies: Essays on the Philosophy of Tyler Burge, edited by M. Hahn and B. Ramberg. Cambridge, Mass.: MIT Press. Loar, B. 1994. “Self-Interpretation and the Constitution of Reference.” Philosophical Perspectives 8: 51–74. McLaughlin, B. and M. Tye. 1998. “Externalism, Twin Earth, and Self-Knowledge.” In Knowing Our Own Minds, edited by C.Wright, B. Smith, and C. MacDonald. Oxford: Clarendon Press. Moran, R. 2001. Authority and Estrangement: An Essay on Self-Knowledge. Princeton: Princeton University Press. Peacocke, C. 1992. A Study of Concepts. Cambridge, Mass.: MIT Press. ———. 1998. “Conscious Attitudes, Attention, and Self-Knowledge.” In Knowing Our Own Minds, edited by C. Wright, B. Smith, and C. MacDonald. Oxford: Clarendon Press. Pitt, D. 2004. “The Phenomenology of Cognition or What Is It Like to Think That P.” Philosophy and Phenomenological Research 69: 1–36. ———. 2009. “Intentional Psychologism.” Philosophical Studies 146: 117–138. Shoemaker, S. 1996. “Self-Knowledge and ‘Inner Sense’.” In The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press. Siewert, C. P. 1998. The Significance of Consciousness. Princeton: Princeton University Press.

3 Knowledge of Perception Daniel Stoljar

I’m not asking whether you know you are not a zombie. Of course you do. I’m asking how you know it. The answer to that question is not so obvious. Indeed it is hard to see how you can know it. Wittgenstein . . . didn’t think he saw anything that allowed you to infer he saw it. The problem is more serious. There is nothing you are aware of, external or internal, that tells you that, unlike a zombie, you are aware of it. Or, indeed, of anything at all. (Dretske 2003, 1)

1 Suppose I am watching my son do somersaults in the living room; and suppose (having nothing better to do) I ask myself whether I am watching my son do somersaults. Would I then come to know that I am? Of course the answer is yes. I am not asking you to suppose that my living room is populated with the sort of things epistemologists like to discuss: fake sons, gurus, mad scientists, etc. The imagined situation is not that at all. Rather it is just a normal day in which I am watching my son do somersaults. If you like, imagine also that I fully understand the notions involved—I, watching, son, living room, etc.— that I have asked myself seriously (and not just in a half-hearted way) whether I am watching my son, and that I am not so busy as to have my mind crowded with other things. If it is distracting, forget also that watching is some sort of action, and so that to know that I am watching is to know that I am acting; concentrate just on seeing

66

Skepticism about Introspection

since watching entails seeing.1 Do I know that I see my son? In the imagined situation, of course the answer is yes. But there is an apparently simple argument for the opposite conclusion. The first premise of the argument is e1: e1. If I know that I am watching my son, there is an answer to the question “How do I know that I am watching my son?” The claim here is not that I must know the answer to the question; I do not have to know how I know. The claim is rather that there must be an answer. What motivates this claim? Well, not everyone who is watching his or her son knows it. Perhaps somewhere in Brazil a new world monkey—a pygmy marmoset, say—is watching (and so seeing) his son do somersaults; but maybe he does not know that he is. Or perhaps on another day I am so preoccupied with one thing or another that I do not know that I watching my son even though I am. However, it is not a miracle that I know that I am watching my son while my pygmy marmoset friend does not know that he is watching his; likewise it is not a miracle that I know I am watching in some circumstances but not in others. Hence, there must be an answer to the question “How do I know?” The second premise of the argument is e2: e2. If there is an answer to the question “How do I know that I am watching my son?” that answer must be a proposition that (a) I am aware of and (b) “tells me” that I am watching my son. Very often a question of the form “How to you know that p?” is a request for evidence. If I tell you that the Dean is going to close down the physics department, and you ask me how I know, in many instances you would be satisfied if I responded with something like, “Well, she told Professor X in the strictest confidence that she was going to, and he immediately told me.”These further claims are things I am aware of and are things that support or confirm or speak in favor of the thing that I know. To the extent that this case is typical, what you are looking for when you ask me how I know something is a proposition or fact that in a certain sense I am aware of and which supports what I claim to know; that is, you are looking for my evidence. The third premise of the argument is e3: e3. There is no proposition that meets conditions (a) and (b). This premise is motivated by a plausible idea about the nature of perception. When I watch, and so see, my son, what facts do I come to be aware of? Well, if we set aside

1 I will myself move back and forth between watching and seeing in the discussion to follow; but nothing substantial will turn on this.

Knowledge of Perception

67

skepticism about the external world, the facts that I come to know are facts such as the following: that my son is doing somersaults, that my son is in the living room, that my son exists, and so on. These are the facts that in the first instance I come to know on the basis of perception. However, these facts seem to bear no particular relation to the fact that I am watching my son. For one thing they do not entail or necessitate that I am watching my son. My son might perfectly well be doing somersaults without my watching him—if I closed my eyes, say. Moreover, they do not confirm or support or tell me that I am watching my son. Imagine someone advancing an argument whose premise is that X’s son is doing somersaults to the conclusion that X is watching his son do somersaults. That looks like a spectacularly terrible argument; the premise not only fails to entail the conclusion, it fails to give it any reasonable support. e1 and e2 tell us that if I know that I am watching my son there must be an answer to a certain question and also that the answer must meet certain conditions. e3 says that no answer meets the conditions. Putting these together, I do not know that I am watching my son. But, since—obviously—I do know, something is wrong. How is it possible that I know I am watching my son? The argument I have just given—the evidence argument, as I will call it—is Dretske’s or at any rate is Dretske’s as I understand him (cf. Dretske 2003, this volume; the argument is also mentioned by other writers or at any rate by other writers as I understand them—I will set them aside here.2) In the passage quoted above, Dretske puts the point in terms of (philosophical) zombies, i.e., creatures by definition physically the same as us but lacking conscious experience; his question is “How do you know you are not a zombie?” But as he himself notes, zombies are not really to the point and raise distracting questions. For one thing, a lot of philosophers think that zombies are strictly speaking impossible, or, at any rate, that we have no reason to suppose them possible. The point comes out better if you adopt one of Dretske’s more mundane examples (which I have) and frame matters in terms of that. My aim is to present a response to the evidence argument and to explore the conception of introspection that emerges from that response. I will begin, in section 2, 2

It seems to me that M. G. F. Martin is suggesting an argument along these lines, when he writes: “why should the evidence that the subject has about how the world is have any bearing on what beliefs a particular person has? The kind of evidence that grounds our beliefs about the world would seem in general not to be relevant to the question of what beliefs we ourselves have” (1998, 110). That Martin is discussing belief rather than perception lends support to the idea mentioned in comment 5 below, namely, that the evidence argument is a very general argument and is not limited to the case of perception. Dretske’s argument has also received extensive discussion in Byrne 2005 and Byrne, this volume.

68

Skepticism about Introspection

with some preliminary comments before presenting the response in sections 3 to 7. In sections 8 and 9, I will contrast my proposal with an alternative.

2 Comment 1: You might think that the evidence argument is generated by a careful choice of example. Watching my son involves seeing him, and this involves standing in a relation to him, something I could hardly do unless he exists. But then you might think that the evidence argument is generated by the possibility that I do not know that he exists, and as such is simply the problem of the external world all over again. However, while it is true I cannot see my son unless he exists, this is not what generates the problem. Many views of perception postulate perceptual facts that do not require the existence of my son, and one could perfectly well discuss the basic issue in terms of these facts rather than the fact that I am seeing my son. Talk of seeing simply makes the problem easier to state. Which theories of perception involve facts that do not require the existence of my son? Consider the intentionalist view. At least on a cartoonish understanding of that view, I see my son only if he plays the right sort of explanatory role in my being in a certain representational state, the state of visually representing (the proposition) that my son exists. But we could if we like focus on the question “How do I know that I am in the state of visually representing that my son exists?” That question does not presuppose that he exists. Likewise, consider the disjunctivist view. At least on a cartoonish understanding of that view, I may be in the situation of either seeing my son or it seeming to me as if I am seeing my son. This disjunctive fact does not require the existence of my son. But we could if we like focus on the question “How do I know that either I am seeing my son or it seems to me that I am seeing my son?” So far as I can see, adopting either of these theories of perception does not affect the issue I want to focus on, and so I will concentrate here on the simpler idea that I am seeing my son. Comment 2: It sometimes happens that a person knows something but does not know (because they cannot remember) how. Suppose I know that the Battle of Marston Moor was fought in 1644 but I do not remember that Mrs. Beamish (my fourth grade teacher) told me this, and in fact I have no idea how I know. If so, I may not be (currently) aware of any proposition that speaks in favor of what I know. Does this show that the condition on evidence set out in e2 is mistaken? Well, as stated it does. However, I take this to be a somewhat superficial problem for a proponent of the evidence argument; for example, e2 could be adjusted to accommodate this by replacing “is” with “is or was.” I will therefore ignore this complication in what follows.

Knowledge of Perception

69

Comment 3: It might be that if either a perceptual model or the inferential model of introspection were true, e3 would be false. Nevertheless I will set both aside here. As regards the perceptual model, Dretske himself assumes that Shoemaker has refuted it.3 I am not sure about that, since it seems to be doubtful that the perceptual model is clear enough to be refuted. But I do agree with both Dretske and Shoemaker that we should do without it if we can, and that will be my policy. As regards the inferential model, as this is usually developed I infer that I am watching my son from premises about my behavior. This idea may be stipulated to be irrelevant in the case at hand. Even if there are occasions on which I come to know that I see my son by inferring this from my behavior, I did not in the case I began with; or at any rate so we may imagine. In recent work,Alex Byrne has developed and defended the inferential model in a nonstandard version.4 For Byrne, I infer that I see my son from certain premises that Byrne calls ‘v-propositions’, where a ‘v-proposition’ encodes information about how my son looks as he is doing somersaults. Byrne’s proposal is interesting and controversial; however, as I understand matters, one of its main motivations is the idea that it solves the problem presented by the evidence argument. If, as I think, that problem may be solved in a different way, this part of the motivation for Byrne’s view evaporates. Comment 4: I said that support for e3 derives from a fact about the nature of perception, viz., that when I see my son, in the normal case I come to know certain facts about my son, e.g., that he exists, that he is doing somersaults, etc. It might be thought that support for this premise derives also or instead from another source, viz., the so-called diaphanousness of perception.5 This claim is somewhat notorious in philosophy of mind. Some people take it to mean that you are never aware of the fact that you see but only the fact (or thing) which you see. In my view, this claim is too incredible to be true. For surely we are sometimes aware that we see; for example, I know (and so am aware) that I am seeing my son. Others (more plausibly) take the diaphanousness claim to be primarily a claim about attention, viz., that when I try to attend to my seeing of my son, I wind up attending to my son. If this claim is true, it raises a number of questions—in which of the many senses of ‘attention’ is it true, for example, and what is it about attention in the relevant sense 3

“I can’t improve on Shoemaker” (Dretske 2003, 11 n. 12). For Shoemaker’s criticism of the perceptual model, see Shoemaker 1996, and for other criticisms, see Siewert, this volume. For defenses of the view, see Gertler, this volume, and Horgan, this volume. 4 See Byrne, this volume; see also Byrne 2005. 5 It has often been suggested to me (e.g., by David Chalmers in conversation) that Dretske 2003 is relying on the diaphanousness of perception. I am not sure that this is so, but even if it is, the evidence argument seems to me to be stronger if we decouple it from any implausible version of diaphanousness. For my own treatment of the diaphanousness issues, see Stoljar 2004.

70

Skepticism about Introspection

that makes it true? But while these questions are important and interesting, they also seem remote from the issues we are discussing and so I will set them aside. Comment 5: One might think that the evidence argument is limited in scope to perception since that is our lead example. If so, the issue we are discussing might appear narrow. However, a similar argument can be mounted in the case of many other mental states. To illustrate, take imagination. Suppose I imagine that my son is doing somersaults in the living room. In the normal case I know perfectly well that I am imagining this. How do I know? Well, imagining my son doing somersaults might make me aware of many propositions: e.g., that it is possible that my son is in the living room, that it is possible that my son is doing somersaults, etc. But none of these propositions, even if true, make it likely that I am imagining him. Or take intention. Suppose I intend to join my son doing somersaults in the living room in ten minutes. In the normal case I know perfectly well that I intend this. How do I know? Well intending to do somersaults might make me aware of various propositions: e.g., that I will in ten minutes do somersaults in the living room, that there will in ten minutes be two people doing somersaults in the living room, etc. But none of these propositions, even if true, make it likely that I am currently intending to do somersaults. The question of whether the proposal I will make about the evidence argument can be extended from perception to other cases is interesting but I will not pursue it here. Comment 6: The question “How do you know?” is a close cousin of the question “Why do you believe?” The first presupposes that the addressee knows the thing in question, while the second presupposes only that he or she believes it. But both questions are normally requests for evidence. If you ask me why I believe that the Dean is going to close down the physics department you would very often be asking me for some propositions that confirm or support what it is that I believe. In consequence, the evidence argument could be formulated in terms of the question “Why do I believe that I see my son?” rather than the question “How do I know that I see my son?” So far as I can see, the argument stated in these terms would have the same persuasiveness of the original and so I will set it aside as well. Comment 7: What about an uncritical appeal to the notion of introspection? That is, why not try to close down the issue by saying that the answer to the question “How do I know that I am seeing my son?” is “by introspection” and leave it at that? Dretske thinks that introspection is by definition that (distinctive) way in which I come to know (e.g.) that I am watching my son. But, as he says, an appeal to introspection in this sense gives us no insight into what if anything is wrong with the argument. It may be true that I come to know that I am seeing my son by introspection, but this tells us only that the evidence argument is unsound. It does not tell us why it is unsound, and it is this latter question that will be our focus.

Knowledge of Perception

71

3 Why then is the evidence argument unsound? My answer will proceed as follows. First, I will consider a particular proposal about how to respond, which I will call “the foot-stamping view.”Then I will consider a different proposal, which I will call “the Austin-inspired view.” Neither the foot-stamping view nor the Austin-inspired view is correct on its own. However, if they are combined together in the right way, we have (I think) a plausible proposal about what is wrong with the evidence argument.

3.1. The Foot-Stamping View According to the foot-stamping view, the third premise (e3) of the evidence argument is false. In the case we are imagining, e3 says that there is no proposition (a) of which I am aware and (b) which supports the proposition that I am watching my son. Against this premise, the foot-stamper insists that there is such a proposition, viz., the very proposition that I am watching my son. How so? Well, in the first place, I know that this proposition is true and in this sense I am aware of it. After all, nobody is denying that I know that I am watching my son; “Of course you do,” says Dretske in a related context. Second, it is a proposition that tells me that I am watching my son, because sometimes propositions tell me about, or speak in favor of, themselves. (Both the “telling” metaphor and various accounts of epistemic support permit this possibility.) What if anything is wrong with the foot-stamping response? Dretske offers one objection when he remarks, “To insist that we know it despite having no identifiable way we know it is not very helpful. We can’t do epistemology by stamping our feet” (2003, 9). However, while this remark is what inspired the name of the view, it is in fact hard to see exactly what objection Dretske intends by it. For the footstamper (as least as I understand him) is not insisting that I know that I am watching my son and yet I have no identifiable way that I know. On the contrary, the footstamping response agrees that there is an identifiable way (at least if this means: an identifiable answer to the question “How do I know?”). That way is the (easily identifiable) proposition that I am watching my son. It is puzzling, therefore, that Dretske can say otherwise. We will be in a better position to respond to this puzzle at a later stage in our discussion. For the moment, I want instead to suggest that a proponent of the evidence argument might respond to the foot-stamping response in a way that is different from Dretske, viz., by finessing it rather than confronting it directly. Right at the start of the argument we agreed that the question “How do I know

72

Skepticism about Introspection

that I am watching my son?” has a nontrivial answer. If so, the argument can be adjusted to avoid the foot-stamping response. The reformulation would go as follows: e1. If I know that I am watching my son do somersaults, there is a nontrivial answer to the question “How do I know that I am watching my son?” e2. If there is a nontrivial answer to the question “How do I know that am watching my son?” that answer must be a proposition that (a) I am aware of; (b) “tells me” that I am watching my son; and (c) is distinct6 from the proposition that I am watching my son. e3. There is no proposition that meets conditions (a), (b), and (c). e4. Therefore, I do not know that I am watching my son. On the face of it, this version of the argument—which I will call version 2 of the evidence argument, to distinguish it from the version already in play—has whatever plausibility the original had, and yet foot-stamping is no response to it. Hence, if the foot-stamping response is the only response available to the evidence argument, it looks as if we so far have no response.

3.2. The Austin-Inspired View To this point we have been mostly uncritical about the question “How do I know that I am watching my son?” assuming only that it is often a request for evidence. But there is a famous discussion, J. L. Austin’s, that bears on what this question may mean. Here is Austin: Suppose I have said ‘There is a bittern at the bottom of the garden’, and you ask ‘How do you know?’ my reply may take very different forms: a. I was brought up in the fens b. I heard it c. The keeper reported it d. By its booming e. From the booming noise f. Because it is booming. We may say, roughly, that the first three are answers to the questions ‘How do you come to know?’, ‘How are you in a position to know’, or ‘How do you know?’ understood in different ways: while the other three are answers to ‘How can you tell?’ understood in different ways. (Austin 1961, 79) 6 As Mark Jago pointed out to me, there are a number of questions about how to interpret ‘distinct’ here. I acknowledge this, but I will leave this notion intuitive in the discussion to come.

Knowledge of Perception

73

There are various interpretative questions about what Austin says here. For one thing, the answers (a–f ) seem to concern, not the question “How do you know that there is a bittern at the bottom of the garden?” but rather the question, “How do you know that it (that) is a bittern?” For another thing, it is not quite clear how precisely to characterize the distinction at issue. Elsewhere Austin says that (a–c) provide my “credentials” while (d–f ) provide my “facts” (i.e., the facts that “prove” (as he says) it is bittern). Finally, it is natural to see a connection between Austin’s use of the idea of “telling” and Dretske’s: on the one hand, there is something I am aware of that tells me it is a bittern, viz., its booming; but, on the other hand, there is (apparently) nothing I am aware of that tells me I am watching my son. If so, answers (d–f ) are responses to a request for evidence. All of these interpretative points may be developed in different ways. For our purposes it is sufficient to draw from Austin (without any claim to scholarly accuracy) a distinction between (what I will call) an evidence-seeking version of the question “How do I know?” in which we ask for evidence for what I know—this is the version we have been operating with all along—and an explanation-seeking version, in which we ask for an explanation for the fact that I know. So, for example, if you ask me how I know that the Dean is going to close down the physics department, you might be asking me for evidence for the proposition that I know, viz., that the Dean is going to close down the physics department, or you might also or instead be asking me for an explanation for a certain fact, i.e., the fact that I know that the Dean is going to close down the physics department.7 Armed with this distinction, a straightforward objection to the argument comes into view, viz., that it does not go through if what is at issue is an explanationseeking question: e1. If I know that I am watching my son do somersaults, there is a nontrivial answer to the explanation-seeking question “How do I know that I am watching my son?”

7 Objection: The question “How do you know the Dean will close down the physics department?” is certainly a request for evidence. But it is also a request for explanation, at least of a sort. Hence, there is no distinction of the sort that you want to draw from Austin. Response: It is true that requests for evidence might be thought of as requests for explanation too; at least it does not stretch the concept of explanation to suppose so. But there are nevertheless requests for explanation that are not also requests for evidence, at least not in the sense of ‘evidence’ that is at issue in the evidence argument. To mark this, we could if we liked work with a distinction between evidence-seeking and explanation-but-not-evidence-seeing questions rather than the simpler distinction I have adopted in the text. However, since doing so would affect anything I have to say, I will continue with the simpler distinction.

74 Skepticism about Introspection e2. If there is a nontrivial answer to the explanation-seeking question “How do I know that am watching my son?” that answer must be a proposition that (a) I am aware of; (b) “tells me” that I am watching my son; and (c) is distinct from the proposition that I am watching my son. e3. There is no proposition that meets conditions (a), (b), and (c). e4. Therefore, I do not know that I am watching my son. The problem with the argument in this version—version 3 of the evidence argument, as we can call it—is that e2 is false: it is not true that if there is an answer to the explanation-seeking question “How do I know that I am watching my son?” then I must be aware of that answer. Of course, I do need to be aware of the answer if I am to give the answer to someone else; but for it to be an answer I do not need to be aware of it. To illustrate, consider again the explanation-seeking question, “How do I know it is a bittern?”What might answer this question, and so explain the fact that I know that it is a bittern? Well, one answer is provided by Austin, namely, that I was brought up in the fens, or rather, to fill out the story a little bit, that I was brought up in the fens and people who are brought up in the fens know a bittern when they see one. These facts, assuming for the moment (implausibly!) that they are facts, would indeed explain that I know that it is a bittern. However, nowhere here is it required that I am aware of the facts in question. For one thing, I do not need to know that I was brought up in the fens, even if I was; I may be completely mistaken about my origins and insist that I was brought up in the tropics. Nor do I need to know that people who are brought up in the fens know a bittern when they see one, even if this is so; I may insist that there is some sort of ambient gas in the fens that blocks people from knowing bitterns when they see them, and that the only way to be able to do this is by growing up in the tropics. What is true for the question about bitterns is true also for the question about seeing my son—at least if what is at issue is a request for explanation. What explains the fact that I know that I am watching my son? Here is an initially plausible suggestion. I know that I am watching my son because (a) I am watching my son; (b) I am suitably situated—here, “suitably situated” summarizes the facts of the case I began with: viz., that I understand the concepts involved, I have asked myself whether I am watching my son, and I am not distracted by other things; and (c) that suitably situated people who are watching their sons know that they are. Taken together these claims explain the fact that I know that I am watching my son. On the other hand, it is obviously not a requirement on (a–c) answering the explanation-seeking question that I am aware of (a–c). If so, e2 is false, and the argument can be rejected.

Knowledge of Perception

75

3.3. Foot-Stamping Mingled with Austin The Austin-inspired response is clearly not going to do on its own. By itself, the response reveals only that any successful version of the evidence argument will employ one legitimate version of the “How do you know?” question, i.e., the evidence-seeking version. That there is another legitimate version—i.e., the explanation-seeking version—is (one might think) beside the point. However, at this stage we may fruitfully combine the Austin-inspired response with the foot-stamping response considered earlier. For if we combine these two ideas we may present a proponent of the evidence argument with a destructive dilemma, as follows. If you want to advance the evidence argument, you must operate either with the explanation-seeking version of the “How do you know?” question or with the evidence-seeking version. If you operate with the explanation-seeking version—that is, if version 3 of the argument is in play—then the second premise (e2) is false, for answers to the relevant requests for explanation do not need to meet the conditions laid down in that premise; in particular one does not need to be aware in any sense of these answers for them to be answers. If, on the other hand, you operate with the evidence-seeking version, then either the first premise (e1) or the third premise (e3) is false, depending on whether you require a nontrivial answer to the question; that is, depending on whether version 1 or version 2 is in play. If version 1 is in play, then (e3) is false, for there is a proposition of which I am aware and which speaks in favor of what I know, viz., the proposition that I am watching my son. If version 2 is in play, then (e1) is false, for it is not the case that there must be a nontrivial answer to the evidence-seeking question “How do I know that I am watching my son?” Of course, there must be a nontrivial answer to the explanation-seeking question, but that is a different matter.

4 How plausible is the Austin-inspired foot-stamping response just sketched? In my view it provides the right form of a solution to the problem posed by the evidence argument. However, before it is adopted we need to bring out and deal with two major issues. The first arises when we look more closely at the explanation that I have just offered of the fact that I know that I am watching my son. The second returns us to Dretske’s objection to the foot-stamping view. In the next few sections, I am going to address both issues. It is convenient to start with the first issue, for thinking it through will provide materials for dealing with the second.

76

Skepticism about Introspection

To bring out this first issue, let us focus again on the explanation-seeking question “How do I know that I am watching my son?” The answer we have given to this question so far is this: (a) I am watching my son; (b) I am suitably situated; and (c) suitably situated people who are watching their sons know that they are. Now, what is the status of the claim (c)? The first thing to say is that it seems not to be just an accidental generalization but is rather a claim with a certain sort of modal force. The main reason for this is that it “supports counterfactuals,” as it is usually put. For example, I have no idea if (e.g.) Barack Obama is watching his son (or daughter in his case) doing somersaults right now. However, I do know that were he to do so, and if he were suitably situated, then he would know that he is. If we agree that the claim has modal force, the question arises as to what the strength, or nature, of that modal force is. An initially plausible proposal is that the claim is naturally necessary; that is, it is or involves an empirical counterfactualsupporting generalization—a psychological law, though presumably of a rough and ready kind, and one that tolerates exceptions, an “other things being equal” law. On this view the claim that suitably situated people who are watching their sons know that they are is analogous to the (rough and ready, other things being equal) claim that people who are brought up in the fens know a bittern if they see one. However, while this assumption is certainly plausible, it creates some difficulties for the response to the evidence argument that we have put forward. To illustrate this, consider a possible world W. W is very much like the world I imagined at the outset in which it is normal day and I am watching my son. The difference between W and the actual world is that, at W, the psychological law we just described does not obtain. Now, is it true at W that I know that I am watching my son? On the assumption that the law in question does not obtain at W, the answer would seem to be “no.” After all, the thing that explained why I know that I am watching my son in the original case is missing in W and nothing has taken its place. Hence (barring a miracle) I do not know that I am watching my son in W. However, it is extremely implausible to say that I do not know that I am watching my son at W, for two reasons. In the first place, what would stop me doing so? After all, I am watching my son, I have asked myself whether I am, I understand the relevant concepts, and—finally—I am rational, or at least as rational as I am in the actual world. But why then would I not come to know that I am watching my son? What else do I have to do? In the second place, regardless of whether I know that I am watching my son in W, it is very plausible that I ought to believe this; that is, I ought from a rational point of view, to believe at W that I am watching my son. But if it is the case that I ought to believe that I am watching my son, then we have a

Knowledge of Perception

77

further reason for supposing that I know this at W. For it seems plausible to say that if I am rational, then other things being equal I will believe what I ought to believe, and the belief in question will amount to knowledge. Hence, if it is true in W that I ought to know that I am watching my son, then, since I am rational there, I will know this.8 The problem for the suggestion that (c) is merely naturally necessary, therefore, is that it entails that at W I do not know that I am watching my son. But, since it seems obvious that I do know this at W, it cannot be that (c) is merely naturally necessary. This is not to deny that (c) is naturally necessary of course; it remains plausible that it is. But it is to deny that it is merely naturally necessary, i.e., it must be necessary in some further way as well. In the face of this, a very natural suggestion is that (c) is not merely naturally necessary but is in addition metaphysically necessary; that is, it is true in all possible worlds (in the widest sense of “all possible worlds”) that suitably situated people who see their sons know that they are. On this view the claim that suitably situated people know that they are watching their sons if they are is analogous to the claim that water is H2O. However, just as it is peculiar to say that (c) is merely naturally necessary, it is likewise peculiar to say that (c) is metaphysically necessary. The problem this time is not the implication that at W I do not know that I am watching my son. If it is metaphysically necessary that suitably situated people who are watching their sons know that they are, then, since I am suitably situated at W, and am watching my son there, then I will indeed know that I am. The problem rather is that the idea that (c) is metaphysically necessary rules out lots of other cases that seem on the face of it to be genuinely possible. To illustrate, consider a possible world W. W is very much like the world I imagined at the outset with the exception that at W I do not believe, and so do not know, that I am watching my son. If you like, imagine that at W God has plucked out the fact that I believe that I am watching my son (and so has plucked out the fact that I know that I am) and yet has left everything else intact. Is W a metaphysically possible world? It is hard to see why not. The fact that I know (or believe) that I am seeing my son is distinct from the fact that I see my son. (Of course, the reverse is not the case because of the factivity of knowledge; but that is not the direction we are interested in.) Hence, it does not seem impossible (at least in the sense of metaphysical possibility) that God could have brought the second

8

That (c) is not merely nomologically necessary I take to be a main lesson of Shoemaker’s discussion in his Brown lectures; see Shoemaker 1996. Shoemaker’s “self-blindness argument” is (as I understand it) precisely an argument that one should deny that at W, I do not know that I see my son.

78

Skepticism about Introspection

fact into existence and not the first. Hence, W is possible. On the other hand, if W is possible, then (c) cannot be metaphysically necessary.9

5 At this point, therefore, we seem to be confronted with a dilemma. The claim (c)— “suitably situated people who are watching their sons know that they are”—is clearly a modal claim in some sense for it supports counterfactuals. But the necessity in question is apparently neither merely natural nor metaphysical. It is not merely natural, because to say that is to say that at W I do not know that I am watching my son. And it is not metaphysical, because to say that is to exclude outright W. How to respond? The way out of this dilemma, I think, is to agree that claim (c) is neither merely naturally necessary nor metaphysically necessary, and to propose instead that (c)—or at least something very close to (c)—is rationally necessary or, equivalently, that it is a requirement of rationality. To say that it is rationally necessary is to say that it is on a par with such principles as “if you believe p and believe if p then q, then you will believe q” and “if you intend to F and know that you can only F if you G, then you will intend to G.” These claims (or, better, instances of these claims, or even better, qualified instances of these claims) are not metaphysically necessary. It is not metaphysically necessary that if I believe p and believe if p then q, then I will believe q. But nor is it true that it is merely nomologically necessary. It is not a mere fact of empirical psychology that I tend to believe q when I believe p and if p then q. Rather such principles are rationally necessary, that is, they are principles that one is required by rationality to follow. One may not follow them, of course, and one may follow them mostly but not in every case; but this is just to say that one may not be in these respects rational, and may not be in these respects rational on every occasion.10 So what I would like to propose is that (c)—or at least something like (c)—is a rational necessity or is a requirement of rationality. Of course, this idea is just that: a

9

That (c) is not metaphysically necessary I take to be a main lesson of Armstrong’s discussion of introspection in, most famously, A Materialist Theory of the Mind; see Armstrong 1968. Armstrong’s “distinct existences argument” is (as I understand it) precisely an argument that W is possible. 10 Why rational necessity as opposed to some other sort? One possible reason is that, following Fine 2004, it is reasonable to suppose that there are three fundamental kinds of necessity: natural, metaphysical, and normative. If the connection we are looking for is neither natural nor metaphysical, and if Fine is right that there are three fundamental varieties of necessity, it must be normative.

Knowledge of Perception

79

proposal; it is hard to prove that it is true. But one argument in favor of its truth is that it allows us to avoid the dilemma about the modal status of (c) that we have been considering. On the one hand, the thesis that (c) is rationally necessary does not imply that at W I do not know that I am seeing my son. The reason is that at W I am as rational as I am at the (imagined) actual world with which we began. If we assume that I abide by or follow the relevant principle of rationality at the actual world, I will likewise abide by or follow it at W. But then it will be true at W, no less than at the actual world, that I will know that I am seeing my son. By the same token, the thesis does not imply that W is impossible. That it is a rational necessity that suitably situated people who are seeing their sons know that they are does not entail that it is a metaphysical necessity. In particular, it is quite consistent with the idea that this is a rational necessity, and that I follow it, that there is a case in which I am suitably situated but have been prevented (by God, for example) from believing that I am seeing my son. To be sure, that will be case in which I have been prevented by God from following (at least on this occasion) a requirement of rationality, but since it is not metaphysically impossible to do that, W is not metaphysically impossible. At the end of section 3 we offered an answer to the explanation-seeking question “How do I know that I am seeing my son?” The answer was this. I know that I am seeing my son because (a) I am seeing my son; (b) I am suitably situated; and (c) suitably situated people who are seeing their sons know that they are. This answer ran into difficulty on the issue of what the modal status of (c) is. A better answer, which solves this problem, is this. I know that I am seeing my son because (a) I am seeing my son; (b) I am suitably situated; (d) there is a rational requirement that suitably situated people who are seeing their sons know that they are doing so; and finally (e) I am rational at least to the extent that I abide by this requirement. It is the combination of (d) and (e) that sustains counterfactuals. Even if Barack Obama is not watching his daughter doing somersaults right now, it is true that were he to do so and if he were suitably placed, we would know that he is watching his daughter. Why so? The answer is that it is a requirement of rationality that suitably situated people who are watching their daughters know that they are, and that Barack Obama just as much as me abides by this requirement.

6 I have argued so far that if we view (c) or something like it as a rational necessity rather than either a metaphysical necessity or a mere natural necessity, we may avoid a major problem for the Austin-inspired foot-stamping view. However, this suggestion faces a number of objections. In this section I will try to deal with some of the main ones.

80

Skepticism about Introspection

First, one might object that to say that there is a rationally necessary connection between the fact I am seeing my son and the fact that I believe or know that I am is simply to say that the first is evidence for the second. If so, we seem to have taken back the Austinian distinction between explanation and evidence that was one of the planks of our discussion. However, in general there is no reason to assume that if p rationally necessitates q then p is evidence for q, and this is especially true if we are using the notion of evidence that is at issue in the evidence argument, i.e., the notion according to which a proposition is evidence only if I am aware of it and it supports what I know.11 Suppose it is a requirement of rationality that if I believe that p and believe that if p then q, then I believe q. And suppose now I believe both p and if p then q but on the basis of no evidence at all; imagine if you like, that both have implanted in my mind on the whim of some bored archangel. In that case it remains a requirement of rationality that I believe q. But it does not follow that I have evidence for q; on the contrary, I have absolutely no evidence for q just as I have no evidence for p or if p then q. Hence, one can say that there is a rational connection between various states without the first being evidence for the first.12 Second, one might object that to say that (c) is required by rationality is implausible because it is too demanding. Might there not be possible defeaters, i.e., factors that make it the case that I do not know that I am seeing my son, despite the facts that I am doing so, I am suitably placed and am rational? For example, imagine I am an eliminativist about vision and so think that nobody ever sees his son at all.13 Or imagine that I am a victim of reverse Anton’s Syndrome, and so am convinced (falsely) that I am blind. Or imagine that someone I trust (or some collective of people I trust) has just informed me that, whatever I am inclined to think, I am not seeing my son. In such cases, I may not know or believe that I am seeing my son even if I am rational and am suitably placed. However, while it is true that we should allow for defeaters of this kind, it is a mistake to suppose that this is a reason to abandon the idea that (c) or something like it is rationally necessary. Instead it is a reason to develop a more discerning interpretation of the precise content of the requirement in question; indeed, it is for this reason that I have spoken so far of “(c) or something like it.” There are a number of different ways in which one might try to develop this more discerning

11

I will consider a different notion of evidence in sections 8 and 9 below. The argument given in this paragraph is a version of Broome’s argument against bootstrapping; see Broome 2005, 2007. Broome does not apply the argument to the case of self-knowledge. 13 For some discussion of the eliminativist case, see Siewert 1998, and Silins, this volume. 12

Knowledge of Perception

81

interpretation. For example, when considering the suggestion that (c) is naturally necessary we noted that on such a view, the law in question would presumably be an “other things being equal” law, a law that tolerates exceptions. Perhaps what is true for the laws of empirical psychology is true also for the laws of rational psychology? On this view, the requirement is not that without exception suitably situated people who are watching their sons know that they are, but is rather that that other things being equal suitably situated people are in this position. Another possibility is to weaken the requirement in a different way and say that it is not that suitably situated people who are watching their sons are always such that they know that they are, but is rather that suitably situated people who see their sons are mostly such that they know that they are. These two ideas—the other things being equal idea and the mostly idea—are different, and it would take further discussion to decide which of them is right, or indeed to decide whether a suggestion distinct from both might not be preferable. I will not try to resolve the matter here. The main point is only that it does not seem impossible to accommodate cases of defeat by being careful to formulate (c), or the intention behind it, in the right way.14 Third, one might offer a simpler reason to reject the idea that it is a rational requirement that suitably situated people who are watching their sons know that they are, viz., that it is not plausible that rationality really cares about watching sons in particular. However, while it is implausible to think that rationality cares about watching sons in particular, it is not implausible to see the requirement we have been working with as a specific instance of a more general requirement. This is not to deny the difficulty in formulating the general requirement. For example, suppose one suggested that it is true of all mental states that, mostly, if one is in them, and one is suitably situated, one will know that one is in them. This will obviously not do. Some mental states—e.g., those postulated by psychology or linguistics, or by Freudian psychoanalysis—do not exhibit this property. Or suppose instead one suggested that it is true of all conscious mental states that, mostly, if one is in them, and one is suitably situated, one will know that one is in them. This is more plausible but is nevertheless subject to counterexamples. Consider cases in which one is completely absorbed in

14 One important consequence of weakening the rational requirement in the second way mentioned in the text is that if this is done, the position I am describing will not face the antiluminosity argument developed by Williamson (2000). If seeing my son were a luminous condition in Williamson’s sense, then in every case in which I satisfy that condition I would be in a position to know that I am in it. If this is true simply in most cases, however, seeing my son is not luminous in Williamson’s sense, even in a rational individual. For some discussion of this argument, see Silins, this volume, and Greenough, this volume.

82

Skepticism about Introspection

looking at a particular object; for example, suppose I am seeing my son doing somersaults and am totally absorbed in his doing somersaults. That state—the conjunctive state of seeing X do somersaults and being totally absorbed in X’s doing somersaults—is conscious in the main senses that dominate philosophy of mind. It is phenomenally conscious because there is something it is like to be in it, and it is access conscious because it controls rational thought and action. On the other hand, it is not true that if I am in it, I will know that I am in it on the condition that I am suitably placed and am rational. The reason is that if I ask myself whether I am in this state, I will inevitably shift some of my attention from my son’s doing somersaults to the question of whether I am in the state. But in that case, I will no longer be in the conjunctive state of seeing my son and being totally absorbed in doing so! So the state of seeing my son do somersaults and being totally absorbed in his doing somersaults does not have the property that the state of seeing my son does. Instead of saying that in a rational individual the property I have attributed to seeing my son is true of all mental states, or true of all conscious mental states, I think the better thing to say is that in a rational individual this property attaches to many states that the subject is in, without trying to specify the kind of states in question. So on this view it is a rational necessity that many mental states are such that that mostly when one is in them, and one is suitably placed, one will know that one is in them. This is certainly a bit of a mouthful. But one can capture the general idea more pithily by saying that it is a requirement of rationality that individuals have a capacity for expertise about their own first-person present-tense psychological states. Expertise does not mean omniscience. Expertise tolerates a certain level of ignorance and error. Someone can be an expert on Russian literature without knowing everything about Russian literature. And it need not be that there is some fact about Russian literature that every expert knows. Rather expertise means knowing a lot of facts about Russian literature, and perhaps knowing fairly deep organizational facts as well. Likewise, I am suggesting, a suitably situated rational subject can be an expert on their first-person present-tense psychological facts consistently with failing to know (and failing to be in a position to know) any few of these facts, and consistently with having some false beliefs about them. Finally, one might think it implausible that rational people should exhibit this sort of self-expertise in the first place. But as against this, I think there are examples that support it.15 Suppose George goes on Mastermind with the stated topic of

15

Examples such as these provide support for the idea that the self-expertise is a requirement of rationality, but they do not by themselves provide an explanation of why this is so. However, the question of why the self-expertise rule is a rule of rationality is not something I will attempt to answer in this chapter.

Knowledge of Perception

83

Russian literature. The compère asks various questions: “Did Tolstoy write Anna Karenina?” “Did Chekhov die young?” “Did Dostoyevsky get sent to Siberia?” and so on. George replies as best he can, but after a while a startling fact emerges: not only is George not an expert on Russian literature, he is an almost complete ignoramus. What explains George’s situation? Well it might be that he does not understand the questions, or that he is irrational, or that he is distracted. But in this case it need not be anything like this. Another possibility is that he simply has not had enough exposure to Russian literature and has gone on Mastermind only because of some false belief that he has—e.g., about what the rules of Mastermind are. An explanation along these lines makes sense of what George is doing, even though of course in another way we might wonder what on earth he was thinking. All right so far; but suppose now George goes on Mastermind with the stated topic of his own first-person present-tense psychological states. (A somewhat strange topic! But recall that just as the topic of Russian literature is in some sense a system of propositions, so too is the topic of George’s first-person present-tense psychological states.) The compère again asks various questions: “Do you see your son?” “Do you have a tingle in your elbow?” “Do you believe that there will be a third world war?” and so on. George replies as best he can, but once again a startling fact emerges: not only is George not an expert on his first-person present-tense psychological states, he is an almost complete ignoramus. What explains George’s situation this time? Well, this time there is no possibility of saying that George has not had enough exposure to his first-person present-tense psychological states. Either it makes no sense to talk of exposure in this case at all or if it does George has been exposed as much as it is possible to be. But if the lack of exposure possibility is ruled out, our set of options for explaining George’s situation is somewhat narrow. It remains possible of course that George does not understand the questions put to him, or that he is distracted. But assuming that we can rule out these hypotheses, it would seem that the only option left is to suppose that he is irrational. To put it another way: on the assumption that George is suitably placed, then either he has considerable expertise on the topic of his first-person present-tense psychological states or he is irrational. But he does not have such expertise. So he must be irrational.

7 I said earlier that our Austin-inspired response to the evidence argument had the right form but that there were two issues that needed to be confronted before it was adopted. The first, which we have just been discussing, is that the explanation for the fact that I know should itself be a rational explanation. The second, as we noted

84

Skepticism about Introspection

above, returns us to Dretske’s objection to the foot-stamping view, the objection that we “can’t do epistemology by stamping our feet.” We have so far made no real attempt to answer that objection. However, it might be insisted that we must make an attempt to answer it. For in developing our own response to the evidence argument, we have relied on the foot-stamping view. If you rely on something, you cannot very well note a major objection to it, and say next to nothing about how that objection might be met. How then might Dretske’s objection be met? Well, what is that objection? As we saw earlier, if the objection is that it is a mistake to assert that one knows without offering an identifiable way that one knows, it does not seem very forceful. For a proponent of the foot-stamping view (and so a proponent of the Austinian footstamping view) has an identifiable way, namely, the very proposition at issue. There is, however, a different way to take Dretske’s objection. On this different way of understanding it, the objection does not rely on the highly questionable claim that one cannot ever offer the very proposition known as evidence for what one knows. Rather it relies on the very plausible claim that normally one cannot do this; the objection then challenges the foot-stamper to explain why it can be done in this case. To put it differently, when Dretske says “we can’t do epistemology by stamping our feet” what I think he means (or at least is best taken to mean) is this: we do not normally do epistemology by stamping our feet; why then can we do it in this case? Dretske’s objection as understood here is closely related to an objection to the footstamping view that comes up in a dialectical context. When you normally ask “How do you know that p?” you are fishing around for a proposition distinct from p. Now a proponent of the foot-stamping view is committed to the idea that at least in certain contexts this is mistaken; at least sometimes it is appropriate to repeat the proposition p in answer to a question of that form. How can this be so if the question “How do you know that p?” presupposes that an answer to it is distinct from p? What a proponent of the foot-stamping view must hold here is that this presupposition is a pragmatic presupposition, something that is cancelled in the present (highly eccentric) case. However, even if that is right, it would seem that there is a challenge here nevertheless, viz., why is the presupposition cancelled in this case? Unless that further question is answered, we have as yet no good answer to the opponent of the foot-stamping view. It is this challenge that I think is the key idea behind Dretske’s objection. If that is the key idea behind Dretske’s objection, however, then the proposal I have advanced meets it. For on this proposal, there is a reasonable account of why the presupposition is cancelled. Suppose I know that I am seeing my son, and suppose that the explanation for this fact is that it is rational necessity that people who are seeing their sons know that they are, that I abide by this necessity, and that I am suitably situated and that I am seeing my son. Suppose now I am asked “How do

Knowledge of Perception

85

you know that you are seeing your son?” As we have seen, this question can be understood in various ways. If it is understood as a request for explanation for the fact that I know, the answer is the one I just gave. If it is understood as a request for evidence of a nontrivial sort, then it is a question whose presupposition I reject. There is (or need be) no evidence of a nontrivial sort that supports what I know. On the other hand, it is in this case reasonable of me to reject the presupposition. Mostly, if one fails to have evidence of a nontrivial sort for what one claims to know, this would be an epistemic embarrassment, for it would suggest intuitively that one has no rational basis for one’s claim to knowledge. But not in this case: for in this case I have arrived at my knowledge by abiding by rational principles. Finally, suppose the question is understood as a request for evidence of either a trivial or a nontrivial sort, i.e., as a question that permits a trivial answer. In that case, I can simply repeat the proposition that I know, and so provide that trivial answer. Once again, in the circumstances this is a reasonable thing for me to do. Mostly, if one can only provide evidence of a trivial sort for what one claims to know, this would be an epistemic embarrassment, for it again would suggest that one has no rational basis for one’s claim to knowledge. But not in this case: again in this case, I have arrived at my knowledge by abiding by rational principles.

8 The Austin-inspired foot-stamping response I have offered to the evidence argument alleges in effect that the argument exhibits a confusion of evidential and explanatory concerns. This response was of course always on the cards. For the evidence argument is framed in terms of “how do you know” questions, and such questions notoriously oscillate between these two readings. What I have tried to bring out in the course of our discussion, however, is that in drawing the evidence/ explanation distinction, it is crucial to acknowledge that there is a significant admixture of rationality in the explanatory side of the ledger. Unless one does that, one will so far as I can see have no response to the problem about natural and metaphysical necessity that I reviewed in sections 4 and 5. Nor will one have a response to Dretske’s objection to the foot-stamping view, which I mentioned in section 3 and again in section 7. I have also said a little bit about what the underlying requirement of rationality is, and in particular that it involves the idea that rational people exhibit a certain capacity for self-expertise. The picture of introspection that emerges from the response to the evidence argument might be summarized in the following terms. Suppose we say that I know that I am watching my son by introspection. What do we mean, i.e., what is

86

Skepticism about Introspection

the‘by’-phrase doing? Answer: at least in central cases, it is indicating that there is a distinctive answer to the question “how do I know?” What is distinctive about the answer is this: 1. If we fix it so that the ‘how’ question has its evidence-seeking reading, the answer will be trivial if there is one at all (and there will only be a trivial answer if I know merely by introspection). 2. If we fix it so that the ‘how’ question has its explanation-seeking reading, the answer will involve the idea that it is requirement of rationality that people exhibit a capacity for self-expertise, and that I am rational in at least this respect, and that I know what I do through an exercise of this capacity. This kind of picture leaves a lot of questions unanswered: (1) it does not say what the underlying scientific facts are; (2) it leaves open how extensive knowledge by introspection might be; (3) it does not (so far) take into consideration cases in which we arrive at introspective knowledge by inference; (4) it does not (so far) take into consideration cases in which I believe mistakenly that I am watching my son; (5) it does not speak to the sense in which the notion of introspection may be introduced as a mode of attention; (6) it does not connect up this notion introspection with consciousness in any of its many varieties; and finally (7) as we have seen it does not go very far into the analysis of what the requirement is exactly. All of these questions deserve detailed discussion, but we do not have the space for that here. Instead I will close the chapter by briefly contrasting the position I have outlined with what is an apparently different response to the evidence argument. This alternative response focuses, not on what is meant by the ‘how’ question, but rather on what is meant by ‘evidence’. In particular, the response says, there are two slightly different notions of evidence in play. The first notion is the one that we have been using all along. According to this notion, something is evidence just in case (a) I am aware of it, and (b) it supports what I know—we may call this evidence 1. But, says a proponent of this response, there is also a different notion of evidence that is sometimes in play in philosophical discussions. According to this notion, a proposition is evidence just in case the truth of the proposition makes it appropriate for me to believe what I know—we may call this evidence 2.16 16 In introducing the concept of evidence 2, I am following Pryor 2005, though his main focus is on justification rather than evidence. Likewise, the concept of evidence 1 is very closely related to an idea about evidence (and justification) which Pryor (2005, 189) calls “the Premise Principle” and states as follows: “The only things that can justify a belief that p are other states that assertively represent propositions, and those propositions have to be ones that could be used as premises in an argument for p. They have to stand in some kind of inferential relation to p: they have to imply it or inductively support it or something like that.”

Knowledge of Perception

87

Armed with this distinction, a different response to the evidence argument comes into view, viz., that it does not go through even if the question “how do I know” is given its evidence-seeking interpretation. For it might be that the evidence sought is in evidence 2 rather than evidence 1: e1. If I know that I am watching my son do somersaults, there is a nontrivial answer to the evidence 2-seeking question “How do I know that I am watching my son?” e2. If there is a nontrivial answer to the evidence 2-seeking question “How do I know that am watching my son?” that answer must be a proposition that (a) I am aware of; (b) “tells me” that I am watching my son; and (c) is distinct from the proposition that I am watching my son. e3. There is no proposition that meets conditions (a), (b), and (c). e4. Therefore, I do not know that I am watching my son. The problem with the argument in this version—version 4 of the evidence argument, as we may call it—is that e2 is false; the conditions set out in e2 will not apply. In particular, that I am watching my son is on this view a proposition that makes it appropriate to believe what I know, but it does not follow from this that I am aware of that proposition or of its truth.

9 How plausible is this ‘two notions of evidence’ response, and how does it compare with the response I have given? Clearly it is only a legitimate response if evidence 2 is a legitimate notion. Is it? Well, a major problem for evidence 2 is that it is very unconstrained. For example, there is a sense in which the total physical state of the universe at a certain time will make it appropriate to believe that I am watching my son at that time (if this is appropriate for me), but the total state of the universe is not my evidence on any reasonable conception of evidence. In response, what proponents of evidence 2 will say17 is that the fact that makes it appropriate for me to believe I am watching my son must be available to me. In other words, on this refinement of the notion, evidence 2 is a proposition (a) whose truth makes it appropriate for me to believe what I know and (b) which is available to me. So for example, if I am watching my son, and this fact is available to me, then it is appropriate for me to believe that I am watching my son.

17

Here again I am following Pryor (2005, 195).

88 Skepticism about Introspection At this point, however, the key question is this: What could this ‘availability’ come to? So far it is only a label for a relation I must bear to a fact if that fact makes it appropriate to believe that that fact obtains and is therefore my evidence. What is it? One answer explains availability directly in terms of awareness, and in turn explains this either in terms of perceptual or inferential model of introspection. However, first, this is an option that we set aside long ago (see section 2), and second, it makes it a bit of a mystery why the detour through evidence 2 was required. A second answer explains availability in terms of rationality. In fact this is exactly what Pryor—whose discussion of justification we have been adjusting to our own concerns—does when he explains the notion of availability. First, he connects availability and the idea of a ground: A justifier is available to you at a given time—it will be something you can “take as” a reason—if it’s something that could then ground a belief of yours. (Pryor 1995, 195) He then argues that the relevant notion of grounding can be explained in terms of following a rule or norm of rationality: What does it take for your belief to be grounded on some fact or condition C that you’re in? A natural thought is that your belief counts as so grounded iff it’s formed (or sustained) in a way that’s guided by the epistemic norm “When in C, believe p.” If that’s right, then the best way to understand the grounding relation is by inquiring into what it takes to be guided by such a norm. (ibid., 195) I do not want to assess this proposal in detail here. Rather I want simply to note that, if ‘availability’ is understood this way, then it looks as if the ‘two notions of evidence’ response is a variation on the response I have given, for a norm as Pryor intends it is just the notion of a requirement of rationality. In turn, if that is right, we are in the presence here, not of two responses to the evidence argument, but of one. On the other hand, if what I have argued is right, that response is successful.18

18 I gave (in some cases rather different) versions of this chapter to audiences at ANU, Melbourne, Oxford, and Sydney. I am very grateful to the extremely good suggestions I received from the audiences at those universities. I am also indebted to specific suggestions from Alex Byrne, David Chalmers, Mark Jago, John Maier, Tori McGeer, and Declan Smithies. I am very grateful also to two anonymous referees for Oxford University Press who gave extremely helpful comments on the chapter that (I think) improved it considerably.

Knowledge of Perception

89

References Armstrong, D. M. 1968. A Materialist Theory of the Mind. London: Routledge and Kegan Paul. Austin, J. L. 1961. “Other Minds.” Pp. 76–116 in Philosophical Papers. Oxford: Clarendon Press. Broome, J. 2005. “Does Rationality Give us Reasons?” Philosophical Issues 15: 321–337. ———. 2007. “Requirements.” In Homage a Wlodek: Philosophical Papers Dedicated to Wlodek Rabinowicz, edited by T. Ronnow-Rasmussen et al. Byrne, A. 2005. “Introspection.” Philosophical Topics 33: 79–104. Dretske, F. 2003. “How Do You Know You Are Not a Zombie?” Pp. 1–14 in Privileged Access: Philosophical Accounts of Self-Knowledge, edited by B. Gertler. Burlington: Ashgate Publishing. Fine, K. 2004. “The Varieties of Necessity.” Pp. 254–282 in Conceivability and Possibility, edited by T. Gender and J. Hawthorne. Oxford: Oxford University Press. Martin, M. G. F. 1998. “An Eye Directed Outward.” Pp. 99–121 in Knowing Our Own Minds, edited by C. Wright, Barry C. Smith, and Cynthia MacDonald. Oxford: Clarendon Press. Pryor, J. 2005. “There Is Immediate Justification.” Pp. 181–201 in Contemporary Debates in Epistemology, edited by Matthias Steup and Ernest Sosa. Oxford: Blackwell Publishing. Shoemaker, S. 1996. The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press. Siewert, C. 1998. “Self-Knowledge and Rationality: Shoemaker on Self-Blindness.” Pp. 131–146 in Privileged Access: Philosophical Accounts of Self-Knowledge, edited by B. Gertler. Burlington: Ashgate Publishing. Stoljar, D. 2004. “The Argument from Diaphanousness.” Pp. 341–390 in New Essays in the Philosophy of Language and Mind, supplemental volume of The Canadian Journal of Philosophy, edited by M. Ezcurdia, R. Stainton, and C. Viger. Calgary: University of Calgary Press. Williamson, T. 2000. Knowledge and its Limits. Oxford: Oxford University Press.

This page intentionally left blank

Part II Theories of Introspection

This page intentionally left blank

4 Renewed Acquaintance Brie Gertler

Pinch yourself (gently). By focusing your attention on the phenomenal quality of the sensation that results, you can come to know something about your current experience. Philosophers generally agree on this much. Yet there is widespread and profound disagreement about what this kind of knowledge consists in, and how it is achieved. These issues have recently gained prominence because of their perceived significance for the question of physicalism about the mind.1 My concern here is not, however, with that question, but only with the nature of this introspective knowledge itself. I will elaborate and defend a set of metaphysical and epistemic claims that comprise what I call the acquaintance approach to introspective knowledge of the phenomenal qualities of experience. The hallmark of this approach is the thesis that, in some introspective judgments about experience, (phenomenal) reality intersects with the epistemic, that is, with the subject’s grasp of that reality. This thesis—or something close to it—is implied by the claim that we sometimes grasp our experiences directly, by using an experience’s defining phenomenal quality to form an epistemically substantive conception of the experience itself. BonJour (2003), Chalmers (2003), Fales (1996), Feldman (2004) Fumerton (1995), Gertler (2001), Horgan and Kriegel (2007), and Pitt (2004) give accounts of introspection along these lines.

1

Awareness of the phenomenal qualities of experience plays a central role in the leading arguments for mind–body dualism—the modal argument (Kripke 1972), the knowledge argument (Jackson 1982), and the zombie argument (Chalmers 1996). Several of the chapters in Alter and Walter (2007) discuss the ontological implications of views about phenomenal knowledge.

94 Theories of Introspection The acquaintance approach is a set of claims about our introspective knowledge of the phenomenal that captures the spirit of these accounts. This is not to say that all of those just mentioned will accept each of the claims that comprise the acquaintance approach. My goal is to explicate and defend the core thesis shared by these views, namely, that a grasp of a current experience can represent an intersection between phenomenal reality and the epistemic. But some of those just listed may take issue with my way of unpacking this thesis and my means of defending it. The acquaintance approach is inspired by Russell’s theory of acquaintance, from which it derives its name. But the approach diverges from that theory in significant ways. In section 1 of the chapter I outline the acquaintance approach by drawing on its Russellian lineage. A more detailed picture of the approach emerges in succeeding sections, which respond to a range of objections. Some critics charge that approaches of this sort are overly idealized, in that they ignore the cognitive flaws and limitations of actual human beings. I begin to address these worries in section 2, by arguing that the epistemic commitments of the acquaintance approach are in fact relatively modest. In section 3, I sketch a picture of introspective reference that explains how phenomenal reality can intersect with the epistemic in a phenomenal judgment, as the acquaintance approach requires. Drawing on this picture of introspective reference, section 4 sets out a practical strategy for achieving knowledge by acquaintance. Some contemporary acquaintance theorists (BonJour 2003, Fumerton 1996) employ demanding epistemic standards for knowledge by acquaintance, standards beyond those mandated by the acquaintance approach. In section 5 I show that instances of introspective knowledge that meet less demanding standards can satisfy the acquaintance approach’s epistemic commitments. The final sections concern the most direct challenges to the acquaintance approach, which target the claim that phenomenal reality intersects with the epistemic. According to one such challenge, this claim is belied by the fact that possessing a phenomenal concept is a matter of having certain dispositions. Section 6 draws on a discussion by Sosa (2003) to articulate this challenge, and responds to it on behalf of the acquaintance approach. Section 7 addresses Stalnaker’s (2008) worry that, if phenomenal reality intersected with the epistemic, phenomenal information would be incommunicable.

1. The Acquaintance Approach The acquaintance approach is inspired by Russell’s claim that we are acquainted with certain aspects of mentality. Here is Russell’s characterization of acquaintance.

Renewed Acquaintance

95

We shall say that we have acquaintance with anything of which we are directly aware, without the intermediary of any process of inference or any knowledge of truths. (Russell 1912, 73) To determine whether you are acquainted with an object, Russell says, you should consider whether you can doubt the object’s existence. If you cannot, your awareness of the object is direct; you are therefore acquainted with the object. Russell does not provide an analysis of the acquaintance relation. Perhaps on his view this relation can be fully grasped only through acquaintance (ibid., 79). But it seems clear that Russellian acquaintance has both an epistemic and a metaphysical dimension. When I am acquainted with an object, my awareness of that object is epistemically direct: it is noninferential and does not epistemically depend on an awareness of anything else. My awareness is also metaphysically direct: there is no object, fact, event, or process that mediates my access to the object. Russell’s picture suggests that, if my access to the object were not metaphysically direct, I would be able to doubt the object’s existence. For instance, on Russell’s view my awareness of the table I see before me is metaphysically indirect; it is mediated by a causal process (involving light reflecting off of the table, striking my retina, and causing a visual experience). The presence of this mediating factor enables me to doubt the existence of the table, since I can recognize that, for all I know, my visual experience has an aberrant cause. So while the criterion for acquaintance is indubitability—an epistemic and/or psychological phenomenon— the acquaintance relation itself is both epistemic and metaphysical. The idea that awareness can be both metaphysically and epistemically direct is the basis for the acquaintance approach developed here. But that approach diverges significantly from Russell’s theory. The acquaintance approach is exclusively concerned with introspective knowledge, whereas Russell’s theory has a much broader scope.2 The acquaintance approach takes introspective knowledge to consist in occurrent judgments, whereas on Russell’s theory knowledge by acquaintance is a nonpropositional knowledge of things. And the acquaintance approach remains neutral on some especially controversial aspects of Russell’s view, including the commitment to sense data, the idea that acquaintance is the foundation of all knowledge, and the claim that acquaintance with an object suffices for knowing it “perfectly and completely” (ibid., 73).

2 According to Russell, introspective knowledge by acquaintance constitutes a relatively insignificant portion of our knowledge by acquaintance, even within the empirical realm (that is, putting aside our knowledge of abstract universals). The majority of such knowledge arises through ordinary perceptual experiences. Every perceptual experience involves sense data, and every sense datum is known by acquaintance.

96 Theories of Introspection The acquaintance approach’s defining claim is that some introspective knowledge possesses modified versions of the characteristics Russell ascribes to knowledge by acquaintance. As we will see, even this sharply restricted position is highly controversial. The acquaintance approach adopts, in qualified form, three key elements of Russell’s theory. The first element concerns the metaphysical dimension of the acquaintance relation. On Russell’s view, acquaintance is a relation between subjects and things known: in the case of introspective knowledge of mental states, the objects of acquaintance are “events which happen in our minds” (ibid., 77). But because the acquaintance approach takes knowledge of phenomenal states to be knowledge of truths rather than knowledge of things, on that approach the salient direct relation obtains between a judgment—such as the judgment I am experiencing pain—and the mental event that is its truthmaker, e.g. the pain experience itself.3 Russell does not explain what, precisely, his “directness” requires. But it is clear from his remarks about perceptual awareness that no merely causal link between a subject and an object is direct in the sense required for acquaintance. The acquaintance approach similarly contends that introspective knowledge is sometimes metaphysically direct, in that the relation between an introspective judgment and its truthmaker is not merely causal. This is the first defining commitment of the acquaintance approach. Direct Tie to Truthmakers; Some introspective knowledge consists in judgments that are tied to their truthmakers directly, where the relevant notion of directness is metaphysical and not (merely) causal. How might an introspective judgment like I am experiencing pain be directly tied to its truthmaker? Before answering this question, a brief note about the ontology of experiences is in order. I take experiences to be events: namely, instantiations of phenomenal properties in subjects at times. So my current pain experience is an instantiation of the phenomenal property pain in BG at time t. On this construal, the phenomenal property pain is strictly a property of the subject, rather than a property of an experience. But for convenience I will usually speak of experienced phenomenal properties, or phenomenal properties of experience. “I am now

3

This is not to say that the contemporary acquaintance theorist must reject Russell’s claim that knowledge by acquaintance is underwritten by a direct relation between the subject and the object of knowledge. But that relation will not render introspective knowledge direct, in the relevant sense, unless it is reflected in a suitably direct relation between the introspective judgment that constitutes knowledge and its truthmaker.

Renewed Acquaintance

97

experiencing pain” or “this experience is painful” can be parsed as “pain is instantiated (in me, now).”4 Now to the question of how a judgment such as I am experiencing pain could be directly tied to its truthmaker. The leading possibility is that the truthmaker—the subject’s instantiation of the phenomenal property pain at a time—directly supplies part of the judgment’s content. This general idea is accepted by numerous philosophers, some of whom reject other commitments that define the acquaintance approach. For instance, Loar ([1990] 1997) says that an experience can directly contribute to the content of a judgment concerning it, by serving as its own mode of presentation (compare Horgan and Kriegel 2007). Others say that an experience embedded in a judgment can directly contribute to the content of that judgment (Gertler 2001), or that an experience can be directly responsible for constituting the content of a judgment (Chalmers 2003), or that an experience can be used to refer to the phenomenal property it exemplifies (Papineau 2007), or that an experience can be cognitively present within a judgment (Levine 2007).5 Of course, much remains to be said about how, precisely, an experience can directly supply part of a judgment’s content. I address this issue in later sections. Any judgment that is directly tied to its truthmaker will be true. But this fact carries no immediate epistemic implications. After all, any judgment causally tied to its truthmaker will also be true, though such judgments sometimes fall short of knowledge. I do not know that an evil genius is controlling my thoughts even if my judgment to that effect is caused by its truthmaker. This brings us to the remaining two elements of Russell’s theory that contribute to the acquaintance approach, which are epistemic. Because Russell denies that knowledge by acquaintance is knowledge of truths, his epistemological position is somewhat difficult to spell out. For instance, we cannot isolate the epistemic basis for knowledge by acquaintance by specifying how such knowledge differs from merely true belief, since on Russell’s view knowledge by acquaintance does not consist in belief or, for that matter, anything that bears a truth value. 4 When appropriately based on introspection, judgments like pain is instantiated (in me, now) are arguably immune to error through misidentification of the first-person pronoun (Shoemaker 1968). Moreover, one who judges that pain is instantiated (in me, now) can conceive of herself indexically, as the thing in which the experience occurs: in that case, “in me” has the force of “here” or “in this subject of experience.” Howell (2006) advances an introspectivist account of self-reference along these lines. 5 See also Balog (forthcoming). Some of these accounts specifically concern phenomenal concepts. The connection to judgments stems from the fact these concepts are exercised in introspective judgments.

98 Theories of Introspection The best way to identify Russell’s epistemological position as regards knowledge by acquaintance is to ask what makes acquaintance with an object sufficient for knowing the object. That is, what makes this relation to an object truly epistemic rather than, say, a matter of brutely metaphysical contact with the object? Russell’s answer seems to be that acquaintance with an object involves—or perhaps simply consists in—that object’s being immediately present to consciousness, where such presence is an epistemic matter. This is not very illuminating as an analysis of acquaintance. But the salient point for our purposes is that Russellian knowledge by acquaintance is epistemically grounded exclusively in the presence of certain objects to consciousness.6 The acquaintance approach endorses a qualified version of this view, as regards some introspective knowledge. It says that some introspective knowledge is epistemically grounded exclusively in the subject’s conscious states. More precisely: the doxastic justification for some introspective judgments that qualify as knowledge consists in conscious states alone. (These may include conscious states that constitute the subject’s awareness—e.g., of an introspected pain.) This does not mean that the knowing subject must be aware of the fact that her conscious states play this justifying role. The salient claim, which constitutes the second commitment of the acquaintance approach, is as follows. Justified by Conscious States: Some introspective knowledge consists in judgments that depend, for their justification, only on the subject’s conscious states at the time of the judgment. The final element of Russell’s theory that contributes to the acquaintance approach concerns the epistemic security of knowledge by acquaintance. Recall Russell’s claim that, because perceptual knowledge rests on a causal process that renders perceptual awareness indirect, we can doubt the existence of perceptual objects. Russell concludes that no perceptual knowledge is as certain (epistemically secure) as knowledge by acquaintance. More generally, he contends that knowledge by acquaintance is more certain than any other type of knowledge. The acquaintance approach is not committed to saying that introspective knowledge is absolutely certain, or that every instance of introspective knowledge is especially secure. But it does say that, among introspective judgments that satisfy 6 This explains Russell’s hesitation about whether the self can be known by acquaintance: he is hesitant about this question precisely because he is not certain that factors immediately present to consciousness epistemically suffice for knowledge of the self. “So far as immediate certainty goes, it might be that the something which sees the brown colour is quite momentary, and not the same as the something which has some different experience the next moment. Thus it is our particular thoughts and feelings that have primitive certainty” (Russell 1912, 29–30).

Renewed Acquaintance

99

the acquaintance approach’s first two commitments, at least some are more strongly justified—possess a greater degree of doxastic justification—than any empirical judgments that do not satisfy these commitments. That is: Especially Strong Justification: Some introspective judgments are more strongly justified than any empirical judgments that are not directly tied to their truthmakers or that depend, for their justification, on factors other than the subject’s conscious states at the time of the judgment. A brief note about justification. The notion of justification here is neutral between various conceptions: as a starting point, it can be thought of simply as “the sort of justification that, in sufficient strength, is a necessary condition for knowledge” (Conee and Feldman 2004, 54). Of course, since the introspective judgments in question are justified exclusively by conscious states, the justification in question will meet the requirements of “mentalist” internalism (ibid). And insofar as these conscious states are accessible to the subject, it will also meet the requirements of (at least some versions of ) “accessibility” internalism. However, endorsing the acquaintance approach does not commit one to any particular conception of justification. The acquaintance approach concerns only a limited class of introspective judgments; so one could endorse this approach while allowing that judgments outside that class are justified by factors external to the mind and inaccessible to the subject. And even as concerns the class of introspective judgments at issue, this approach is compatible with a range of answers to the question of how the conscious states at issue justify these judgments. (We will briefly examine two competing answers to this question in section 5.) So the current claim, that some introspective judgments are especially strongly justified, is cashed out in different ways by contemporary proponents of the acquaintance approach, corresponding to differences in their conceptions of epistemic justification. I have isolated three elements in Russell’s picture of knowledge by acquaintance.7 These three elements of Russell’s picture, when suitably modified, define the contemporary acquaintance approach to introspective knowledge. The defining thesis of the acquaintance approach is as follows. Acquaintance Approach: Some introspective knowledge consists in judgments that 1. are directly tied to their truthmakers; 2. depend, for their justification, only on the subject’s conscious states at the time of the judgment; and 3. are more strongly justified than any empirical judgments that do not meet conditions (1) and (2). 7

For a more detailed discussion of Russell’s position, see Gertler 2011, ch. 4.

100 Theories of Introspection I will use the term “knowledge by acquaintance” to refer to knowledge consisting in judgments that satisfy conditions (1)–(3).8 So the distinctive thesis of the acquaintance approach is that some introspective knowledge is knowledge by acquaintance. The defining thesis of the acquaintance approach is existential in form. The approach is vindicated so long as some introspective knowledge constitutes knowledge by acquaintance; introspective knowledge by acquaintance may be achieved only rarely, and may be possible only as regards a narrow class of states. The existential form of the acquaintance approach’s defining thesis means that this approach cannot be refuted by counterexamples. Knowledge of phenomenal states is widely—and, I believe, rightly—considered the most plausible candidate for knowledge by acquaintance. (For this reason, phenomenal states will be my exclusive focus.) But the acquaintance approach is highly controversial even as regards introspective judgments about phenomenal states. For instance, contemporary versions of the inner sense theory imply that no introspective judgments meet conditions (1) or (2). (See Armstrong [1968] 1993; Goldman 2006; Lycan 1996; Nichols and Stich 2003; for a critical discussion of the inner sense theory see Gertler 2011, ch. 5.) A central tenet of these views is that the relation between an introspective judgment and its truthmaker is essentially similar to the relation between perceptual judgments and their truthmakers: that is, it is a causal relation. Since these inner sense theorists take introspective knowledge to be justified (or warranted) by the reliability of this causal connection between judgments and truthmakers, they also deny that introspective justification depends only on conscious states.9 Others reject the idea that introspective judgments are especially well justified, as compared with perceptual judgments, purely on the basis of epistemic considerations.

8 I have not identified a particular relation that is the analogue of Russellian acquaintance. The relation between an introspective judgment and its truthmaker, mentioned in condition (1), is not Russellian acquaintance, since Russellian acquaintance is a relation between a subject and something else. A proponent of the contemporary acquaintance approach might accommodate this restriction by claiming that, when an introspective judgment satisfies conditions (1), the subject is acquainted with the mental state it concerns. I will remain neutral on that issue. My use of the term “knowledge by acquaintance” is simply intended to register that this thesis about (some) introspective judgments is Russellian in spirit. 9 And since these inner sense theorists deny that such judgments satisfy (1) or (2), they also deny that these judgments satisfy (3). However, inner sense theorists generally allow that introspective judgments may be more epistemically secure than other empirical judgments. On their view, this difference derives from the fact that introspective detection mechanisms may be more reliable than the causal mechanisms involved in perception or memory.

Renewed Acquaintance

101

I suspect [that] . . . [o]ur judgments about the world to a large extent drive our judgments about our experience. Properly so, since the former are the more secure. (Schwitzgebel 2008, 268) Even among those who accept that introspective knowledge can be especially secure, many will deny that conscious states can supply justification (or warrant) sufficient for knowledge. That idea is at odds with those brands of epistemic externalism that take justification or warrant to consist in certain regularities, such as those involved in the tracking relation. The acquaintance approach takes the doxastic justification for an introspective judgment to consist in events—namely, the occurrence (or presence) of certain conscious states—and these events cannot constitute or entail such regularities. So the acquaintance approach implies that justification does not always consist in such regularities. Relatedly, the acquaintance approach implies that purely dispositional features of the subject, which are not reflected in her conscious states, play no essential role in justifying these introspective judgments. This consequence runs afoul of a widely held view about phenomenal concepts, namely that the possession and justified exercise of a phenomenal concept (like pain) at least partly consists in dispositions to recognize states falling under it. (I discuss this issue in section 6.) Perhaps the most controversial commitment of the acquaintance approach is the claim that reality can intersect with the epistemic: that is, that one’s epistemic grasp of a bit of reality—a fact, event, or property—can be partly constituted by that reality itself. This claim will be rejected by those who accept certain semantic views according to which reality is grasped only through representations, either linguistic or nonlinguistic, which mediate between minds and reality. Such views are sometimes motivated by the assertion that thoughts depend, for their representational content, on relations to public objects. (Section 7 responds to an objection along these lines.) A final source of objections to the acquaintance approach concerns its alleged antiphysicalist implications.10 I will not address this issue here. The acquaintance approach is, then, highly controversial. It conflicts with prominent views about the processes involved in introspection, the nature of knowledge, and mental semantics; it may also threaten physicalism. I cannot hope

10 Here is one particularly clear statement of the view, which is held by numerous philosophers on both sides of the physicalism–dualism debate. “[V]arious theorists have suggested that we are ‘acquainted’ with our phenomenal states, or that tokens of phenomenal properties themselves ‘partially constitute’ our concepts of them, or that we possess a concept-forming mechanism that somehow ‘quotes’ these property-tokens themselves when we think about them. Physicalists, however, should resist these suggestions” (Levin 2007, 97).

102 Theories of Introspection to establish the acquaintance approach here, as that would require showing that the approach is preferable to competing views about introspection. But given the range and significance of the approach’s controversial implications, demonstrating that key obstacles to the approach can be overcome will constitute genuine philosophical progress. I embark on this project in the next section.

2. The Limited Epistemic Commitments of the Acquaintance Approach Perhaps because of their Cartesian–Russellian ancestry, acquaintance-type approaches to introspective knowledge are often associated with ambitious epistemic claims—for instance, that we are infallible or omniscient about our own phenomenal states. But the acquaintance approach is not committed to these claims. The acquaintance approach is defined by an existential thesis: namely, some introspective knowledge is knowledge by acquaintance. That thesis does not imply that every experience is introspectively accessible. In fact, it does not suggest that a subject’s having an experience—even an experience that is of a type sometimes known by acquaintance—bears any substantial epistemic implications for her. So the acquaintance approach is not committed to the claim that some mental states are “self-presenting” in that anyone who has such a state, and considers whether he does, will self-attribute it (Chisholm 1982). It is even compatible with the idea that some persons never achieve introspective knowledge by acquaintance, because they abstain from—or perhaps are incapable of—forming judgments that satisfy (1)–(3). An analogy will help to illuminate these points. Consider a traditional rationalist claim about mathematical knowledge: some mathematical judgments are based in pure reason and are therefore more strongly justified than any empirical judgments. Those who accept this claim can recognize that we are fallible as regards our mathematical judgments. They can admit that certain mathematical questions are simply too complex for us to resolve (using reason alone), and that questions that could be settled by reason are sometimes addressed through empirical means, as when one relies on a memory of a multiplication table. They can also allow that attempts to arrive at a mathematical judgment by reason alone do not always succeed: one may be careless in one’s calculations or biased by antecedent expectations about the result. So claiming that some judgments based in a particular method (such as pure reason or introspection) are very strongly justified does not commit one to saying that subjects are generally reliable in their judgments about issues to which that method is applicable (mathematical truths, one’s own current experiences).

Renewed Acquaintance

103

Recognizing these limits to the acquaintance approach’s epistemic commitments will defuse some historically prominent objections. For instance, since the acquaintance theorist (a proponent of the acquaintance approach) need not claim that every experience is introspectible, she can neutralize the famous speckled hen argument. According to the speckled hen argument, when I see a hen with 48 clearly visible speckles on its facing side, my visual experience will differ, phenomenally, from the experience I have when seeing a similar hen with 47 speckles. But I may be unable to introspectively discern this phenomenal difference. The speckled hen argument does not threaten the acquaintance approach, since that approach does not imply that every phenomenal state can be introspectively known. The acquaintance theorist should allow that “phenomenal properties can outstrip a subject’s ability to make justified judgments about them.” (Fantl and Howell 2003, 380) Our introspective powers are simply unequal to the task of discriminating highly complex phenomenal properties. (Compare: our powers of reason are unequal to the task of resolving highly complex mathematical questions.) Another objection to acquaintance views is based on the claim (made by Hill 1991, among others) that attempts at introspection can alter the target experience. Stepping back from the epistemic ambitions traditionally associated with such views enables us to accommodate this objection as well. Imagine a radical possibility along these lines. Suppose that certain kinds of experiences are systematically altered by attempts at introspection: whenever one tries to form an introspective judgment about (what is actually) a phenomenal F experience, that experience ceases. In this scenario, we may be unable to achieve introspective knowledge of our F experiences. But this does not threaten the acquaintance approach, since we might nevertheless know some of our experiences by acquaintance. (Compare: a rationalist theory of mathematical knowledge is not threatened by the fact that certain mathematical truths are inaccessible to human reason.) Moreover, attempts at introspection that alter the target state may nonetheless yield knowledge by acquaintance, though not of the original experience. The introspective judgments that are candidates for knowledge by acquaintance concern one’s current experience, that is, the experience one has at the time of the judgment itself. Such judgments are noncommittal as to whether that experience began before one engaged in introspection, and as to what caused it. While limiting the acquaintance approach’s epistemic pretensions allows it to sidestep some familiar objections, this move also prompts new worries. The main epistemic commitments of the acquaintance approach are that some introspective judgments are very strongly justified, and depend, for their justification, only on conscious states. But some critics contend that these epistemic features require a more general connection between judgments and truthmakers. They claim, for

104 Theories of Introspection example, that a single introspective judgment to the effect that one is in pain cannot qualify as knowledge—or even as a genuine exercise of the concept pain—purely in virtue of one’s conscious states. One must also have certain nonconscious dispositions, for instance to recognize pains across a suitable range of circumstances. This type of requirement could be met by reinstating some of the weighty epistemic claims that I have put aside. For example, we might ensure that subjects have the appropriate dispositions by maintaining that pains and other phenomenal states are self-presenting, in that anyone who is in pain and has the conceptual resources to consider “am I in pain?” will recognize that he is in pain.11 Since my defense of the acquaintance approach rests on avoiding epistemic commitments of this sort, the charge that knowledge requires this kind of general connection between judgments and truthmakers presents a challenge for my project. I address that challenge in later sections. My next task is to explain how phenomenal reality and the epistemic can intersect in a judgment.

3. Demonstrative Attention and Epistemic Appearances The most promising way of explaining how phenomenal reality can intersect with the epistemic, in an introspective judgment, holds that subjects can refer to the phenomenal qualities of experience with introspective demonstratives. A brief review of the more familiar type of demonstratives, perceptual demonstratives, will provide a useful introduction. The referent of a perceptual demonstrative is the object whose presence and properties causally contribute (in an appropriate way) to the relevant aspect of how things seem to the subject.12 Consider the following modification of a famous example from Donnellan (1966). I point in a certain direction and say “that man drinking a martini is nattily dressed.” My demonstrative “that man drinking a martini” refers to a particular man—Mr. Smith, say—because it is Mr. Smith’s appearing in my line of vision and holding a martini glass that make the appropriate causal contribution to its seeming to me that there is a man over there drinking a martini. As Donnellan notes, 11

Although the acquaintance approach implies that concept possession or knowledge does not consist in sensitivity, if sensitivity is construed dispositionally. More on this in section 6 below. 12 I am concerned here only with so-called “speaker’s reference.” And this is only a rough, preliminary description of demonstrative reference. A fuller theory would spell out various details, such as what makes a causal contribution “appropriate” in the relevant sense.

Renewed Acquaintance

105

Mr. Smith can make this contribution even if he is not actually drinking a martini. The crucial point is that it is his presence and properties that appropriately contribute to its seeming to me that there is a man drinking a martini, and this might be true even if his martini glass contains only water.13 In this sense, perceptual demonstratives are grounded in (an aspect of ) how things seem to the referring subject.14 Like perceptual demonstratives, introspective demonstratives are also grounded in how things seem to the subject. And as in the perceptual case, the referent of my introspective demonstrative is the thing that appropriately contributes to how things seem to me. But on the version of the acquaintance approach I have in mind, there is a key difference between perceptual and introspective demonstratives (of the type which concerns us15). In introspective demonstration the referent’s contribution to how things seem is not, or not merely, causal. When you pinch yourself and attend to the sensation that results, it seems to you that your experience has a certain phenomenal property. This aspect of how things seem to you can ground your demonstrative reference to the property, as “this phenomenal property.” The referent of this demonstrative is (let us assume) the phenomenal property pinching, since that property is what appropriately contributes to this aspect of how things seem to you. But whereas in the previous example Mr. Smith’s presence and properties causally contributed to how things seemed, in this case the nature of the phenomenal property (that it is pinching rather than, say, tickling), and its instantiation in your experience, constitutes the relevant aspect of how things seem. Two questions about this type of introspective demonstrative reference immediately arise. First, what are the referents of introspective demonstratives? And second, what kind of “seeming” grounds these demonstratives? We should be liberal about the referents of introspective demonstratives, just as we are liberal about the referents of perceptual demonstratives. Depending on the subject’s referential intentions, a perceptual demonstrative might refer to an object, 13

There are, however, limits as to how much the referent can diverge from one’s conception. My judgment the man drinking a martini is nattily dressed would lack a referent if it were the presence of a naked penguin holding a martini glass that explained its seeming to me that there was a man there, drinking a martini. If I voiced this judgment, it would be plainly inappropriate for someone to respond “No, he’s not—he’s naked!” (Whether a perceptual demonstrative refers, in a given case, may depend on context and pragmatic matters.) 14 Of course, the demonstrative term “that man drinking a martini” is unusual in that it expresses the relevant aspect of how things seem. A less descriptive term like “that guy” may be similarly grounded, if the subject mentally singles out Mr. Smith as the man drinking a martini (over there). 15 The acquaintance theorist can allow that some types of introspective demonstratives share the causal structure of perceptual demonstratives: e.g., that I can demonstratively refer to an event as the cause of an experience I am now introspecting.

106 Theories of Introspection property, or event. For instance, I might refer to that man drinking a martini (an object), or that color of a martini (a property), or that spilling of a martini (an event). Similarly, an introspective demonstrative might refer to a mental object such as a sense datum (if there are such), a mental property, or a mental event. My discussion here will focus on introspective demonstrative reference to properties—in particular, phenomenal properties. Such reference can occur in a judgment of the form this property is instantiated (in me, now), where “this” refers to a phenomenal property. So the proposal is this. When I attend to the feel of my pinching experience, that experience—the instantiation of pinching—constitutes an aspect of how things seem to me. This aspect of how things seem to me grounds my demonstrative reference to the phenomenal property, when (while attending to the experience) I judge that this property is instantiated (in me, now). The demonstrative “this property” refers to the phenomenal property (pinching) whose instantiation constitutes that aspect of how things seem to me. (Compare: in the perceptual case, “that man” refers to the man whose presence and properties causally contribute to the relevant aspect of how things seem to me.) In the judgment this property is instantiated (in me, now), the term “this” expresses what Chalmers calls a “direct phenomenal concept.” The clearest cases of direct phenomenal concepts arise when a subject attends to the quality of an experience, and forms a concept wholly based on the attention to the quality, “taking up” the quality into the concept. (Chalmers 2003, 235) I have said that an aspect of how things seem to me grounds the demonstrative “this” in my judgment this property is instantiated (in me, now). Notice, however, that that judgment also expresses the relevant aspect of how things seem to me, e.g., that I am now having a pinching experience. This is precisely the aspect of how things seem that is constituted by my experience itself. So my judgment this property is instantiated (in me, now) is directly tied to its truthmaker, my pinching experience. It thus appears to satisfy condition (1) of the acquaintance approach. But this last claim invites an objection, concerning the nature of the “seeming” at issue. Philosophers who speak of “appearances” or “seemings” in discussing consciousness invite conflation of the epistemic and phenomenal senses of these terms. They thus risk breathing an illegitimate air of indefeasibility into our reflections about phenomenology. “It appears that it appears that such-and-such” may have the look of redundancy, but on disambiguation the redundancy vanishes: “it epistemically seems to me that my phenomenology is such-and-such.” No easy argument renders this statement self-verifying. (Schwitzgebel 2008, 263)

Renewed Acquaintance

107

This distinction between epistemic and phenomenal appearance threatens the proposal just sketched. How things phenomenally seem to me is simply a matter of the phenomenal reality: that is, of the phenomenal properties I instantiate. But what my judgments express, and what grounds my demonstratives, is how things seem to me epistemically: roughly, what I believe, or am inclined to believe. The “air of indefeasibility” Schwitzgebel mentions stems from a familiar thought: that “in the case of phenomenal consciousness there is no gap between appearance and reality, because the appearance just is the reality: how the phenomenal character seems, to the agent, is how it is” (Horgan, this volume). This thought is plausible as regards phenomenal appearances, since how an experience phenomenally appears just is its phenomenal reality. But as both Schwitzgebel and Horgan note, it is mistaken as regards epistemic appearances. For instance, what epistemically appears to me to be a 47-speckledness experience may, in (phenomenal) reality, be a 48-speckledness experience. What I am inclined to believe about the experience may diverge from its phenomenal reality. Schwitzgebel’s objection suggests that introspective judgments express how things epistemically appear, not how they phenomenally appear. Since there is a gap between epistemic appearance and phenomenal reality, such judgments are not directly tied to the phenomenal reality itself. According to this objection, then, they are not directly tied to their truthmakers. One might respond to this objection by maintaining that introspective judgments, and the demonstratives they involve, differ from perceptual judgments and demonstratives. Whereas perceptual judgments express (and perceptual demonstratives are grounded in) epistemic appearances, introspective judgments express (and introspective demonstratives are grounded in) phenomenal appearances. However, this response is not promising. Denying that introspective demonstratives express how things epistemically seem would cast doubt on the cognitive significance of introspective judgments. If an introspective judgment does not express how things seem to me epistemically, it is hard to see how the judgment could have genuine cognitive significance for me, or why it should even be regarded as a judgment. The acquaintance theorist should, then, acknowledge that it is epistemic appearances that are expressed in introspective judgments and that ground introspective demonstratives. Moreover, she should accept that epistemic appearances sometimes diverge from the phenomenal reality, as Schwitzgebel’s objection implies. Since the defining commitment of the acquaintance approach is an existential thesis about some instances of introspective knowledge, that approach requires only that epistemic appearances sometimes converge with phenomenal reality. The acquaintance theorist should respond to Schwitzgebel’s objection by saying that when an introspective judgment about experience qualifies as knowledge by acquaintance, the aspect

108 Theories of Introspection of how things epistemically seem that is expressed in that judgment is constituted by how they phenomenally seem—that is, by the phenomenal reality. According to this response, some introspective judgments about the phenomenal qualities of experience play dual, epistemic-and-metaphysical roles. They express how things epistemically seem to the subject, and they are directly tied to their truthmakers, the phenomenal reality they concern. (As we will see in sections 6 and 7, many who reject the acquaintance approach do so precisely because they deny that a judgment can play these dual roles.) So the acquaintance theorist should say that, while there is sometimes a gap between epistemic appearances and phenomenal reality, in some introspective judgments the epistemic intersects with phenomenal reality. To understand how this could work, let us consider in more detail the kinds of factors that can affect epistemic appearances. Factors shaping how things epistemically appear include: one’s situation, relative to the object or state of aff airs in question; the direction of one’s attention; and one’s background beliefs. These factors clearly contributed to its seeming to me that there was a man (over there) drinking a martini. Mr. Smith was in my line of vision; I directed my attention towards him; and I had the background belief that glasses of a certain shape usually contain martinis. These factors can also shape how experiences epistemically appear. My position relative to an experience—most importantly, whether the experience is mine or someone else’s—plainly shapes how it appears to me. And the direction of my attention has an obvious effect as well. When I turn my attention to the taste of the coffee I have been absentmindedly drinking, the epistemic appearance of this gustatory experience may grow more nuanced.16 Background beliefs also affect how sensations epistemically appear, though their influence may be less obvious. Consider the blindfolded fraternity pledge who is told that his throat will be cut by a razor (Shoemaker 1996). When an icicle is held against his throat, he cries out, apparently believing that he is in pain. The background expectation of pain shaped how his experience epistemically seemed to him. Assuming that he was not actually in pain, this epistemic appearance diverged from the phenomenal reality.17

16 While attention may change the phenomenology of an experience, phenomenal features sometimes withstand a shift in attention, as when one notices phenomenal features of one’s coffeedrinking experience that, while previously present, had gone unnoticed. 17 This is only one interpretation of the fraternity case. The important point is that background beliefs sometimes have this effect, not that this is the best interpretation of the fraternity case in particular.

Renewed Acquaintance

109

So the key task, for one who would achieve knowledge of an experience by acquaintance, is to bring epistemic appearances into line with phenomenal appearances and, thereby, with phenomenal reality. The next section explains how one might go about this.

4. Achieving Knowledge by Acquaintance I now outline a practical strategy for achieving introspective knowledge by acquaintance. This strategy draws on another factor that shapes epistemic appearances: the subject’s degree of credulousness. If I had been in a skeptical mood, the sensory experiences, background beliefs (etc.) involved in the martini case would not have inclined me to believe that there was a man drinking a martini. Depending on the degree of skepticism involved, these may have inclined me to believe that there was a man over there holding a martini glass, or simply that there was someone over there holding something. And if I had been more credulous, those same experiences and beliefs might have disposed me to a more specific belief, e.g. that there was a man about forty-five years old drinking a martini. So how things epistemically appear to one depends, in part, on one’s degree of skepticism or credulity. Obviously, subjects do not occupy fixed positions on the spectrum from skepticism to credulousness. One may grow more or less skeptical, and one may be skeptical as regards some issues or sources of information while comparatively credulous as regards others. Even relative to a topic or source, one’s readiness to believe can vary according to one’s purposes at the moment. This flexibility is illustrated by Descartes’s First Meditation, which is essentially a set of exercises intended to increase skepticism by instilling an exceptionally cautious doxastic attitude—albeit one that is highly context-specific. Adopting a more cautious doxastic attitude has the effect of restricting the epistemic appearances, by raising the bar as to the strength (and kinds) of evidence regarding p that will dispose one to believe that p.18 Raising the bar is not easy, of course. As Descartes acknowledges, a resolution to believe only what is indubitable will not eliminate the appeal of familiar beliefs. My point is just that adopting a more skeptical attitude, while holding fixed the experiences, beliefs, and other factors that bear on your inclination to believe that p, can 18

As Daniel Stoljar pointed out to me, there are plausibly limits as to how credulous—and perhaps how skeptical—one can be, while remaining rational. The kind of skepticism at issue here seems not to court these limits, for two reasons. First, it does not impose an impossibly high bar for belief; and second, it applies only to beliefs within a narrowly circumscribed realm.

110 Theories of Introspection decrease p’s subjective probability for you. If this subjective probability falls below a certain level, it does not epistemically appear to you that p. By increasing skepticism and thereby decreasing subjective probability, as regards a topic or source, one can restrict the corresponding epistemic appearances. Suppose that you are especially concerned to avoid error in your judgments about your own experiences, though not so intent on this that you refrain from such judgments entirely (or adopt unreachably high standards for belief). This goal will naturally lead you to be scrupulously cautious in your judgments. Even if you are carefully attending to a 48-speckledness experience, you will not be inclined to believe I am now experiencing 48-speckledness. If you are not inclined towards that belief, it does not epistemically appear to you that you are now experiencing 48-speckledness. (Although I am now experiencing 48-speckledness may retain a higher subjective probability for you than any comparable statement, such as I am now experiencing 47-speckledness.) In this way, the attitude of scrupulous caution effectively restricts—or prunes—the epistemic appearances. The strategy for achieving introspective knowledge by acquaintance should now be clear. By adopting a scrupulously cautious doxastic attitude toward one’s own experiences, one seeks to prune the epistemic appearances to the point where those that remain are exclusively determined by how things seem phenomenally—that is, by the phenomenal reality. Success in this endeavor will neutralize the influence of background beliefs; one will thereby avoid the fraternity pledge’s mistake. The introspective judgments based on the remaining epistemic appearances will play the dual role envisioned by the acquaintance approach. They will express how things epistemically seem to one. And since how things epistemically seem is directly and fully determined by the phenomenal reality, they will be directly tied to the experiences that are their truthmakers. When how an experience epistemically appears, to the subject, is determined exclusively by its phenomenal reality, the epistemic appearance amounts to what Horgan and Kriegel (2007) call a “bracketing mode of presentation” of the phenomenal reality.19 [A bracketing] mode of presentation suspends any such presuppositions [about relational features of the experience], so that their truth or falsity does not affect the content of the specific belief that employs such a mode of presentation. This is a mode of presentation that brackets out all relational information about the experience and its phenomenal character, including how experiences of this sort

19

I am grateful to Declan Smithies for bringing this comparison to my attention.

Renewed Acquaintance

111

are classified by other subjects, how they are classified by oneself on other occasions, what their typical causes are, etc. It focuses (so to speak) on how the experience appears to the subject at that moment. (ibid., 128) The process described here, of suspending presuppositions so as to reach a purely phenomenal or “bracketing” mode of presentation, is similar to the strategy I have described as adopting a scrupulously cautious doxastic attitude so as to prune the epistemic appearances. If one carries out this task successfully, the experience’s epistemic appearance will be determined exclusively by its phenomenal appearance. In Horgan and Kriegel’s terms, the remaining information about the experience will be constituted by “how the experience appears to the subject”—where the relevant appearance here is phenomenal. And an experience’s phenomenal appearance just is its phenomenal reality. A judgment expressing this information will be directly tied to its truthmaker and will thereby meet the first of our three requirements for knowledge by acquaintance. But can we successfully carry out this strategy? The best way to settle this question is to attempt it oneself. So I invite you to adopt a scrupulously cautious doxastic attitude as regards your current experiences. You need not go so far as imagining that a Cartesian evil genius is bent on deceiving you. Instead, simply set yourself the task of registering how your phenomenal experience seems to you (epistemically), while doing all you can, short of suspending belief entirely, to avoid error. It will help to make available a relatively sharp, clear-cut sensation, perhaps by pinching yourself. You can then focus your attention on the corresponding aspect of how things seem to you. When I try this, I find it nearly impossible to doubt that my experience has a certain phenomenal quality—the phenomenal quality it epistemically seems to me to have, when I focus my attention on the experience. Since this is so difficult to doubt, my grasp of the phenomenal property seems not to derive from background assumptions that I could suspend: e.g., that the experience is caused by an action of pinching. It seems to derive entirely from the experience itself. If that is correct, my judgment registering the relevant aspect of how things epistemically seem to me (this phenomenal property is instantiated) is directly tied to the phenomenal reality that is its truthmaker. Proponents of the acquaintance approach should concede that experiences do not always epistemically appear as they actually are. This was illustrated by the speckled hen example and the fraternity case. The acquaintance theorist must only maintain that sometimes, experiences are as they epistemically appear—and that, in at least some of these cases, this is because the epistemic appearance is constituted by the phenomenal reality.

112 Theories of Introspection

5. The Justification of Introspective Judgments Here, once again, is the distinctive commitment of the acquaintance approach. Acquaintance Approach: Some introspective knowledge consists in judgments that 1. are directly tied to their truthmakers; 2. depend, for their justification, only on the subject’s conscious states at the time of the judgment; and 3. are more strongly justified than any empirical judgments that do not meet conditions (1) and (2). The last two sections focused on how introspective judgments can meet condition (1). We now turn to considering how such judgments can meet conditions (2) and (3). How might conscious states justify an introspective judgment? And what gives this justification its special strength? Answers to these questions can be gleaned from statements by two leading acquaintance theorists, Richard Fumerton and Laurence BonJour. My suggestion is that one has a noninferentially justified belief that p [e.g., I am in pain] when one has the thought that p and one is acquainted with the fact that p, the thought that p, and the relation of correspondence holding between the thought that p and the fact that p. (Fumerton 1996, 75) On my view, . . . a foundational belief results when one directly sees or apprehends that one’s experience satisfies the description of it offered by the content of the belief. (BonJour 2003, 191) These proposals are similar in spirit. Putting aside some details—e.g., about whether direct awareness of an experience involves acquaintance with a fact or, instead, with an event—the following proposal emerges. One achieves knowledge by acquaintance of an F experience (that is, that phenomenal property F is currently instantiated) iff: i. One is directly aware of an F experience. ii. One is directly aware of the judgment an F experience is now occurring. iii. One is directly aware that the F experience mentioned in (i) makes the judgment mentioned in (ii) true. This is a highly demanding conception of knowledge by acquaintance. It requires, for knowledge by acquaintance of an experience, that one is directly aware not only of the experience itself but also of one’s judgment registering the experience and of the appropriate relation (truthmaking, correspondence, or satisfaction) between these.

Renewed Acquaintance

113

Is this conception of knowledge by acquaintance implausibly demanding, or can we meet conditions (i)–(iii)? One might seek to answer this question by attempting to meet these conditions, as regards a current experience. But in order to verify that one has succeeded here, one must not only meet these conditions but also recognize that one does so. And this is a tall order. Just to meet these conditions, one must achieve a highly complex state of awareness: one must be aware that the experience (of which one is aware) bears the truthmaking relation to the judgment (of which one is also aware). Recognizing that one satisfies these conditions is harder still, for it requires higher-order awareness of this complex state of awareness. Since recognizing that one has met these conditions is considerably more difficult than simply meeting them, an inability to directly verify that one can meet these conditions does not demonstrate that one is incapable of meeting them. In any case, the acquaintance theorist need not place such stringent requirements on knowledge by acquaintance. Fumerton and BonJour effectively require not only that the subject be aware of the basis for her judgment (the experience itself) but also that she be aware that the experience provides a good basis for her judgment. This latter requirement is met by recognizing that the experience stands in the appropriate relation to the judgment (truthmaking, correspondence, or satisfaction). In a discussion of BonJour’s view, Richard Feldman questions the need for this requirement. .  .  . BonJour adds a requirement for foundational justification that involves a meta-level perspective, a (mysteriously non-cognitive) comparing of the contents of beliefs and experiences. My own inclination is to resist the idea that any such perspective is required. (Feldman 2006, 726) If we put aside the requirement that one grasp that one’s experience is a good basis for one’s judgment, what justifies the judgment is one’s awareness of the experience (or, perhaps, simply the experience itself 20). Knowledge by acquaintance does not require awareness that the experience appropriately supports the judgment; in fact, it does not require awareness of the judgment at all. We have, then, two competing conceptions of the justification required for introspective knowledge by acquaintance. On the more demanding conception,

20 In fact, I think it is more promising to take the justification for the judgment to consist in awareness of the experience, rather than the experience itself. By contrast, Smithies (this volume) claims that having an experience suffices for having justification to believe that one is having that experience. But this may not represent a real difference between our views, since Smithies is concerned with propositional justification, whereas my concern is with doxastic justification. Smithies leaves open the possibility that attention to the experience is required for using the justification it provides.

114 Theories of Introspection embraced by Fumerton and BonJour, a judgment cannot constitute such knowledge unless the subject has reason to believe that her judgment is appropriately based—e.g., that it is based on an experience to which it corresponds. The less demanding conception, suggested by Feldman, rejects that requirement. On that conception, the fact that a judgment is appropriately based on an experience (or awareness thereof) can constitute justification adequate for knowledge by acquaintance. I remain neutral between these two conceptions of justification. On either conception, introspective knowledge by acquaintance can clearly satisfy condition (2) of the acquaintance approach. On the more demanding conception, justification consists in the satisfaction of conditions (i)–(iii) above, where one’s satisfying these conditions is purely a matter of one’s conscious states. On the less demanding conception, it is a conscious experience (or conscious awareness thereof) that provides the justification suitable for knowledge by acquaintance. This brings us to condition (3) of the acquaintance approach, which says that some introspective knowledge is more strongly justified than empirical judgments that do not meet conditions (1) and (2). (The acquaintance approach says that some introspective judgments meet all three of these conditions. It does not say that every introspective judgment meeting the first two will also meet the third.) Arguably, one benefit of the more demanding, Fumerton–BonJour conception is that any judgments meeting their stringent requirements will be extremely well justified. But since some doubt that these stringent requirements can actually be met, the acquaintance approach will be strengthened if judgments meeting only the less demanding conception of knowledge by acquaintance will satisfy condition (3). Let us spell out what is involved in satisfying condition (3). A judgment that satisfies (1) and (2) will be more strongly justified than any judgment that does not meet these conditions if: •

•

The judgment’s justification is immune from certain defeaters to which the justification for empirical judgments that do not meet (1) and (2) is vulnerable; and The judgment’s justification is not vulnerable to any defeaters from which the justification for empirical judgments that do not meet (1) and (2) is immune.

I cannot offer a comprehensive theory of justification here (or anywhere else, for that matter). But a contrast between introspective judgments that meet (1) and (2), and perceptual judgments, will go some way toward establishing these points about the comparative immunity from defeaters. Consider the following case. While attending to a current pinching experience (and exercising scrupulous caution, etc.), I judge this phenomenal property is instantiated

Renewed Acquaintance

115

(in me, now). Suppose this judgment is directly tied to the experience that is its truthmaker—that is, it meets condition (1). Moreover, I am consciously aware of this pinching experience, and this awareness justifies my judgment. My judgment thereby meets condition (2), and satisfies the epistemic requirements of the less demanding conception of knowledge by acquaintance just sketched. In that case, the justification for my judgment is immune from defeaters concerning how the experience, or my awareness thereof, is caused. For instance, the judgment remains justified even if my pinching experience is caused by an evil genius, or the actions of a mad scientist caused me to become aware of that experience. And it remains justified even if I have good reason to suspect that these forces are at work. Now consider the case where I have a visual experience as of a table before me, and judge on that basis that there is a table before me. My justification for this belief will be defeated by a well-grounded suspicion that my visual experience has an aberrant cause—an evil genius, mad scientist, or hallucinogen. So perceptual justification is vulnerable to some defeaters from which the justification for the introspective judgment just described (which meets (1) and (2)) is immune. Might the justification for judgments meeting (1) and (2) be vulnerable to some defeaters from which the justification for other judgments is immune? It is hard to see how a judgment’s violating condition (1) or (2) could immunize its justification from defeaters. A direct tie to truthmakers seems epistemically neutral at worst. And the fact that my judgment depends, for its justification, only on conscious states does not seem to make it less secure than judgments that do not meet this condition. Arguably, empirical judgments generally depend, for their justification, on some conscious states of the subject: perceptual judgments depend on perceptual experiences; memory judgments depend on conscious memories; etc. If these judgments do not meet condition (2), this means that they depend on additional justifying factors as well. A dependence on additional justifying factors would seem to introduce potential defeaters rather than neutralize them. What allows judgments constituting knowledge by acquaintance to be especially strongly justified is that the justification for such judgments is directly tied to the truthmaker. By contrast, in other types of empirical knowledge (memory, perception, etc.) justification is linked with the truthmaker through a less direct, merely causal relation. Because of this difference, the latter justification is vulnerable to defeaters (such as aberrant causes) from which the former justification is immune. Obviously, much more could be said about the epistemological implications of the acquaintance approach. But this brief discussion suggests that introspective judgments that meet the requirements of even the less demanding conception of justification—requirements which plainly can be met—can satisfy conditions (2) and (3) of the approach.

116 Theories of Introspection

6. Acquaintance and the Dispositional Analysis of Concepts At the core of the acquaintance approach is the idea that, in knowledge by acquaintance, phenomenal reality intersects with the epistemic (the subject’s grasp of that reality). When one knows a phenomenal experience by acquaintance, one’s judgment regarding that experience is directly tied to phenomenal reality (the experience itself) and directly expresses one’s epistemic perspective. Not surprisingly, some of the most serious objections to the acquaintance approach take issue with the idea that phenomenal reality intersects with the epistemic in this way. We consider objections of this type in this section and the next. The first objection is based on the observation that, if an introspective judgment truly represents an intersection between phenomenal reality and the epistemic, it must express an epistemically substantive grasp of the phenomenal property to which it refers. In such a judgment, of the form this phenomenal property is instantiated (in me, now), the demonstrative term “this property” must be cognitively significant. Wittgenstein challenges the idea that an introspective demonstrative can express an epistemically substantive grasp of phenomenal properties. He contends that demonstratives directly linked to reality will be cognitively insignificant. Imagine someone saying: ‘But I know how tall I am!’ and laying his hand on top of his head to prove it. (Wittgenstein 1953, §279) Suppose that, while his hand is on his head, the subject says “I’m this tall.” So long as he succeeds in demonstrating the distance from the floor to the top of his head, he has said something true. But this statement does not reflect knowledge of his height. He has no substantive conception of the referent of his demonstrative: he does not know what distance he is indicating. In this sense, his demonstrative is cognitively insignificant, or blind. The acquaintance theorist must maintain that introspective demonstrative reference sometimes involves a substantive phenomenal concept.21 For the envisioned

21

The acquaintance theorist should allow that demonstrative reference to the phenomenal qualities of experience is sometimes blind, that is, that it does not always involve the exercise of a substantive conception of the phenomenal property. Here is one such case. Fred lives in a loud neighborhood and has learned to “tune out” the ambient noise when reading. But this does not mean that he ceases to hear the noises: his hearing is fully intact. Fred might demonstratively refer to an auditory experience he is now having, by thinking to himself I will not allow this experience (or the instantiation of this phenomenal quality), whatever it is, to ruin my concentration! Because he is deliberately not attending to the experience in question, he does not know whether it involves the phenomenal quality usually associated with hearing an ambulance siren, or with hearing a car backfiring (etc.).

Renewed Acquaintance

117

intersection between phenomenal reality and the epistemic requires a substantive grasp of the phenomenal property. Moreover, the acquaintance theorist must maintain that a substantive conception of a phenomenal property (of the sort that could help to justify an introspective judgment) does not depend on anything beyond conscious states. These requirements provide the basis for an objection to the acquaintance approach. On that approach, a conception of a phenomenal property takes the form of a conscious state directly tied to an instantiation of that property. But, the objector claims, this conception is insubstantial, and grounds only blind demonstrative reference to the property. The price of tying an introspective judgment directly to phenomenal reality is that the judgment cannot express a substantive epistemic grasp of the phenomenal property at issue. (Hawthorne 2007, among others, makes this objection.) In evaluating this objection, it will be useful to have before us an account of phenomenal concept possession that exemplifies the kind of account that is (allegedly) unavailable to the acquaintance theorist. The most plausible such accounts construe phenomenal concept possession in dispositional terms. Ernest Sosa offers an attractive account along these lines. According to Sosa, possession of a phenomenal concept “is defined in part by sensitivity to the relevant feature of which it is a concept. It is defined in part by the ability to tell when that feature is present or when absent in our experience,” across a suitable range of circumstances (Sosa 2003, 125). To accommodate the fact that concepts can be lost over time, he construes this sensitivity requirement in counterfactual terms. On Sosa’s view, then, my grasp of the phenomenal quality of my current experience partly consists in counterfactuallyspecified dispositions or abilities: to recognize the presence of this quality in other appropriate circumstances and to avoid mistaking a different quality for this one. Sosa further claims that this understanding of phenomenal concept possession helps to explain why introspective judgments about the phenomenal qualities of experience are so reliable. We are maximally reliable [in introspective judgments employing these phenomenal concepts] because our very grasp of those concepts requires reliability in the right circumstances, and the circumstances are nearly always right, leaving little scope for possible failure. (ibid., 126) If Sosa is correct, our dispositions or abilities play two crucial epistemic roles in introspective knowledge of the phenomenal qualities of experience. First, in order for reference to a phenomenal property to be cognitively significant, the referring subject must be able to recognize the property across a suitable range of circumstances, and to avoid mistaking different phenomenal properties for that one.

118 Theories of Introspection Second, judgments involving cognitively significant reference to phenomenal properties, like this property is instantiated (in me, now), qualify as knowledge (at least partly) because anyone who can entertain these judgments will have those recognitional abilities. Sosa’s proposal is appealing. It is hard to imagine someone who genuinely possessed the phenomenal concept pinching, and whose exercise of this concept qualified as knowledge of his current experience, but who was prone to mistaking pinches for tickles and vice versa. However, Sosa’s proposal is incompatible with the acquaintance approach. His claim that certain dispositions are epistemically critical for knowledge of phenomenal qualities is at odds with the acquaintance theorist’s claim that some instances of introspective knowledge (of the phenomenal) consist in judgments that depend, for their justification, only on the subject’s conscious states. If the possession or justified exercise of phenomenal concepts is a matter of dispositions, then dispositions play a crucial epistemic role in justifying judgments about the phenomenal qualities of experience. But dispositions are not reducible to conscious states. So if dispositions play a crucial epistemic role in judgments employing phenomenal concepts, then no introspective knowledge of phenomenal qualities will satisfy condition (2) of the acquaintance approach. The acquaintance theorist must, then, reject the dispositional account of phenomenal concept possession. But in order for her view to remain plausible, she must do justice to the intuitions that make the dispositionalist account appealing. One such intuition is this: the true judgment this (pinching) property is instantiated (in me, now) does not constitute genuine introspective knowledge if the subject is disposed to frequently misclassify pinches as tickles and vice versa. Absent the disposition to correctly apply this concept, the judgment seems merely lucky. The acquaintance theorist can do justice to this intuition: the acquaintance approach is compatible with the claim that any subject who achieves introspective knowledge (of phenomenal qualities, by acquaintance) will have the dispositions Sosa describes. The acquaintance theorist can maintain that these dispositions are explained by one’s meeting the requirements for introspective knowledge by acquaintance. When I adopt a cautious doxastic attitude, and direct my attention exclusively to an experience, how the experience epistemically seems, to me, will be constituted by (an aspect of) its phenomenal reality. The acquaintance theorist can say that it is because I apprehend this phenomenal quality, at the moment I reflect on it, that the counterfactuals are true of me at that moment: my grasp renders me able, at that moment, to recognize this quality and to avoid mistaking a different quality for this one, across a suitable range of circumstances.

Renewed Acquaintance

119

The disagreement between the acquaintance theorist and the dispositionalist has the form of a Euthyphro question. Do I possess the phenomenal concept pinching because I can recognize instances of pinching sensations, or can I recognize instances of pinching sensations because I possess the phenomenal concept pinching? Here are the options spelled out. Dispositionalist: My possession of the phenomenal concept F—that is, my grasp of the phenomenal property F—(partly) consists in my disposition to apply this concept to instances of F, and to avoid applying it to non-F experiences (in relevant circumstances). Acquaintance Theorist: My disposition to apply phenomenal concept F to instances of F, and to avoid applying it to non-F experiences (in relevant circumstances), is explained by my possession of the concept F—that is, my grasp of the phenomenal property F. The dispositionalist option has one clear advantage. It provides a partial analysis of what it is to grasp a phenomenal property, whereas the acquaintance theorist leaves concept possession unanalyzed. However, the acquaintance theorist’s position better reflects the simple idea that I can recognize pinching experiences because I know what it’s like to undergo these experiences. If my grasp of a phenomenal property explains my ability to recognize instances of it, then my grasp does not consist in this ability. So the acquaintance theorist can do justice to the intuition that one who possesses a phenomenal concept (who grasps a phenomenal property) will have certain dispositions, while maintaining that this grasp consists in conscious states. For she can claim that conscious states explain these dispositions. (One way to do this is to say that awareness of a pain grounds the disposition to recognize pains; that is, the conscious state is the categorical basis of this disposition.22) The crucial point is that introspective knowledge of a phenomenal state can be justified entirely by conscious 22

Suppose that I achieve introspective knowledge by acquaintance that I am now experiencing pain. This judgment is justified, according to the acquaintance theorist, by conscious states alone. Among these is my conscious awareness of my experience, achieved through careful attention exclusively to this aspect of phenomenal reality (my pain). The acquaintance theorist can say that this conscious state grounds my disposition to recognize pains; that is, the conscious state is the categorical basis of this disposition. On this account, my current, actual awareness of a pain makes it the case that the relevant counterfactuals are true of me. (Compare: the vase’s molecular structure is what grounds its fragility: its actual structure makes it the case that it would break if dropped.) Introspective awareness of a phenomenal property can also explain the subject’s ability to recognize an instance of that property at a later time, though not in the same way. A conscious state cannot ground a later disposition; and this sort of explanation will involve memory. (I am indebted here to Trenton Merricks.)

120 Theories of Introspection states—in particular, the subject’s direct awareness of an experience. These conscious states may ground or issue in dispositions, but so long as dispositions make no independent contribution to justification, the acquaintance approach is preserved. The objection discussed in this section charged that if introspective judgments directly reflect phenomenal reality, they do not express a substantive epistemic grasp of the phenomenal. I have shown that the acquaintance approach can do justice to some of the key intuitions that inspire this objection. Yet I have not tried to prove that introspective demonstrative reference to phenomenal properties is (even sometimes) nonblind. This would require establishing a basic tenet of the acquaintance approach, namely that introspective awareness of an experience can yield a substantive grasp of its phenomenal character. Introspective reflection does suggest, to me at least, that that is the case. But those who deny this are ordinarily motivated by theoretical reasons rather than by introspective reflection. So the best way to support this facet of the approach is to highlight its resources for responding to these objections. I now turn to the final objection, which charges that direct reference to a phenomenal property can only be blind.

7. Acquaintance and Phenomenal Information Robert Stalnaker explicitly denies that phenomenal reality intersects with the epistemic in the way the acquaintance approach requires. In this passage, he describes an acquaintance-style claim about what happens when Frank Jackson’s character Mary sees red for the first time. [According to the view at issue] when Mary saw the red star, and named her experience “wow,” she knew what she was naming, since she was acquainted with the experience; she had acquired a pure phenomenal concept of it. I want to question the assumption that there is something—phenomenal experience—that has both an autonomous place in a conception of the world as it is in itself and also this kind of distinctive epistemic role. (Stalnaker 2008, 88) If Stalnaker is correct, a judgment can either be directly tied to phenomenal reality (“the world as it is in itself ”) or directly express the subject’s epistemic perspective. But, he contends, no single judgment can play both roles. Stalnaker objects to the claim that phenomenal reality intersects with the epistemic by arguing that, if this claim were correct, information about phenomenal reality would be “incapable of being communicated” (ibid., 77). As he notes, his reasoning here is similar to David Lewis’s (1990) argument concerning the “Hypothesis of Phenomenal Information.”

Renewed Acquaintance

121

Here is how the acquaintance approach implies that phenomenal information is incommunicable. Suppose that in my judgment this is what it’s like to feel a pinch, the phenomenal quality pinching supplies the content expressed by “this.” So “this” marks an intersection between phenomenal reality and my grasp thereof. Now if the acquaintance approach is correct, an instantiation of that phenomenal quality is the only thing that can supply that content.23 This means that no one can think that thought unless pinching is instantiated in his or her experience. The point is not that one must have the appropriate experience in order to demonstratively refer to that phenomenal quality as “this.” Nor is it an epistemic point about other minds, e.g., that one cannot be sure that others have experiences similar to one’s own. Rather, it is that even entertaining this content requires having the appropriate experience. Acquaintance theorists are committed to this point, by the claim that phenomenal judgments constituting knowledge by acquaintance derive (part of) their cognitive value directly from the experiences that are their truthmakers. If an aspect of my phenomenal reality (my current experiences) partly constitutes how things epistemically appear to me, then no one can share or even understand my epistemic outlook unless they are in a position to draw on a similar experience. Since understanding this is what it’s like to feel a pinch requires an appropriate relation to a pinching experience, this information cannot be conveyed in the usual way, by uttering statements that express it. Stalnaker claims that this consequence— the incommunicability of phenomenal information—is unpalatable. This consequence derives, in part, from the fact that the phenomenal information at issue is expressed with “direct” or “pure” phenomenal concepts, which are definitionally tied to their instances. However, phenomenal concepts are not the only concepts definitionally tied to their instances—at least, if influential externalist views about content are correct. Externalists (including Stalnaker) hold that possessing the concept water essentially involves a relation to actual water (H2O): one must be appropriately situated, relative to H2O, in order to entertain thoughts like water quenches thirst.24 This parallels the claim at issue here: that one must be appropriately situated, relative to a pinching experience, in order to entertain thoughts like this is what it’s like to feel a pinch. So on Stalnaker’s own externalist view, phenomenal

23 The acquaintance approach seems committed to this claim. For if phenomenal reality and the epistemic truly intersect in a judgment, then any epistemically equivalent judgment—any judgment with the same cognitive value—will also be directly tied to a type-identical portion of phenomenal reality, that is, to an instance of the same phenomenal property. 24 The “appropriate situation” relative to H2O may be that H2O is in one’s environment or that members of one’s linguistic community have interacted with it (etc.).

122 Theories of Introspection concepts would not be unique in requiring, for their possession, that one is appropriately related to items falling under them. Of course, there remains a difference between the acquaintance theorist’s construal of phenomenal information and the externalist’s construal of water information. The acquaintance approach implies that, in order to possess phenomenal concepts, the subject herself must instantiate the properties they express. By contrast, externalism’s restriction for possessing the concept water is only that one inhabits a watery environment. This difference makes the communication of phenomenal information more difficult. Since you and I live in the same environment, the presence of water in my environment is the presence of water in your environment. So the fact that the concept water is tied to the presence of water does not impair communication between us. But particular instances of phenomenal properties cannot be similarly shared between us. The communication of phenomenal information is more problematic because (to borrow from the title of Stalnaker’s book) we inhabit different “internal worlds.” But even if phenomenal information is incommunicable in Stalnaker’s sense, it can nonetheless be conveyed. If you have never had a pinching experience, conveying phenomenal information about such experiences requires introducing the appropriate experience into your “internal world”—perhaps by pinching you. A brute action of this sort will not qualify as communication in Stalnaker’s sense. However, a similarly brute action would be required to enable Twin Earthians (who inhabit a waterless world) to acquire water information. We would have to bring it about that they stood in an appropriate relation to H2O, perhaps by bringing water to their planet or by transporting them to Earth.25 Stalnaker is right to say that phenomenal information is incommunicable, so long as communicability requires (roughly) that the information at issue can be conveyed simply by uttering a descriptive statement expressing it. But the acquaintance theorist will argue that the incommunicability of phenomenal information is not objectionable. The acquaintance theorist can draw on the point just illustrated. If Stalnaker’s own externalist position about thought contents is correct, this kind of incommunicability is not peculiar to phenomenal information: information about water may be similarly 25 Might we instead simply give the Twin Earthians a chemistry lesson, explaining to them that water is a compound of hydrogen and oxygen? Perhaps. But content externalism’s defining thesis is that thought contents are individuated by environmental factors. Even if the Twin Earthians acquire the concept water through a chemistry lesson, that concept is not purely descriptive. The chemistry lesson would enable them to think water thoughts only by forging a relation (involving a long causal chain) between the Twin Earthians and actual H2O (or its atomic components).

Renewed Acquaintance

123

incommunicable to Twin Earthians. And the latter incommunicability need not be limited to hypothetical interplanetary communication. Peter Ludlow (1995) claims that there is a difference between American and British English, regarding the concept expressed by “chicory,” that parallels the difference between Earthian and Twin Earthian English regarding “water.” This suggests that chicory thoughts may be similarly incommunicable between some members of the actual terrestrial population. Moreover, the differences between phenomenal information and water (or chicory) information may redound to the benefit of the acquaintance approach. According to content externalism, whether one can think a water or chicory thought depends on environmental factors such as the presence of H2O or the linguistic community’s use of “chicory.” This dependence fuels the principal objections to content externalism. Critics charge that, because these environmental factors lie beyond the thinker (are “outside her ken”), individuating thought contents by reference to them will fail to capture the subject’s epistemic perspective. However, the acquaintance theorist’s claim that phenomenal information (the content of phenomenal judgments) is tied to experiences escapes this objection, since experiences seem to be squarely within the subject’s ken. The acquaintance theorist can thus maintain that her view is preferable to content externalism, in at least one important respect. This is not to say that the acquaintance theorist cannot be a content externalist. The point is only this. Both the acquaintance approach and content externalism take the possession of certain concepts to essentially depend on the presence of things falling under those concepts. But because the acquaintance approach takes such essential factors to be mental (e.g., experiences), it is immune from the doubts about epistemic significance that confront externalism. Why does Stalnaker place such a premium on communicability? Like most externalists, Stalnaker gives special weight to the idea that thought contents must be knowable by others in certain ways. Thinkers are things with a capacity to make their actions depend on the way the world is, and with dispositions to make their actions depend on the way they take the world to be. Theorists and attributors of thought characterize these capacities and dispositions by locating the world as the thinker takes it to be in a space of relevant alternative possibilities. The theorist uses actual things and properties to describe these possibilities, and that is why content depends on facts about the actual world. (Stalnaker 2008, 131) If we begin with the assumption that thought contents must be understood by reference to “actual things and properties” accessible to “[t]heorists and attributors of thought,” it is a short step to concluding that my thought contents cannot

124 Theories of Introspection intersect with (or be partly constituted by) unshareable phenomenal reality. But, of course, the acquaintance theorist does not accept that assumption. So she will not be disturbed by the idea that phenomenal information cannot be communicated (in the usual way). The core of the acquaintance approach is that knowledge by acquaintance represents an intersection between reality and the epistemic. On Russell’s view, the intersection between reality and the epistemic anchors all de re reference and is the necessary foundation for all de re empirical knowledge. But one need not take acquaintance to play this foundational role, in order to recognize the appeal of the acquaintance approach. If any elements of empirical reality intersect with the epistemic, the most plausible candidates for this role are elements of mental reality. Because these features of reality occur (or are instantiated) within the individual, the claim that there are such intersections leads to the type of incommunicability that concerns Stalnaker. Acquaintance theorists should not be troubled by this consequence.

8. Conclusion The acquaintance approach to introspective knowledge of the phenomenal does not imply that we are generally reliable in our phenomenal judgments; or that subjects have epistemic access to all of their phenomenal states; or that phenomenal judgments are generally better justified than other empirical judgments. But it is nonetheless a substantial philosophical view, with significant implications not only for debates about introspection but also for other questions in epistemology, for mental semantics, and for accounts of the relation between mind and world. If the acquaintance approach is correct, some introspective judgments possess a level of justification that is unmatched by the justification available for other empirical judgments. This epistemological fact is linked to a semantic feature: introspective judgments that achieve this level of justification derive their (cognitively significant) content directly from their truthmakers. In this way, phenomenal reality intersects with our grasp of that reality. Russell plainly takes acquaintance to involve an intersection between reality and the epistemic. Thus the sense-data which make up the appearance of my table are things with which I have acquaintance, things immediately known to me just as they are. (Russell 1912, 73)

Renewed Acquaintance

125

The acquaintance theorist need not follow Russell in claiming that “things with which I have acquaintance [are] immediately known to me just as they are.” This claim construes our grasp of the objects of acquaintance as exhausting their reality. (Hence Russell’s contention that the presence of a sense datum suffices for knowing it “perfectly and completely.”) Mental reality intersects with the epistemic so long as my grasp of the phenomenal character of an experience—however limited and partial—is directly tied to phenomenal reality in the way elaborated here. The intersection between reality and the epistemic may not be as broad as Russell believed. But a central and distinctive component of Russell’s theory will be preserved if reality intersects with the epistemic to some extent. I have argued that our conceptions of our experiences are sometimes directly grounded in those experiences themselves: that is, in an aspect of their phenomenal reality. In such cases, experiences are as they are immediately known. Phenomenal reality thus intersects with the epistemic.26

26 Declan Smithies and Daniel Stoljar provided enormously helpful comments on a draft of this paper; two anonymous referees also offered useful suggestions. For valuable discussion, I am indebted to participants in a 2011 conference at the University of Texas—Austin, honoring the centenary of “Knowledge by Acquaintance and Knowledge by Description,” and especially to Trenton Merricks, Eric Schwitzgebel, and Lisa Shabel.

126 Theories of Introspection

References Alter, Torin, and Sven Walter. 2007. Phenomenal Concepts and Phenomenal Knowledge. Oxford: Oxford University Press. Armstrong, David M. [1968] 1993. A Materialist Theory of the Mind. London: Routledge. Balog, Katalin. Forthcoming. “Acquaintance and the Mind–Body Problem.” In Identity Theory, edited by S. Gozzano and C. Hill. Cambridge: Cambridge University Press. BonJour, Laurence. 2003. “A Version of Internalist Foundationalism.” In Epistemic Justification: Internalism vs. Externalism, Foundations vs. Virtues, by L. BonJour and E. Sosa. Oxford: Blackwell. Chalmers, David J. 1996. The Conscious Mind. Oxford: Oxford University Press. ———. 2003. “The Content and Epistemology of Phenomenal Belief.” In Consciousness: New Philosophical Perspectives, edited by Q. Smith and A. Jokic. Oxford: Oxford University Press. Chisholm, Roderick M. 1982. The Foundations of Knowing. Minneapolis: University of Minnesota Press. Conee, Earl, and Richard Feldman. 2004. Evidentialism: Essays in Epistemology. Oxford: Oxford University Press. Donnellan, Keith S. 1966. “Reference and Definite Descriptions.” Philosophical Review 75: 281–304. Fales, Evan. 1996. A Defense of the Given. Lanham, Md.: Rowman & Littlefield. Fantl, Jeremy, and Robert J. Howell. 2003. “Sensations, Swatches, and Speckled Hens.” Pacific Philosophical Quarterly 84: 371–383. Feldman, Richard. 2004. “The Justification of Introspective Beliefs.” In Evidentialism: Essays in Epistemology, by Earl Conee and Richard Feldman. Oxford: Oxford University Press. ———. 2006. “BonJour and Sosa on Internalism, Externalism, and Basic Beliefs.” Philosophical Studies 131: 713–728. Fumerton, Richard. 1995. Metaepistemology and Skepticism. Lanham, Md.: Rowman & Littlefield. Gertler, Brie. 2001. “Introspecting Phenomenal States.” Philosophy and Phenomenological Research 63: 305–328. ———. 2011. Self-Knowledge. London: Routledge. Goldman, Alvin. 2006. Simulating Minds. New York: Oxford University Press. Hawthorne, John. 2007. “Dancing Qualia and Direct Reference.” In Phenomenal Concepts and Phenomenal Knowledge, edited by Torin Alter and Sven Walter. Oxford: Oxford University Press. Hill, Christopher S. 1991. Sensations: A Defense of Type Materialism. Cambridge: Cambridge University Press.

Renewed Acquaintance

127

Horgan, Terry, and Uriah Kriegel. 2007. “Phenomenal Epistemology: What Is Consciousness That We May Know It So Well?” Philosophical Issues 17: 123–144. Howell, Robert J. 2006. “Self-Knowledge and Self-Reference.” Philosophy and Phenomenological Research 72: 44–70. Jackson, Frank. 1982. “Epiphenomenal Qualia.” Philosophical Quarterly 32: 127–136. Kripke, Saul. 1972. Naming and Necessity. Cambridge, Mass.: Harvard University Press. Levin, Janet. 2007. “What is a Phenomenal Concept?” In Phenomenal Concepts and Phenomenal Knowledge, edited by Torin Alter and Sven Walter. Oxford: Oxford University Press. Levine, Joseph. 2007. “Phenomenal Concepts and the Materialist Constraint.” In Phenomenal Concepts and Phenomenal Knowledge, edited by Torin Alter and Sven Walter. Oxford: Oxford University Press. Lewis, David. 1990. “What Experience Teaches.” In Mind and Cognition: A Reader, edited by W. Lycan. Oxford: Blackwell. Loar, Brian. [1990] 1997. “Phenomenal States.” In The Nature of Consciousness, edited by N. Block, O. Flanagan, and G. Güzeldere. Cambridge, Mass.: MIT Press. Ludlow, Peter. 1995. “Social Externalism, Self-Knowledge, and the Prevalence of SlowSwitching.” Analysis 55: 45–49. Lycan, William G. 1996. Consciousness and Experience. Cambridge, Mass.: MIT Press. Nichols, Shaun, and Stephen Stich. 2003. Mindreading. Oxford: Oxford University Press. Papineau, David. 2007. “Phenomenal and Perceptual Concepts.” In Phenomenal Concepts and Phenomenal Knowledge, edited by Torin Alter and Sven Walter. Oxford: Oxford University Press. Pitt, David. 2004. “The Phenomenology of Cognition, or,What Is It Like to Think That P?” Philosophy and Phenomenological Research 69: 1–36. Russell, Bertrand. 1912. Problems of Philosophy. New York: Henry Holt & Co. Schwitzgebel, Eric. 2008. “The Unreliability of Naïve Introspection.” Philosophical Review 117: 245–273. Shoemaker, Sydney. 1996. The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press. Sosa, Ernest. 2003. “Beyond Internal Foundations to External Virtues.” In Epistemic Justification: Internalism vs. Externalism, Foundations vs. Virtues, by L. BonJour and E. Sosa. Oxford: Blackwell. Stalnaker, Robert. 2008. Our Knowledge of the Internal World. Oxford: Oxford University Press. Wittgenstein, Ludwig. 1953. Philosophical Investigations, translated by G. E. M. Anscombe. Oxford: Blackwell.

This page intentionally left blank

5 On the Phenomenology of Introspection Charles Siewert

1. Introduction What is the special way in which we each know our own minds? “Introspection,” we say. But this word may seem too theoretically loaded—suggesting a faculty of “inner sense” whereby we turn attention away from the “external world” to perceive what’s inside. And that picture of how we get self-knowledge cannot be taken for granted. However, while the term ‘introspection’ seems practically inevitable, we can elect to use it—as I will—without committing ourselves to the existence of an “inner sense.” I will in fact reject inner sense views. But I also want to do justice to what I think makes them attractive. This means explaining how consciousness and attention to experience make first-person “introspective” knowledge possible. My approach will be phenomenological. Phenomenology, as I see it, aims at a reasoned and unified understanding of philosophically or theoretically salient distinctions, by examining cases in which we judge (or imagine judging) about our own experience with a distinctively first-person warrant or authority. I have argued for its legitimacy elsewhere.1 To rely on such judgments is not to accept self-reports uncritically—as correct portrayals either of experience or of beliefs

1

Siewert 1998, chs. 1–2; 2007a, b; 2011a.

130 Theories of Introspection about it.2 On the contrary, what I propose is that, by means of self-examination, we clarify concepts in terms of which we understand our mental lives, in a way open to criticism and engaged with argument. How this works and why it matters I hope to make plain by what follows. I will first engage in this sort of phenomenology to see how it can help us evaluate the notion of inner sense (or higherorder perception). I will then proceed to criticize two prominent contemporary incarnations of that idea, in William Lycan (2004) and Peter Carruthers (2004).3 Finally, I will explain how my rejection of inner sense leads to a positive account of introspective knowledge.

2. Generic Introspection and the Idea of Inner Sense We should distinguish inner sense from (what I’ll call) generic introspection. It is enough to accept the latter if you agree: you have warrant for first-person judgments about your own experience that is distinct in kind from what others have for speaking about it. In short, you judge of your own experience with a special sort of right. Well, do you accept this? Suppose I ask you what color a shirt looks to you. Or I ask you whether it smells to you like something is burning in here, or whether this milk tastes sour to you. Or I ask you to picture the face of an American politician—and then want to know which you imagined. Or you may ask yourself these or similar questions. In such cases, it sometimes happens that you just do not observe yourself doing

2

On my approach it is wrong to assume that criticism can proceed only by abandoning a “firstperson” perspective for a “third-person” one. For we need not assume that all warrant for your judgments about your experience ultimately derives from some theory justified as an explanation of observations of your behavior. And my way of using first-person judgment to criticize itself need not assume infallibility in stating our own beliefs, since it can also yield a correction of our belief reports. As I explain in Siewert (2007a), in these respects my phenomenology differs from what Daniel Dennett (1991) calls his “heterophenomenology.” 3 On Lycan’s view some of our mental states are “. . . the object[s] of a kind of internal scanning or monitoring by a quasi-perceptual faculty” (2004, 95)—where this “quasi-perceptual” scanning is distinct in kind from representation in higher-order thought. And Carruthers maintains, “It is because it proposes a set of higher-order analog—or ‘experiential’—states, which represent the existence and content of our first-order perceptual states, that the theory deserves the title of ‘higher-order perception’ theory . . .” (2004, 118). Carruthers finds the mark of perception (as distinct from thought) in the possession of “analog” or “nonconceptual” representational content. It is true that Carruthers calls his a “higher-order perception” (as distinct from “inner sense”) theory—while Lycan (like me) uses the terms interchangeably. But the convictions just noted are enough to make both Lycan and Carruthers “inner sense” theorists, as I interpret this.

On the Phenomenology of Introspection

131

what others would ordinarily need to observe, to be warranted in giving the same answers about your experience as you have warrant to give. That this is so might be clear in various ways. Maybe you are lying to them. (So what you offer others to see and hear is at odds with what you know about yourself.) Or maybe you have simply remained silent. (So others just don’t have enough to go on.) Or perhaps you are just not then in a position to observe about yourself what others can— e.g., your eye movements, your facial expressions, your brain scan. Still, often enough, your lack of what others would need in order to render warranted judgment about your experience would not destroy your capacity to do so. Thus, what warrants you in your judgment differs from what would warrant others in theirs about you. If you agree, you agree there is generic introspection. Whatever gives you a distinctively first-person warrant for your judgments about your experience— that counts as introspection in this nonspecific, catchall sense. (Whether there are in fact a number of different “introspective” ways of knowing can initially be left open.)4 What counts as “experience” here? Clearly, from my examples: your imagining something, or things looking or tasting somehow to you. But generally, for me, “experience” is the same as “phenomenal consciousness”—that which there is, essentially and nonderivatively, “something it’s like for one to have” (Siewert 2011b). And I am liberal about this category—I do not restrict it to some “raw,” or even “nonconceptual” sensory aspect of mental life. Wondering whether the milk has gone bad, judging that Nancy Pelosi’s smile is strangely strained, and wishing she wouldn’t smile like that—all these can be experiences. Even if my liberality makes you unhappy, you will likely admit what matters here: often there is nothing beyond at least some of what I call “experience,” the explanation of which you can plausibly cite to justify first-person judgment about it. I judge that this shirt looks (not quite purple but) blue to me, that the milk tastes sour to me, that I am imagining the smiling face of Nancy Pelosi. If I ask myself: “In explanation of what further fact do I offer such judgments?” at some point I come up empty-handed. And just here we may reach for something like inner sense. “To know what I myself am experiencing, I don’t need to get at it by trying to explain something else; I just directly perceive it happening!” The idea may be elaborated by analogy with vision. You are looking into a box whose contents it happens only you can see. Maybe we should similarly say, only you can now sense or perceive (and thus know about) what is in your mind. But while the box may be turned to give me the same sort of view you had of its contents, nothing can be turned to allow me to “look into your mind” as you do. 4

See Schwitzgebel, this volume.

132 Theories of Introspection Some friends of inner sense may hasten to disclaim any such visual analogy. And they may insist (like David Armstrong 1968) that some other mode of sensing (such as proprioception) actually furnishes a better image of the faculty by which one “scans” the “contents” of one’s own mind. (Much as only you proprioceptively feel your own body, only you “innerly” sense or perceive your own states of mind.) Exploring variations on the theme of inner sense could send us on some long detours. I will simply proceed by spelling out the core notion of inner sense I want to examine. That is this. First, whatever the sensory modality, we can distinguish between: (a) your sensing something (e.g., what’s in the box looks somehow to you, or you feel the position of your own limbs); and (b) your thinking about it (e.g., you wonder whether the item in the box is a tooth, or judge that your ankles are crossed). Second, sensing performs an epistemic service—it somehow furnishes you with either knowledge of the item sensed or warrant for thinking what you do about it. Finally—and here is the crucial point—this sensing/thinking distinction can somehow be iterated. It can be drawn not only where something “without the mind” is concerned but also with respect to events in, or states of one’s mind—for instance, one’s own sensing. So there is both first-order sensing (of, e.g., colors, shapes, smells, and sounds) and also—distinct from this in kind—a sensing of such sensing (and perhaps of other mental states/occurrences). Importantly for my discussion: some versions also maintain there is a distinctively sensory form of attention to experience. Sense—whether “outer” or “inner”—brings one knowledge via attention to its proper objects. The key distinctions here (sensing/thinking, first-order/higher-order) need more explanation. However, I hope it is clear enough that we can accept the reality of “generic introspection,” while withholding judgment on inner sense. But we also now may glimpse its allure. A solely self-directed “inward” mode of sensing would seem to informatively specify the mode of access first identified generically and would make this—appropriately—distinct from explanatory inference and exclusively first-personal. Please bear this firmly in mind throughout. I do not assume that the iteration of a univocal sensing/thinking distinction commits the inner sense theorist to an extra layer of the “qualia” supposedly found in “sensation.” (Qualia, I take it, are the seemingly “nonrelational” (or “intrinsic” or “monadic”) components or aspects of sensory experience, commonly thought to be somehow “ineffable,” such as the “raw feel” of a color shade, a trumpet note, or some migraine agony.) Nor do I invoke some other idea of “sensory quality” here. These notions will play no role in my understanding of what sensing is, or in my criticisms of inner sense. And my argument will not depend on the (for me, utterly distinct) assumption that wherever

On the Phenomenology of Introspection

133

there is sensing, there must be some state with a distinctively sensory phenomenal character.5 I will, however, ask whether higher-order sensing (and not just thinking) is—as I will put it—“phenomenologically discernible.” But I do not here adopt the recent usage on which ‘phenomenology’ is but another term for ‘phenomenal character’. Unless otherwise indicated, ‘phenomenology’ here names a style of inquiry—one involving the critical use of first-person reflection to clarify distinctions. Thus, I am asking what phenomenology, so understood, has to say about inner sense.

3. First-Order Thinking and Sensing There will be inner sense only if a sensing/thinking distinction iterates beyond the mind’s ground floor. So to address this issue at all, we must be able to distinguish sensing from thinking. But then we need to understand how this distinction applies at least at the first level. You may want to describe the difference by saying something like: thinking (but not sensing) has “propositional” (or perhaps “conceptual”) “content.” But I do not think theoretical claims like these or professional jargon like “sensory qualities,” “qualia,” “representation,” or “content” provide us a good place to start. Let’s first cultivate some concrete understanding of what we want to distinguish. Otherwise, I fear we won’t adequately understand what we are talking about. 5 To clarify: I distinguish my case from Rosenthal’s (2002) “no extra quality” argument against inner sense, not because I would grant Rosenthal his conception of sensory qualities at the firstorder level, and just don’t see why all sensing (even higher-order) should need them. Rather, my point is (in part): I don’t recognize the qualities that figure in his story. They are to be that by which we ordinarily introspectively distinguish our sensory states—but at the same time they are said to belong to pains that don’t feel anyhow, and to characterize blindsighters’ vision. However, I do not ordinarily introspectively think of the differences among my own states with reference to such qualities. Rather, I do so precisely in terms of how certain states feel (as they would not, to the anesthetized), and with respect to how colors and figures look to me (as they would not, to a blindsighter). I would also make no demand that inner sense display its own “qualia,” partly because I cannot endorse all that term often drags in with it. (I do, however, endorse notions of phenomenal character, and of phenomenal features (as described in Siewert 1998, chs. 3–4, and in Siewert 2011b).) Now admittedly, in previous argument for the indiscernibility of inner sense I did appeal to the absence of inner sensory phenomenal character (Siewert 1998, 212–213). But I now see that as unfortunate, both because the confusion of sensory qualities with phenomenal character facilitates conflation of my complaint with Rosenthal’s, and because I thereby obscured the fact that my crucial point really had to do—not with an absence of phenomenal character—but (as I try to explain better here) with the phenomenological absence of perceptual constancy, and of objectual sensing, beyond the first-order level.

134 Theories of Introspection I begin with this elementary point: some things are sensed when others are only thought of. For instance, staring at the mess on my desk, I am absorbed in thought about someone I plan to meet this afternoon. But I certainly do not now sense her here, and in this way “sense her presence.” However, we cannot simply understand what is special about sensing by speaking of the “presence” of what or whom is sensed. What is not sensed may still be thought present. Suppose I am now looking at the very person I was waiting for. I may contrast her looking to me as she does with my continuing merely to think about her, and confidently think of her as still present, when I close my eyes. The former state (eyes open) will constitute a kind of “first-order” sensing, whereas the latter (eyes closed) will not be sensing—but only thinking about her presence. The sensing and thinking here are “first” (not “higher-”) order, because here what looks to me somehow (and what I am thinking of) is not my own sensing (or some other mental/cognitive state). I trust you can make similar observations. And we could make analogous points about sensory modalities other than vision. The shape or texture of what I feel, I still judge to be there unfelt when I withdraw my touch. The scent or odor in the air no longer smells any way to me, when I pinch my nose, though I think it is still there. And when I am no longer tasting the ice cream, I can certainly think it is still coconut. As for hearing—I cannot so effectively shut my ears against the leaf blower racket outside. But if I plug my ears and shut the window, it may no longer sound any way to me—though still (I think) it blares on there undiminished. In all these ways we may make ourselves stop sensing what we are still thinking of, and thereby exhibit to ourselves the difference between sensing and thinking. Can we do just the same with the feeling that belongs to proprioception? To make this sort of feeling “go away,” while the thought of what is felt remains, I can shut no bodily organ, nor withdraw from this body, the position and movement of which I feel. (How can I move my entire body away from itself ?) Nevertheless, it seems bodily feeling can diminish when my attention is directed or drawn away from my body or some part of it: away from my body as a whole when, for example, I am absorbed in something I am reading; away from one part to another when, e.g., I follow the instructions: “Attend to your left foot; . . . attend now to your legs; . . . now attend to your breathing; . . . now to your face.” Such shifts in attention permit me to recognize that a prior feeling of my body can “recede” and be “swallowed up” in the background of my experience, even if I happen to be still thinking of my body, or the part of it formerly distinctly felt. In the reading case: what I am thinking about with such concentration may be my medical report, describing this very body or part—my own—my feeling of which has faded into the background. Also: I may think of my feet before I think of my face, when, for instance, I am answering

On the Phenomenology of Introspection

135

questions about my “distinguishing marks”: “I have no scars on my feet; . . . I have two on my face.” But this may well not coincide with a change in how my feet and face feel to me (of the sort that may be invoked by my attention instructions). Now maybe it is unusual that I think of my body, when I have no feeling of it whatsoever—not even of a marginal sort. Still the cases just mentioned reveal independent variation in how one feels and how one thinks of one’s own body, and that is enough to highlight the contrast between them. Anyway, cases of local anesthesia and severe nerve damage show you can think about a part of your body while no longer feeling it. And for a “whole body” case, consider: I might intelligibly (and correctly) think to myself, as I float in a sensory deprivation tank: I can no longer feel my body. If you agree, then you and I both recognize—for each commonly accepted modality by which we sense something other than our own sensing—a contrast between merely thinking something is present and sensing its presence, which demonstrates our grasp of the sensing/thinking distinction. A more exotic demonstration may be found in “blindsight.” As is well known: subjects suffering from damage to the visual cortex deny that they see anything where, prior to the damage, they would have claimed they did. Nevertheless, forced choice tests reveal some can still discriminate (verbally and otherwise) shapes, orientations, and movements caused by optical stimuli in that region of purported deficit. One way to interpret this: using the notion of ‘look’ illustrated above, suppose the stimuli the subjects deny seeing do not look any way to them (any more than what you look at looks anyhow to you after you shut your eyes). There is a form of sensing—the kind reportable by something of the form ‘That looks . . . to me’—which we may intelligibly suppose such blindsighters lack. Nonetheless we can say their “guesses,” their responses to verbal “forced choices” (“Was that an X or an O?” and so on) express thoughts—thinkings as distinct from sensings as were your “shut eye” thoughts about what you had been looking at. Analogous reflections seem available with other standard modalities of first-order sense. It is at least conceivable that you may discriminate in thought the presence of a sound that does not then sound any way to you (“there was just a beep”), or an odor that does not then smell any way to you (“this one’s coffee, that one’s lemon”). We may even imagine something like this happening in proprioception: eyes closed, I cannot feel my arm being raised and lowered by the experimenter, but when required to say whether I think it’s up or down, as it happens, I answer correctly. The contrast at issue reveals itself in another noteworthy way—not, as just now, because thought persists as sensing withdraws—but because sensing persists in the face of thought’s opposition. Even as I affirm that the lines of the drawing in my psychology text are straight (because a moment ago, they looked straight against a

136 Theories of Introspection ruler), now that the ruler is removed, they again look curved to me. And two green patches look different to me against different backgrounds, though I think they are the same shade (due to their appearance when their background is the same). Disturbingly, it still sounds to me as if someone is murmuring my name, but I think there is really no speaking I hear (for I see no one there, and the voice sounds the same even as I try to approach or flee its source). That illusion and hallucination may thus persist against our better judgment again displays the difference between sensing something and thinking it is there. More should be said about this distinction. But by applying this contrast as indicated, we exhibit the basic grasp we need if we are to think clearly about these matters. I will later return to such concrete demonstrations, to consider whether they also are available, where what is thought of and ostensibly sensed is one’s own sensing, and so whether sensing can be “iterated”—as a “sensing of sensing.”

4. Do Sensing and Sensed Sometimes Coincide? It may seem this iteration can be easily confirmed—and with it, the reality of inner sense. Take ordinary nausea. This is a feeling. And it is something you feel. So it is a feeling that you feel. And to “feel” here is—not to think—but to sense. So does it not follow that you sense your own sensing? But then wouldn’t this be a more-thanfirst-order sensing—the very inner sensing we seek? Let’s consider this more closely. What I just said about nausea seems to hold for “bodily sensations” generally— e.g., feeling an itch, feeling dizzy, and feeling pain, and (not to be so negative) feeling relaxed, feeling warm, and feeling sexual pleasure. In each case we may say some feeling is felt (thus “sensed”). We may take this to say that here something felt is identical with the feeling of it. Sensed and sensing “coincide.” I do not mean to say that all that is felt in these cases is the feeling itself. Someone may, for example, hold that when you feel pain, we can also take the pain felt to be something other than a feeling felt—some “object of awareness” felt to be “in” a body part. Though puzzled by such talk, my aim here is not to reject it. Nor do I wish to deny (something I actually believe) that, in feeling pain, I also (“proprioceptively”) feel something that is not a feeling or a pain (in any sense): the movement of my foot, for instance—which feels painful when it hurts to move my foot. Right now I just want to point out that we can truthfully speak of “feeling our feelings” in the way just indicated, and so (by taking feeling to be a kind of sensing) call this “sensing of sensing.” This, what I’ll call “coincident” sensing is reported by means of an “internal accusative.” That is, its report is like those of dancing a last dance, or laughing a hearty laugh, or living a long life)—the dance is not distinct

On the Phenomenology of Introspection

137

from the dancing of it, the laugh laughed is none other than the laughing, and the life lived is the living of it. Just so: the feeling (sensation) felt is none other than the feeling (sensing) of it. Coincident sensing contrasts markedly with that earlier exemplified by the visual appearance of what’s in my box. When the tooth it contains appears to me, when it looks somehow shaped (e.g., like an incisor, like a molar) and colored (yellowish, gold, or pearly white), what I sense is what looks/appears to me somehow: the tooth. But then what is sensed is distinguished from the sensing of it. The tooth with its color and shape is distinct from its looking as it does to me. So here, in this case of what I will call “objectual sensing,” it seems sensing and sensed do not coincide. (More on this later.) I have introduced coincident sensing by reference to feeling bodily sensations. But could we maybe extend this notion to visual and other sensory experience generally, even while keeping the contrast with “objectual” sensing? Yes. When the tooth (or the flower, or the cloud, or the contents of my desk drawer) look somehow to me, I suppose I could also say I “sense” them to look somehow to me, or (better, and more idiomatically) that I experience their visual appearance. But in this case—I may equally suppose—the “sensing of the looking” (i.e., my experience of the appearance) is none other than the looking sensed (the appearance experienced). This goes to show ‘experience’ can display the same ambiguity that we allowed ‘feel’ (and consequently, if awkwardly) ‘sense’. I may be said to experience the appearance when that appearance (the appearing) is an experience—and what is experienced is none other than the experiencing of it. By contrast, sometimes what I experience is not an experience: the tooth that looks somehow to me. I “experience” this in a different sense—objectual, not coincident.

5. Inner Sense: What We’re Seeking Why dwell on these distinctions? We need them to see better just what is this inner sense we seek on the way to understanding introspection. Inner sense would be a higher-order sensing of sensing. To confirm its reality we would then need to find sensing being sensed in a manner that takes us beyond the first-order. And that means: to a noncoincident sensing of sensing. For any sensing of sensing that is simply coincident with—that is not distinguished from—the sensing that is sensed is clearly of no higher order than it. Thus, the idea recently aired—that bodily sensations exemplify inner sense—was confused. Neither feeling your feelings nor experiencing sensory appearances will count as inner sense—as long as no distinction opens up between the feeling felt and

138 Theories of Introspection the feeling of it (the appearance experienced and the experience of it). While no difference has clearly emerged for us between sensing and sensed, we have as yet no license to speak of sensing of a higher-order sort. And that is what we mean here by inner sense. This talk of “orders” needs comment. When I ask whether we can find a noncoincident “sensing of sensing,” this is not the same as asking whether there are two numerically distinct states or events of sensing: one of color, say, and a second, which is “of ” the first. For, to have what is sought, we just need two distinguishable features—e.g., the property of being a sensing of a color and the property of being a sensing of that color-sensing. If someone says that it is numerically one and the same state of mind that “doubles back and senses itself,” that still counts as secondorder sensing. The basic question of inner sense then is whether there is some way to clearly distinguish and attribute two features—the sensing and that sensing’s being sensed. I do realize that, on a certain usage, there is “higher-order” sensing, only where there is a higher-order state numerically distinct from a lower-order state. Thus, on that view a “self-representing” state would not exhibit higher-order-ness. But for my purposes, if you say a single state sensorially represents both: a specific color and itself representing that color, you would be saying there is a more than first-order, noncoincident sensing—an “inner sensing.”6 But isn’t all this funny talk of coincident sensing absurdly redundant? (“I feel a feeling,” “I experience an experience,” “I sense my sensing.”) No, not really. For such locutions would permit us to speak—not just of the sheer fact of feeling/ experiencing/sensing—but also of differences in how the feeling feels, or how the experience is experienced, or (as we may now say, albeit with some strain) how the sensing is sensed. We want to recognize, for instance: how pain is experienced by someone differs from how nausea or an itch is experienced by him or her. The seemingly redundant forms of expression give us a means for saying this. That is to say, they give us devices for speaking of the “phenomenal character” of experience, of what it’s like for you to have it. For it seems right to identify this with how you experience an experience (e.g., how your feeling feels to you). Note well: even so, it is just not clear that with any of this we have gone a hair beyond first-order sensing or experiencing. For when you speak only of how the

6 Kriegel (2009) understands “higher-order” thought and perception as implying a distinctness of states. Carruthers, however, distinguishes higher-order perception from inner sense partly in virtue of the fact that the former does not assume such distinctness: “it is one and the same set of states that have both first-order and higher-order analog/nonconceptual contents” (2004, 118). My use of ‘higher-order’ is thus more like Carruthers’.

On the Phenomenology of Introspection

139

experience is experienced, it is not clear you have marked any difference between a feature of the experience and the manner in which you experience it. And only then would you have the distinction between manner of sensing and features sensed that is required for higher-order sensing. To sum up: we now recognize a form of sensing (as distinct from thinking) that is not a sensing of our own sensing or mental state—visual appearance, for example. And we can recognize a sensing that is of our own sensing—feeling a feeling (an itch, for example). But so far we have no license to take this latter in something other than an internal accusative sense. And such “coincident” sensing would not rise above the first order, either as state, or as property. Your sensing is first-order just when what you sense (if anything) is not your own sensing (or any other mental state)—unless this is merely coincident sensing. By contrast, your sensing would be higher-order just when what you sense is your own sensing (or other mental/cognitive state), and the manner of sensing does not simply coincide with, but is distinguishable from, what is sensed. There is a way to allow that you have coincident sensing (that is, experience) of your own first-order sensings across the board (in vision, hearing, tasting, and so on). But this would not be to admit inner sense.

6. Objectual Sensing We are now clearer about how inner sense calls for a sensing/sensed distinction above the mind’s ground floor. But we still need to understand better what, if anything, allows us to draw this distinction even at the first-order level. So far I have taken for granted a contrast between “objectual” cases (like the tooth visually apparent) and “coincident” cases (like nausea felt). But now—just how is it that I can identify cases of the first sort, in which I distinguish between (e.g.) that which looks somehow to me, and the manner of its appearance—i.e., how it looks to me? We need to take into account how closely this is bound to the direction of attention. In vision attention is “directed” (“overtly”), when I look at something, whether I direct my attention to it, or my attention is “grabbed” or “captured” by something—“drawn” to what I look at. As I look at different things the way things look to me—how they appear—changes. And yet, often enough, what I am looking at does not appear to change (in color, shape, size, location). More generally, I may draw a distinction between two kinds of change in how I sense. 1. I sense something to change. 2. How I sense something changes, but I do not sense it to change. (Or how I sense some feature of it changes, but I do not sense it to change in that respect.)

140 Theories of Introspection When how I sense coincides with what I sense, type (2) change is not phenomenologically discernible. For instance, if what I feel is nausea, when I judge introspectively that there is change in how that feeling feels to me, I cannot also say that what I feel does not change at all. Nor can I identify a feature of what is felt (a feature of the nauseated feeling), with respect to which it feels just the same, or is sensed as just the same—even through a change in the way it feels to me. Although the feeling may constantly feel “like nausea” (that type of feeling) even as it varies, say, in felt intensity, the specific type of nausea-feeling is not felt to be completely the same throughout. Although the feature felt or sensed remains “nausea,” still different variants of it are felt—there is a sensed change (if only slight) in what type of nausea is felt, if indeed the way it feels changes. With each kind of bodily sensation I would say the same: the feeling felt—the pain, the itch, the ecstasy—feels different (I sense it to change) as the way it feels varies. By contrast, in the case of what I have called objectual sensing, type (2) change is phenomenologically discernible. I can identify instances where how something looks to me (or how some aspect of it looks to me) changes, though it does not thereby appear to change—it does not look as if it is changing (in that respect). How something’s shape looks to me changes as I simply look at different parts of it, but it does not appear to change shape at all. Running my hand along a curved surface, I touch it, and so may attend to its shape or texture: how it feels varies, though no change in its shape is felt. Moreover, I can (in the examples just offered) discern type (2) cases that also constitute the positive sensing of constancy. The object I am looking at appears to me to stay the same shape or color during a change in its manner of appearance: it looks the same in shape and color throughout. What I touch feels the same shape as I run my hand over its surface. Note that illustration of type (2) is not confined to cases where the object is sensed to change in no respect whatsoever. So: a cup is tilted toward me, so its rim appears constant in shape, even as the way that shape looks to me somehow changes. It does not affect my point if we say here that this change in appearing is to be understood entirely in terms of the rim’s appearing to change with respect to some feature other than its shape—say, its position: it appears tilted toward me just so. For still: this is not a sensed change in what type of shape appears to me. Wherever this “sensing of constancy” condition is met, sensing is “objectual.” So, when you can determine introspectively that what you sense is sensed to remain just the same in some respect (such as color or shape), even as how you sense this alters (how its color or shape looks to you changes), you are entitled to regard this as a case of “objectual”—and consequently, as “noncoincident”— sensing. Here what is sensed does not simply coincide with how it is sensed, precisely because what is sensed is sensed as totally constant, apparently unchanging

On the Phenomenology of Introspection

141

in some respect, even while how it is sensed varies. And, as I use the term, a mode of sensing is “objectual,” only if it admits of sensory constancy—only if, in this mode one senses features (or their instances) to remain the same even as how they are sensed changes. We may wonder just what modes of objectual sensing are phenomenologically discernible. What about hearing? I listen to something (say, the music on the radio), and in doing so attend to this. And then, as my attention slips away, I start to listen to something else (or just stop listening), and the way the music sounds to me alters, as it “shifts into the background.” But the sound does not itself thereby appear to change (in volume or proximity, for instance): this difference in how the music sounds to me does not constitute its sounding to me as if it is changing. More: it sounds the same loudness and distance throughout. What about proprioception? Sitting here typing, I start to notice my hunched shoulders so that how they feel to me changes. But I do not thereby feel my posture to change; I feel how my shoulders are hunched—I did not feel myself hunching them. And when I began to notice how my shoulders are hunched, did their position not somehow feel the same to me as it already was? Though only now prominent in my feeling, my hunched shoulders fit into and do not contrast with the overall posture I had already at least vaguely and marginally felt my body to occupy. If this is right, there is sensory constancy here too—my overall posture feels constant, though the way it feels changes when I notice how I am holding part of my body.7 What about when I smell a scent or odor? In attending to it, the way it smells to me somehow changes—but I do not then smell something changing its scent. May I not say it continues to smell the very same scent to me, even as how it smells to me intensifies or diminishes with attention? Finally, taste. When I taste some chocolate, how it tastes to me may also alter somehow—as I attend to it more or as it melts in my mouth. But that is not to say I taste it becoming more or less chocolate-fl avored. Does it not taste the very same fl avor throughout? We may still see significant differences within objectual sensing. Do you want to say that, much as a visible object may appear to change shape or color (“before your eyes”), something tastes to you to be changing fl avor (“upon your tongue”)? Probably not. It seems vision exhibits a more advanced form of “objectuality” than taste—what is visually sensed as constant in some respect can at the same time be visually sensed to change in some respect, and then, to change in the very respect in which it first appeared constant. But, however exactly we want to 7 We can still recognize that even if, in my minimal sense, we “objectually sense” our own bodies, there are important ways in which we do not experience them merely “as objects.”

142 Theories of Introspection interpret, extend, and subdivide objectual sensing, we should distinguish a further form of noncoincident sensing. We might call this “registration.” To illustrate: consider another way to interpret the nausea case—different from (but consistent with) that earlier suggested. When we feel nausea, we sense, not just this feeling (nor just this and the—vaguely bounded—region of the body nauseously felt), but also a separate bodily condition that such feeling “registers”: infl ammation of the stomach, for instance, or (more momentously) pregnancy. Similarly, one might say that when one feels pain one does not just sense this (painful) feeling (nor this plus the painfully felt part of one’s body), one senses (registers) tissue damage of some sort in that area. This is not coincident sensing of what is registered, because the condition registered does not coincide with the sensing of it. But neither is it objectual sensing. For when I feel nausea or pain, it doesn’t seem right to say that, as how it feels to me changes (perhaps with gain or loss of attention), the registered state is nonetheless felt to remain just the same. To be sure, I may, for some reason, rightly think the registered condition remains invariant even as the feelingintensity waxes and wanes. But that would not mean that, as the feeling of nausea intensifies, I am feeling my stomach to remain equally infl amed, or that as the feeling of pain diminishes, I am feeling the tissue to remain equally damaged). And while you may be “no less pregnant” as your morning sickness subsides, this would not be because you are then feeling the “degree of pregnancy” to remain constant—but simply because (and only in the sense that) you can’t be just a little bit pregnant. Now some may doubt that sensory constancy is discernible in all of the modalities. And classical empiricist accounts may deny the phenomenological reality of objectual sensing—of sensory object constancy—altogether. On such views, considered subjectively, the “how” and the “what” of sensing always simply coincide, except perhaps where what is sensed is merely “registered.” I do not propose now to go further into the phenomenology of object perception. For present purposes, it is enough to say this. If we can phenomenologically discern, at the first-order level, sensing other than the coincident kind, this will be either objectual sensing or sensory registration. And for objectual sensing, we need something more than a lack of sensed change in what we variantly sense: we need positively to sense it then staying the same in some respect.8 8 Readers may find that my story about objectual sensing and registration parallels to some extent Tyler Burge’s recent (2010) theoretically rich account of what he calls objective representation and registration. My understanding of the notions I explain here was arrived at independently, and I will not try here to examine the relationship between my view and Burge’s. However, it may be helpful to note that my account (by contrast with his) is phenomenologically based and concerned specifically with conscious experience.

On the Phenomenology of Introspection

143

7. Is Inner Sense Phenomenologically Discernible? I do not doubt that higher-order thought occurs and is phenomenologically discernible. We can talk about (hence express thoughts concerning) our first-order sensing and thinking. But is there also, distinguishable from such thought, a higher-order sensing of what it is about? At the ground floor, what is phenomenologically distinguishable from thinking is either coincident sensing, objectual sensing, or sensory registration. If we look for an analog at the higher-order level, we do not want coincident sensing—that would be no ascension in “orders.” So we need to ask whether either objectual sensing or sensory registration is found beyond the first-order. First, consider whether we may discern an objectual sensing of sensing. For this what we would need to find is a change in the way some feature of, say, a visual appearance is sensed, even while that visual appearance is sensed as itself quite unchanged with respect to that feature. We would need something like apparent shape constancy at the second-order level. For example: First-order case. Consider the change in how a stable figure looks to you as you shift attention while it appears unchanged. This allows you to distinguish what is (visually) sensed from how it is (objectually) sensed. Second-order case. Now repeat the same sort of operation on the visual appearance. Without changing the way the figure looks to you, shift your attention with respect to this (constant, stable) appearance, so as to alter how that appearance is sensed by you. In the first-order case the operation seems, again, easy enough. But in the secondorder case I am drawn up short. How do I even try to follow these instructions? How could I vary the way the visual appearance is experienced or sensed by me, without changing the figure’s visual appearance to me? If I cannot do this, then I just cannot find, in first-person reflection, an objectual, higher-order sensing of sensing. We may, as earlier suggested, extend the notion of “sensed sensing” beyond “felt feelings” of bodily sensation, and speak, e.g., of sensing how something looks— that is, of “experiencing its visual appearance.” But the same points then apply to each. Just as the feeling felt does not remain, for first-person reflection, discernibly and completely constant in respect of some feature, while the way it feels changes, so the visual appearance experienced does not stay completely fixed for me in some respect through variation in how it is experienced.9

9 My point here bears some resemblance to Shoemaker’s (1996, 219), when he says that there is no difference between (a) shifting attention from one hand to another, and (b) shifting attention

144 Theories of Introspection But what about the other noncoincident sort of sensing recognized: registration? This seems no more discernible in second-order form than objectual sensing. Consider: in the nausea case, I would need to find a sensing of that feeling, distinct in kind from the feeling itself or any thought about it—a separate sort of sensing that registers this feeling, as that feeling registers the condition of my stomach. But I find nothing meeting that description—no such additional sensing, distinguishable from a coincident feeling, is discoverable to first-person reflection. Where the feeling of nausea is concerned, such sensing as I find is none other than the very feeling to be registered. And similarly for something’s looking somehow colored, or feeling somehow shaped. But for inner sense, I need to find a sensing that is other than that which is to be registered. This reveals the ultimate futility of trying to demonstrate concretely to ourselves a second-order sensing, as distinct from thought. For consider, in the light of what has just been said, how we might attempt to do this, appealing to the ways of distinguishing sensing and thinking at the first-order level earlier articulated. That some of these are nontransferable to the second level should have been obvious from the start. I may shut my eyes to what I see, or withdraw my hand from what I feel—though here what is “out of sight” (or feeling) is not thus “out of mind.” But I cannot similarly contrast inner sensing with reflective thought by shutting some organ of (inner) sense to (or withdrawing it from) the sensing I still hold in my thought. But subtler failures of the analogy now become evident, once we appreciate the absence of sensory object constancy beyond the first-order. I cannot hope to demonstrate an inner sensing/reflective thought distinction by analogy with the (motionless) withdrawal or direction of attention found in proprioception. The disposition of my limbs feels unchanged, even as how this is felt alters with attention—and this difference in feeling may be no difference in thought. But I cannot then ascend a level, and similarly keep my own sensing constant, even while the manner in which it is sensed shifts with attention. For no shift in attention that alters the manner in which the sensing is experienced by me can be distinguished from a change in the sensing itself. Thus, pace Armstrong, making proprioception (rather than vision) the favored analogy for “inner sense” does not

between your visual experience of each—unless it is a difference in your thinking. But I am trying here to further support and refine this sort of point. I claim to support it, by clarifying (with my discussion of sensing vs. thinking, and objectual vs. coincident sensing) what we would need to find, to find something here besides thinking. Because we do not find that beyond the first level, we find only differences in thinking. I aim to refine the point, by making clear (in section 8) that these are not just any old differences in thinking, but rather special ones—which help explain why it might have seemed there was sensing here, as well as thinking.

On the Phenomenology of Introspection

145

really help. Relatedly, pace Carruthers (2004, 118–119), the problem with inner sense is not removed by saying (along with Armstrong) that sensing (or perceiving) does not always need a special sense organ. For the root difficulty remains: how to demonstrate to ourselves independent variation in sensing and sensed. Furthermore, I cannot find some analog of blindsight for inner sense that would enable me to separate second-order sensing and thinking. I may conceive of the stimulus not looking any way to me (as ordinarily it would), even as I am prompted by its effect on my eye to think it is there. But now, how am I to iterate this operation and conceive of an “inner blindsight”? I would need to conceive of something’s still looking somehow to me (in the sense thought to be missing in ordinary first-order blindsight), and of my still thinking this visual appearance remains, even while I no longer sense or experience its looking this way to me. I could perform this conceptual feat, only if I already had some way to pry apart, not just my thought about how it looks to me from its looking to me as it does, but its looking this way to me and my experience of its looking this way to me as it does. I would need some way to make it clear to myself that my experience of something’s visual appearance does not simply coincide with the manner of visual appearance itself. And just that is what we do not find, when we find neither sensory constancy nor registration at the second level. Notice now also how this discredits the notion that a postulated inner sense could play an epistemic role like that of attention in first-order sensory appearance. I have said there is no (phenomenologically discernible) way to experience one’s own sensing as remaining quite unchanged in some respect as one’s manner of sensing it alters. This failure to find sensory object constancy at the second level is also a failure to find any additional, distinctively sensory form of attention in operation there. If there is none to be found, it follows that one cannot attend more closely to one’s own sensing by sensing it better, by “getting a better look at it.” But then there is no senselike way to confirm and correct our judgment about our own sensing, through making it better apparent by directing attention upon it. Thus, nothing like first-order sensory attention is to be found at a higher level. So we can rightly assign no epistemic role to an inner sensing, parallel to the role of first-order appearance. This in turn helps explain why it is also no use trying to exhibit to myself a second-order thinking/sensing distinction by appeal to some analog of first-order sensory illusion or hallucination. For how will I recognize a case of “inner illusion”? Can I find myself confirming a judgment that how I actually feel, or how things really look to me, conflicts with how I (falsely) experience my feeling or visual appearance? I could, if I found myself sensing (and not just thinking of) the feeling or appearance as staying quite constant in some respect, even while my manner of sensing it changed for the better, so that my thought could favor the improved

146 Theories of Introspection “appearance” of my sensing and oppose the defective one—now exposed as illusory. But this is just the sort of constancy found missing. Similarly for “hallucinating an experience”: it seems I might recognize a case of this, if I could identify or reidentify as constant some sort of “setting,” “field,” or “space” of feeling or appearance (distinct from that in which my body is situated), wherein I sensed/experienced some feelings/appearances incompatible with others arising in the same (mental) “place.” But again, the unavailability to reflection of second-order sensory constancy seems to preclude this. We are left with nothing then, but the suggestion that illusions or hallucinations in second-order sensing would be some sort of “misfirings” of mere sensory registration. But this will be of no avail, if still we find only sensings of sensings that simply coincide with what is sensed.10 My argument thus far can now be briefly stated. Inner sense will be phenomenologically discernible to us, only if we can find, in first-person reflection, not just thought about our experience, but either a form of objectual sensing, or a form of sensory registration of our own sensing. Otherwise we will not distinguish what is sensed from how it is sensed, and the only “sensing of sensing” we will discern is coincident sensing (experiencing) of first-order sensing—which is not, as inner 10 With this in mind, we can see why attempts (as in Lycan) to offer illustrations of “illusions of inner sense” are so problematic. He asks us to consider a fraternity prank, where the victim anticipates having something burning hot applied to his body, only to have this covertly switched at the last moment for something icy cold. A brief experience ensues, Lycan tells us, which comprises an illusion of inner sense. “[T]he fraternity pledge mistakes the cold sensation produced on his bare skin by the ice cube for burning heat” (2004, 96). But is this clearly an illusion of inner sense? We might well suppose the intense, fearful expectation of being burned occasions a different sort of mistake. For instance: (1) there was in fact something cold, which for a mere moment, the victim somehow fleetingly (and mistakenly) felt to be hot (or at least: he felt it in some startling way, but not yet as clearly either hot or cold). Or: (2) he mistakenly, and fleetingly thought it felt hot (or at least: not-cold) to him. But these would yield at most (in case (1)) a first-order sensory illusion, or (in case (2)) a mistaken second-order thought. But Lycan needs an illusion of higher-order sense: (3) it definitely felt cold to our miserable pledge as soon as it felt anyway at all, but for the merest moment, this feeling of cold was also just as definitely (not just falsely thought to be but) mistakenly sensed or experienced by him as a feeling of burning heat. But how is one to determine (3) happened, rather than just (1) or (2)? The problem is we do not here have, phenomenologically, a feeling of cold sensed as invariant through a change in how it is sensed/experienced, so that the initial tiny time slice of (first-order) sensing cold (initially wrongly sensed as a feeling of burning heat) can be subsequently sensed as it (allegedly) really was: merely part of a constant feeling of cold. And we also still do not find a sensing of feelings of cold and warmth, clearly distinct from the feelings we might then want to suppose such sensing may “misregister.” So it is hard for me even to understand what it would mean for it really to feel icy cold to me, and yet experience this feeling as one of burning heat. Carruthers, for his part, agrees there are no illusions of inner perception (2004, 119). But I believe he does not appreciate how this is to due to the phenomenological elusiveness of the higher-order perceiving/perceived distinction that his own theory requires.

On the Phenomenology of Introspection

147

sense must be, “higher order.” But the sought-for objectual constancy is elusive: I find no change in how I sense my sensing while sensing it to stay just the same. As for registration: I find no sensing—as distinct from thinking—concurrent with the first-order feelings and appearings to be registered, which is clearly other than they. Thus, phenomenologically I find none but coincident sensing of sensing. If, on working through the distinctions and examples I have presented, and engaging with the questions they allow us to pose, you obtain the same results, then you also find inner sense phenomenologically indiscernible. (And if you do not seriously try to take this journey, you should not conclude anything at all about the phenomenology.) I do not suppose that simply closes the case. For one thing, like everyone, I am a fallible phenomenologist, and phenomenology is an open-ended, ever provisional, never finished enterprise. It is, after all, a form of philosophy. And maybe, even if my phenomenology is basically right, we are still justified in postulating inner sense because that would explain something we cannot otherwise account for. In line with such concerns, I will consider in turn two sorts of criticism, inspired by two recent, prominent accounts of introspection and consciousness. My own positive account of introspection will emerge from this.

8. Attending to Experience Has my phenomenology gone wrong? At least Lycan’s (2004) would seem to be at odds with mine. He holds that, going by the “phenomenology” (implicitly meaning this term, it seems, in both a “way of inquiry” and a “topic of inquiry” sense), there is an introspective form of attention that begs to be understood in sensory terms. He concludes that some of our “higher-order representations” are “outputs of an attention mechanism” that makes them more perceptual than thoughtlike in character. Thus, he thinks phenomenology supports inner sense.11 Now I would agree with Lycan about this much—unlike defenders of “transparency” claims about consciousness (such as Michael Tye), I think we can, in a nontrivial and epistemically significant way, attend to the character of our own experience.12 But now how can I say this, while

11 Under the heading “Phenomenology,” Lycan writes, “When we attend to our own mental states, it feels like this is just what we are doing, focusing our internal attention on something that is there for us to discern” (2004, 101). That some first-order states are “there for us to discern” he links to the notion of “presence”: they are, he says, “phenomenologically present to our minds and not (or not just) represented by them” (ibid., 102). 12 I take Tye’s 2002 defense of transparency to deny this. (See Siewert 2004.)

148 Theories of Introspection siding with philosophers like Tye and Fred Dretske (1995) against Lycan—who say that, phenomenologically, there is no inner sense? My negative thesis, I hope, is now plain enough. We do not attend to our own experience, as we attend visually to what is before us or proprioceptively to our own bodies—via an objectual sensing that allows us to make something differently but stably apparent—so that it reveals its constancy through this difference, becoming more and better apparent to us. But what is my positive claim? And just how do I respond to Lycan’s phenomenological defense of inner sense? We may start by remarking that we can attend not only to things that appear to us—but precisely to how they appear as well. “But,” it will be replied, “this in itself is no admission of attention to the experience, to the appearing and the manner of that appearing. To attend to how something looks (e.g.) may be simply to attend to its color or shape.” Yes, and certainly I would say that when you look at something’s shape, or listen closely to a sound, you are attending to how it looks or sounds, but you are not looking at a manner of visual appearance, or listening to its sounding to you as it does. But consider the form of thought you may employ with regards to how something looks, smells, sounds, etc. to you— where you have not yet informatively classified to a desired degree of specificity the color, shape, odor, sound, etc. that is apparent. With vision: you think of what color or shape something looks to you, in consideration of what general color or shape term to apply to the specific color or shape you are thinking of. You may, prior to such classification, articulate a thought of just what color, shape (etc.) it appears to you, with a phrase like ‘that color,’ ‘that shape,’ ‘that smell,’ and so on. Now you could also speak of the apparent, demonstratively identified feature in question by speaking explicitly of its appearance to you. So you could say: “the color this thing now looks to me”; “the shape this thing now looks to me”; “the way this now smells/tastes/feels to me.” (Maybe also: ‘this way of looking colored’; ‘this way of looking shaped’; ‘this way of smelling (sounding/tasting/feeling)’.) In such cases, there is no alternative formulation of the thoughts expressed that dispenses entirely with appearance words, while distinguishing these thoughts from others I might have. For I can say, ‘that color,’ ‘that smell,’ and so on, of colors, smells, etc., and demonstratively speak and think of these, in the absence of their looking (smelling, etc.) any way to me. And when I do, I may not be thinking of them as: the color that this looks to me (the odor that this smells to me, etc). But when I do in this manner think of color, smell, and so on, then it seems I can mark the difference by my thoughts’ expression only in something like the way just suggested, by introducing, “appearance” words (‘looks,’ ‘smells,’ etc.), which I may in various ways combine with demonstratives (or indexicals) in complex phrases to identify certain

On the Phenomenology of Introspection

149

features. Since appearances in this sense are, by my reckoning, “phenomenal features,” I will label these thoughts “phenomenal-demonstrative.”13 Now, as I have hinted, we can see these thoughts playing what we might call an “anchoring” role with respect to further thought about experience, in which we seek informative classification of what we first fasten onto in a demonstrative manner. For example, when I feel somehow, I may think to myself a thought that could be expressed in some such fashion: “The way this feels to me, is it really nausea? Maybe it is just a form of anxiety that is not quite nausea. Or maybe it is both nausea and anxiety—yes that’s it: anxiety lightly tinged with nausea.” “The color this looks to me—what color is that? Green . . . bluish-green . . . milky bluish-green.” “The shape this looks to me—what shape is that? Kind of circular . . . squashed or stretched . . . elongated . . . an elongated ovoid.” As my dentist pokes my gums, he asks, “Does it hurt?” “Uh-uh,” I grunt—but I am thinking to myself what I might express by saying: “What is this feeling? It’s not painful exactly, but it feels irritating, a little in the way a tickle does, but it really doesn’t feel like a tickle.” We should also note, however—such phenomenal-demonstrative thoughts can be had, even when they anchor no such classificatory reverie. Back home, in happier times (before my dental excursion), I perhaps had occasion to think (articulately or not—and with zero interest in taxonomy) what I might put by saying: “This feels fantastic!” And by ‘this’ I understand: feeling this way. Also: in those everyday aesthetic moments that help keep life bearable, I may simply be “struck” by, e.g., “the way that looks” (to me, now, from here)—thinking, say, of the aloe plant by the window “thrusting its tentacles” into space, or of the smooth sun-bleached patterns of fading paint on the side of the wall. Or—less agreeably—I may simply “be struck by” some strange odor: “the way that smells” (to me, now). When one is so struck, one takes up in thought some manner of appearance—which one could (but need not) go on to try to classify or characterize by analogy or metaphor. To understand this better, we need to examine more closely the way in which thought can “anchor” questioning. Another way of describing this: in thinking the thoughts at issue, I “identify for recognition” what I am thinking of. I do this only if I think of something in a way that enables me to recognize the correctness or incorrectness of how I go on to classify what I am thinking of, provided my 13 Roughly speaking: appearances (i.e., appearings) are “phenomenal,” on my view, because they are features essentially suited for one to claim or desire a subjective knowledge regarding what features they are, underived from other such knowledge—that is, there is essentially and nonderivatively “something it’s like for one” to have them. (See Siewert 2011b.) Note also that the expression of “phenomenal-demonstrative thoughts” in my sense does not necessarily (though it can) attach the demonstrative to the term for a phenomenal feature (e.g., ‘this way of feeling’). (Compare Gertler’s (this volume) discussion of “introspective demonstratives.”)

150 Theories of Introspection understanding of these classifications is not defective. So, I want to say: when I think of the way it now feels to me, or the color this now looks to me, I identify the feeling of which I am thinking or the color of which I am thinking, in a manner that enables me then to recognize the former as, say, itchy (not dizzy), and the latter as, say aquamarine (not royal blue)—provided that there is no defect in my understanding of ‘itchy,’ ‘dizzy,’ ‘aquamarine,’ or ‘royal blue’ that would impair my capacity to make these classifications. (Note: when I say, “I identify the color I am thinking of,” I do not mean that I come to think that I am thinking of this color. I identify the color I was in fact thinking of—though I do not, necessarily, in so doing, think of it as “what I am thinking of.”) That some such thoughts play this role is not guaranteed by the fact that they are expressed with definite descriptions embedding demonstratives or indexicals. I may think of “the color of the first pair of socks I purchased” while having, in a sense, no idea what color that was, that is, without thereby identifying what color I am thinking of, in a way that would put me in a position to recognize blue as a correct, brown as an incorrect classification. And that is so, even though there is no relevant defect in my understanding of ‘blue’ or ‘brown’. Similarly with “the color of these socks,” uttered as I am holding up the very socks to which I refer, in a light too dim for me to see their color. But it is hard to understand how I could express a thought with the phrase “the color this looks to me now” or “the way I now feel,” and thereby think of some color apparent to me, or some manner in which I feel, without identifying these features for recognition, in my sense. In any case, I can, as illustrated, in some contexts think such identifying, phenomenal-demonstrative thoughts. Notice too now that the thought involved in this “identification for recognition” is distinctively first-personal. That is, in such contexts the way I understand: “the color that this looks to me” is not the same as the way I understand “the color this looks to you.” (Although, I can indeed think thoughts expressible by both phrases.) The difference does not simply lie in the difference in my understanding of ‘me’ and ‘you’. For it may be both true and informative that the color this thing looks to me = the color this thing looks to you. But that is not because I may come to find out that I am you. I would go further, and say that, when I do identify for recognition some feature in first-person phenomenal-demonstrative thought, I cannot fail to actually experience the appearance identified, insofar as I successfully form the thought at all. (I mean ‘experience’ here in the coincident sense.) That is to say, if I identify the way of feeling of which I am thinking, by thinking of the way it feels to me now, I do in fact feel the very way I am thinking of. If I identify some color of which I am thinking by thinking of the color this now looks to me, then the color this now looks

On the Phenomenology of Introspection

151

to me is in fact the color I am thinking of. By contrast, I may think a thought I would express by saying “the color (or shape) this thing now has,” and in so doing identify for recognition which color (or shape) I am (in fact) thinking of, even though the color or shape the thing actually possesses is not the color or shape I am thinking of. At least, I cannot make out in what manner I could identify for recognition a specific way of appearing, in phenomenal-demonstrative first-person thought, without actually experiencing that way of appearing. I find no inner sense, so I cannot, by appeal to a false “inner” appearance of appearances make intelligible how, for example, I could identify what shape I am thinking of, as the shape this looks to me, even though the shape this in fact looks to me is not the shape I am then thinking of. And—although this bears more looking into—I do not think there is any other good way to make out how such a scenario could arise in relevant cases.14 Assuming that is right, this aspect of first-person “identifying for recognition” thought makes for a contrast with the corresponding second- and third-person cases. When, for instance, I think of the shape this looks to you, either I simply have not identified what shape I am then thinking of (“the shape this looks to you . . . whatever that may be”), or else, if I have, it could easily happen that the shape this actually looks to you is not the specific shape I have identified in my thought.15 If this is correct, then the sort of first-person thoughts that can anchor reflection on experience are essentially appearance-dependent. I mean: one has these firstperson, phenomenal-demonstrative thoughts that are suited to the identification of features for recognition only if one actually experiences oneself the very appearances identified. Since for me, to “experience an appearance” is to have a phenomenal feature, I will also say: one has first-person thoughts that require one’s possession of the very phenomenal features they are thoughts of. Pulling all this together, we can see how by means of first-person reflection we can demonstrate to ourselves the reality of what we may now fairly call introspective cognitive attention to experience. First, what I have termed “thinking about experience”

14 Notice we will not do this by reference to the sort of cases Schwitzgebel and Smithies (this volume) both discuss, in which an intense, anxious expectation of pain leads one to think, momentarily, that one feels pain. Perhaps, in a panic I briefly think I feel pain, though I do not. It does not follow that I can also then entertain a phenomenal-demonstrative thought of the form the way this feels to me, which enables me to identify a specific way of feeling of which I was then thinking, even though in fact that is not the way this feels to me. 15 One way this could happen: I identify for recognition that shape I am thinking of, by wrongly assuming that the shape this looks to me = the shape this looks to you. Another possible way: I wrongly believe the shape this looks to you is the shape I assume it would look to me from there, where you are.

152 Theories of Introspection is indeed rightly called thinking (hence cognitive) because it is verbally expressible in the manner of thought (e.g., I say, ‘the color this looks to me’)—and it can function in reasoning. Second, what you are thinking of here (hence attending to) is indeed your experience, because you think not merely of what appears to you, but of (an instance of) its appearing to you as it does—i.e., an experience. (This is shown by the need for appearance words when trying to distinguish such thoughts from others, and by the unavailability of such thoughts in the absence of your actually experiencing the manner of appearance they are about.) Third, when you think of experience here, you do indeed attend to it—and not simply because you might be said to “attend” to whatever you (consciously) think of. For this is thought capable of “anchoring inquiry”—of initiating the identification of a continuing common topic for subsequent, otherwise varying cognition—say, a series of questions and answers. (And if something does that, it is not attention in some merely lax or trivial sense.) Finally, this attention is specifically introspective, because it is a form of thought of one’s own experience, distinct from any in which one may consider another’s. If what I have just said is supposed to be at odds with the idea that experience (or consciousness) is “transparent,” then so much the worse for that doctrine. Later I will say more about how this helps us explain the special sort of warrant with which we may judge of our own experience. But I now want to say how it helps explain the appeal of inner sense. If you accept my phenomenology, you accept that there is a kind of first-person thought about experience which: a. is demonstratively expressible; b. anchors inquiry so as to enable one to recognize correct classification of what it is about (it enables “identification for recognition”); c. involves attending to what it is about; and d. is appearance-dependent. Note now that features (a)–(d) can be found not just in a special sort of thought about one’s experience but also in thoughts about the objects of first-order appearance. There too—when, e.g., I am thinking about the thing that I see—I employ a thought that: (a) is demonstratively expressible (‘that shoe’); and (b) anchors questioning and allows me to recognize correct classification of what it is about (“What sort of shoe is that? A loafer—isn’t it?”—“Yes, that’s what it’s called.”) And clearly, in thinking about this shoe, I am (c) attending to it. And lastly, since I think about it by means of looking at it, and in looking at it, it looks some way to me, we may say the perceptual thought is (d) appearance-based. That is, I understand what thing I mean here by ‘that,’ through its visual appearance to me. How does this explain the appeal of inner sense? Since (a)–(d) applies to firstorder thought about perceived objects in virtue of our sensing the objects of such

On the Phenomenology of Introspection

153

thoughts, we are tempted to suppose that it must also apply to reflective thought about experience in virtue of our sensing the objects of these thoughts. So we are tempted to think there must be two levels of sensing, one for each object of thought. But this is one temptation we should resist. We can explain why the same description (a)–(d) can apply to thoughts about one’s experience, and to thoughts about the things one experiences, without appeal to some higher-order iteration of sensing. The same description applies to both simply because some first-person thoughts about one’s experiences are also thoughts about the things one experiences. “The shape this looks to me” can be used in the expression of such a thought. But what do I say to Lycan’s claim that we need inner sense theory to account for the phenomenology of attention? We can now accept, as a phenomenological matter, that when, for example, one looks at a red thing, and one thinks about its looking to one as it does, the acts of attention are somehow similar—or at least, there is something they share. But this requires no extra layer of sensing. Rather: the thought of the appearance of the red object, which thought constitutes attending to that experience, essentially involves the same manner of visual appearance that also constitutes attending to—that is to say, looking at—the red object. In other words, the visual appearance—this first-order sensing—can do a sort of double duty. It can comprise an act of attending to a visible object, even as it helps constitute a cognitively attentive phenomenal-demonstrative thought about itself. That is what is “shared” between the two modes of attention. We would have reason to reject this perspective in favor of something more like Lycan’s, if the phenomenology of attending to experience revealed a higher-order objectual sensing. But we have seen that it does not. So I maintain the phenomenology of introspective attention supports the “double duty/cognitive attention” view I have outlined, as against an inner sense alternative.16 Please note: to speak of

16

Maybe my criticism of Lycan will make me seem close to something like Dretske’s (1995) and Tye’s (2002) “displaced perception” views of introspection. (See also Byrne’s defense of the “transparency proposal” in this volume.) Tye, for example, says: “We attend to one thing—the external surfaces and qualities—and yet thereby we are aware of something else, the ‘feel’ of our experience” (2002, 51–52). Isn’t it also true that on my view, by attending to (looking at) a colored surface, I am (in thought) aware of its looking to me a certain way—since the former “helps constitute” the latter? Yes, you could put it that way. But for the following reasons I do not think of this as displaced perception or indirect awareness. (1) Sometimes I think of the colored surface, and of what color or shape it has, by thinking of (and thereby attending to) its appearance to me (in the phenomenal-demonstrative manner indicated above). In that case I do not think of the color or the shape more directly than I think of their appearance to me, since I conceive of the color/shape in terms of their appearance. (2) I do not think of the appearance, only by means of being in some other way aware of the appearance. (There is no other “mediating awareness” of the appearance.)

154 Theories of Introspection “double duty” here is not to say that visual appearances, in addition to “representing” visible objects, also somehow “represent” themselves. I myself do not really understand how the visual appearance I enjoy—which is not itself part of what is visually apparent to me—can “represent itself.” To my mind, if visual appearances “represent” anything, what they represent is what they are appearances of—that of which they are objectual sensings, that of which they are either accurate or inaccurate appearances. But the visual appearance is never then somehow “of,” or “about” itself, it does not “refer to” itself. Nor for that matter does it “present” itself to you— either just as it is, or as other than it is. For the visual appearance you experience is never an object of appearance to you—as distinct from an object of your thought.17 However, it can figure essentially in a thought about itself. In this way it can be “present to you” (or, as some say, be “given” to you) in the sense that it is (as Lycan says) “there for you to discern” first-personally, and not merely be “thought of ” (represented). And this can be a matter of phenomenology. But this does not mean it has to be presented to you in the sense of: be sensorially apparent.

9. Do Recognitional Concepts Require Inner Sense? I turn now to an objection suggested to me by reading Peter Carruthers: whether or not phenomenology can support belief in inner sense, we should posit such a faculty, because that best explains something we should in any case admit—namely,

(3) I would not say that I tell from (what I believe to be) the actual color or shape of the surface how its color or shape looks to me. For (again) my phenomenal-demonstrative grasp of what I take the actual color or shape to be is precisely in terms of its apparent color or shape (“the color/shape this looks to me”). Also, I might believe the actual color or shape of an object to diverge from its (merely) apparent one, because I have resolved the question of what its actual shape/color is, by taking other of its appearances to me to better reveal this. But then I am determining how the object is from its (favored) appearances, rather than determining how it appears from how it is. (4) It may be true that the surface’s actually looking to me as it does reliably makes me think it looks to me that way. But I am unpersuaded this has to be mediated by some inference from the belief that the content of the visual appearance is correct, to a belief that I experience this visual appearance. 17 Although my views are in some respects close to Horgan’s (this volume), on this point I seem to part ways with him. (See also note 21.) Here I take myself to agree with Husserl when he says: “The appearing of the thing (the experience) is not the thing which appears (that seems to stand before us in propria persona). As belonging in a conscious connection, the appearing of the thing is experienced by us, as belonging in a phenomenal world, things appear before us. The appearing of things does not itself appear to us, we live though [i.e., experience] it” (Husserl [1900–1901] 2001,Vol. II, 83) More generally, I should acknowledge here how much my thoughts in this chapter are indebted to my reading of Husserl’s Logical Investigations.

On the Phenomenology of Introspection

155

that we form “recognitional concepts” of our manner of experiencing the world.18 I do not want to endorse all that may be linked to the idea of a “recognitional concept.” Still, I take it that sensory recognition gives us a way of conceiving of kinds, which requires little or nothing in the way of a “theoretical” account specific to them. Loar (2002) offers the example of a kind of plant, a succulent you come across in the desert for the first time, about which you frame a concept of the form, ‘one of that kind,’ which you can successfully apply to a range of instances perceptually encountered, though you apparently have little or nothing more to say about what that kind is, beyond, perhaps, it is a kind of plant. I am wary of supposing that we really have concepts, strictly speaking, that are radically independent of discursive abilities—so-called “pure” recognitional concepts that do not require any relevant abilities of inference and analogy, to be exercised in thinking about real and hypothetical cases. But I do accept there are appearance-based concepts of features possessed by objects of (first-order) sensing. And I would say we have concepts of the types of appearances themselves (phenomenal features) through experiencing them (in the coincident sense), which concepts (due to their experiential provenance) can be had without a properly “theoretical” understanding of what the conceptualized features consist in, or even anything but fairly minimal capacities for inference and analogy, using the concepts concerned. This bears some similarities to what Carruthers says about recognitional concepts. He also accepts a thinking/sensing distinction—though he interprets this in terms of a difference between a kind of “mental representation” that is “coarsegrained” (“digital,” or “conceptual”) and one that is “fine-grained” (“analog,” or “nonconceptual”), and he does endorse the idea of “pure recognitional concepts.”19 Setting aside for the moment whether we should, and whether “fineness of representational grain” can explain the thinking/sensing distinction (which I also doubt), let’s focus on this idea that we are justified in thinking that distinction iterates, because this would explain how we have (some sort of ) recognitional concepts, of both first- and second-order objects. This case for inner sense, as I understand it, depends on roughly the following line of thought. 1. Our first-order sensing of objects accounts for our having recognitional concepts of them.

18

Carruthers 2004, 121ff. Lycan (2004, 109) also endorses this type of argument for higherorder perception. 19 Carruthers 2004, 121:“. . .We have, or can form, recognitional concepts for our phenomenally conscious experiences that lack any conceptual connections with other concepts of ours. . . .”

156 Theories of Introspection 2. We also have recognitional concepts of our ways of first-order sensing. 3. What makes first-order sensing explanatory of first-order recognitional concepts (as in (1)) can also explain our having the recognitional concepts acknowledged in (2). 4. More specifically: positing a distinct second-order sensing that shares the features relevant to making first-order sensing explanatory (as in (1)) provides the most satisfactory way of explaining our possession of the recognitional concepts in (2). I myself would accept (1)–(3), rightly understood. But I reject the move to (4). A second level of sensing seems to me unnecessary and unable to do the work it is asked to do. First, why it is unnecessary. Consider the “recognitional concept” I have of a certain flavor. I may be helpless to say much about what this flavor is, or even to classify it informatively, beyond saying what things I find it in—ripe figs, let us suppose. Otherwise I may simply identify it as “that flavor.” The sensing that accounts for my possession of this concept of the fig’s flavor is its tasting a certain way to me. And so I might also articulate my way of understanding ‘that flavor’ by saying something like “the way this [paradigmatic fig] tastes to me.” But this certainly does not mean that, unless I can form such thoughts, and articulate concepts in this way, I can possess no appearance-dependent recognitional concept of, e.g., a flavor, at all. As a child, still pre-reflectively toddling around, I first learned to recognize a flavor—and think of it (“It’s yucky!”)—through the way it tasted to me, even when I was as yet unable to think of it as a way of tasting. I may have these appearance-dependent concepts, even though (lacking the cognitive maturity to distinguish, in thought, things from their appearance) I do not as yet also have concepts of appearance. But now to make this developmental advance to reflection, and go from merely having a first-order (“recognitional”) concept of a flavor to having a second-order (“recognitional”) concept of the way of tasting on which the first concept depended, what more do I need with respect to sensing? Do I need, in addition to the tasting, a separate sensing of the tasting, to enable me to think of the way it tastes to me? On my account, that way of tasting itself can and must figure in the formation of such a thought. But then I want to ask, what more could an inner sensing of the tasting furnish to the thought, which it could not simply get from the tasting itself ? The postulated, phenomenologically indiscernible extra “volley” of sensing is supposed to somehow reproduce a representational fineness of grain found in sensing’s first round. But no useful work is done by this covert intermediary that somehow representationally matches the first-order representation it is supposed to join with reflective thought. For such a hidden proxy apparently has nothing relevant to offer

On the Phenomenology of Introspection

157

the thought, which it would not just be able to get more directly from the original manifest appearance—the taste experience, say. And we certainly do not want to suppose, on pain of regress, that to get a representation of a representation, one always needs a third representation as a go-between. All the sensing I need can be found at the first level. The same appearances that first provided me with a firstorder recognitional concept (by being appearances of objects) also provide me with a second-order one (by helping to constitute reflective thoughts about them). Again, since sensory appearances do a kind of double-duty, they do not need to be doubled. Thus, we can accept that what makes first-order sensing explanatory of our recognitional concepts of its objects also explains our having such concepts of these very ways of sensing—but without introducing a second level of sensing, an inner sense. Suppose this is right. Does there still perhaps remain this place for inner sense? You say: “Granted there are these phenomenal-demonstrative thoughts that enable ‘identification for recognition.’ But what do we call the tie that binds phenomenal features in such thoughts? Why not ‘inner sense’?” I allow that you can use whatever label you like for this intimate relationship of experience and thought. You might also suggest: ‘intuition’, ‘inner perception’, ‘self-givenness’, or ‘acquaintance’. And maybe you can explain what you mean by these phrases, so as to add to what has already been said regarding how first-person thought about phenomenal features is possible. But I fear such terminology will obscure the problems with inner sense I have been at pains to point out.20 Anyway, if we do speak in such terms, I would still resist the idea that it licenses us to believe in a more-than-first-order mode of “intentionality,” “mental reference,” or “representation,” distinct from and underlying essentially appearance-dependent first-person thoughts.21

20

I should note however, that there are ways of spelling out an “acquaintance” view that would overlap significantly with mine. For Gertler (this volume) holds that “some introspective knowledge consists in judgments that are directly tied to their truthmakers.” My own view about appearance-dependent thought and how it figures in self-knowledge would seem to be in agreement with at least this aspect of her “acquaintance approach.” 21 One may say that what I am calling “identification for recognition” suffices for perception in a very broad sense—perhaps matching Husserl’s broad use of ‘Wahrnehmung’. And so, with some terminological adjustment, I could allow for “higher-order perception” after all. Also, what I have said so far does not address the (also Husserlian) idea—in the background of Smith’s (2005) and Zahavi’s (2005) accounts—that there is a special nonobjectual form of intentionality, exemplified in (but not exhausted by) the primitive way experience always anticipates experience to come and retains experience just-past, wherein consciousness is always “pointing at” itself. And I still need to address directly the kind of proposal found in Kriegel (2009) and Williford (2006) that consciousness is ubiquitously “its own object” (always “self-representing”), though this may be properly neither a form of sensing nor a form of (propositionally articulable) thought. But I think my

158 Theories of Introspection I have now argued that the claim of an additional stratum of sensory intentionality or representation is unjustified, on the grounds that it is phenomenologically indiscernible and explanatorily superfluous. Now I want to add why I think such a posit would not be up to the explanatory job for which it was introduced. First consider: brain damage might leave intact some sort of vision without phenomenal visual appearance—as in blindsight. And there is reason to think nonphenomenal differences in visual detection modulate our actual motor responses. But, with nothing but that kind of vision, I would not have the recognitional concepts of shape and color that I do. So I could not have the recognitional concept of that shape (or color) that I actually get from vision, if that shape (color) did not look anyhow to me, where we interpret “looking somehow to me” to be sufficient for phenomenality. (Notice we can accept such an interpretation, while remaining neutral on the question of whether some kind of self-representation is buried within the “looking,” and responsible for its phenomenality.) Now we have been urged to accept inner sense on the grounds that, to account for our second-order recognitional concepts, we need to posit at the second level the concept-supplying sensing we needed at the first. But if that is so, then it seems we will also need this second-order sensing to be phenomenal, for only this yielded us recognitional concepts at the first-order level. If nonphenomenal sensing would not account for our first-order recognitional concepts, we have no right to think it could account for our recognitional concepts “higher up.” However, it is admitted that second-order sensing eludes critical first-person reflection; it is phenomenologically indiscernible. In that case then, the proposal in question will tell us that we ourselves have—ubiquitously—phenomenal features that are actually undetectable by our own first-person reflection. Oddly then, there is supposed to be (nonderivatively) something it’s like for us to have these features—even though we can’t know what it’s like; we can only wonder what it’s like for us to have this hidden inner sensing of our own sensing, over and above what it’s like for us to sense at the first level.

remarks prepare a challenge to these views. How can we phenomenologically distinguish this allegedly ubiquitous self-pointing of experience from the coincident “experiencing of experience” that—since it admits no distinction between what is experienced and how it is experienced— admits none between either (nonobjectual) “pointing” and “pointed at,” or (objectual) “representing” and “represented”? Moreover, how does it really help me to judge about my own experience that it is constantly “pointing at itself ” in this rarefied way? How does its pointing at itself get my thoughts pointed at it? Is it supposed to be drawing attention to itself ? Wouldn’t we then need to objectually sense the self-pointing object’s display, to guide our thoughts to it? On my view, we obviate these problems when we see that the form of cognitive attention involved does not need to be guided to experience, since it cannot even arise without it.

On the Phenomenology of Introspection

159

Now it’s not clear to me it is inconceivable I could have such features—phenomenal but stubbornly unintrospectible to me. Still, if there are features I (at least allegedly) have, but I cannot tell what it’s like for me to have them, and—as they are introspectively inaccessible—I cannot even learn to tell what it’s like for me to have them, that seems to be a good reason to conclude there just is not actually anything it is like for me to have them. For what would prohibit me from knowing or learning this, were it there to be known? But if there is not anything it is like for me to innerly sense, if this has no phenomenal character of its own, then inner sense would be no more suited than blindsight to supply me with recognitional concepts of what it senses.

10. Introspection without Inner Sense I have offered a view concerning introspective attention and recognitional concepts that abjures inner sense while honoring its appeal. This is also intended to provide the basis for an alternative account of introspective knowledge. It is not possible to develop that fully here. But I want to say enough to indicate its promise. One traditional theme in discussions of introspection is the question of its fallibility. It may now seem I am poised to say that the phenomenal-demonstrative thoughts central to my account enjoy a kind of infallibility—for didn’t I say that I can have these thoughts only when I actually have the features they are about? So it seems if I have such thoughts at all they cannot fail to reach their target, so to speak. This is a reasonable gloss on my view—but the talk of infallibility can be misleading, so let me try to be precise. Because there is no inner sense, there is no erroneous inner appearance that might allow me to understand which phenomenal features I think I have, even when I do not actually have them. And I am often not in a position to be led to such mistakes by faulty or misconstrued observations of what I am doing, as can happen when I am considering another’s experience. So it seems I could somehow “go wrong” in thinking, for instance, what I would express as “the color this now looks to me,” when I putatively identify a feature for recognition, only if either: (a) there is just nothing answering to my ‘this’—perhaps because I have a (first-order) hallucination; or (b) I do not understand what I am saying (in which case I would not so much have a failed thought, as simply failed to form one). In the (a) case, I would say that I would still understand what I mean by the italicized phrase and would still have succeeded in thinking of some phenomenal feature I possess—some visual appearance. And if we insist that the reference failure should make us say otherwise (or at least view the expression of the thought as defective), it seems there are reformulations that could reduce my propensity to

160 Theories of Introspection make an error here while still understanding what I am saying. I might speak instead of “the color it looks to me as if something there has,” or “this way of looking colored to me,” or “the way this color looks to me.” I will not try to deal now with the questions this raises. For now, I just want to say it seems plausible that, once we get our focus on the sort of phenomenal-demonstrative thoughts that can play an anchoring role in first-person reflection, then we will have focused on a kind of thought we cannot conceive of as an erroneous or mistaken thought about experience as long as we understand its expression. Now whether one does understand—and does not seriously misunderstand— the purported expression of such a thought is open to question. This is why talk of infallibility here can be misleading. Even if there are certain judgments you cannot make falsely, you may remain fallible about just when you have made one of these. However, this need not lead to skepticism. Consider: a presumption that you understand what you are saying is a condition of rational verbal thought. You have to presume an understanding of what you are saying, absent some reason to think otherwise—or else you cannot even get in the verbal reasoning business. Now, if what I said earlier was right, there are thoughts I can have about certain (phenomenal) features at all, only if I am thinking about some such features I actually have. If I am not thinking about one I actually have, when I purport to speak of, e.g., a certain color now appearing to me, via an expression like “the color this now looks to me,” I do not know what feature I am thinking of, and thus would not understand what I meant by this phrase on this occasion. Since I am entitled to presume I do understand what I mean by the expression of my thought, I am warranted in thinking I do have the feature I took myself to speak of. Thus, the conclusion is that I have some special warrant in thinking I have identified certain of my own phenomenal features—a warrant I do not have for taking myself to have identified those of others. Just how might I express this identification of my own phenomenal features in a complete judgment? To be sure, “the color this now looks to me” does not do the trick. It seems I would need to say something like: the color this now looks to me is, . . . well, the color this now looks to me. Or: the way this feels to me is the way this feels to me. Of course, this has the absurd appearance of an empty tautology—but that is misleading. For in the relevant contexts, what would lie on either side of the identity judged is my identification of some specific feature I actually have. And if there is some specific feature I have warrant for judging myself to have, I do not merely have warrant for an empty tautology. Nevertheless, the judgments in question are still pretty meager. But we can get farther if we add this. The warrant we have for the judgments at issue is not globally independent of the warrant we have for others. We cannot reduce our

On the Phenomenology of Introspection

161

phenomenal-demonstrative thoughts to some epistemically self-standing, bare demonstrative core, stripped of all informative classificatory terms. I would not be able to think of my experience of color if I could not classify it somehow— e.g., as a case of something’s looking somehow to me. In other words, I do not think we are endowed with some power to think of our own experiences purely demonstratively (“This is this”), independently of any ability to correctly classify them. For I do not think one could understand what such a ‘this’ would refer to—an experience, not just an object experienced—without some prior competence in the use of “appearance” vocabulary. Now further: there is no warrant to be had for thinking someone has come to understand the terms of that vocabulary (‘looks’ and so on) without being able to form correct judgments joining it with other types of predicates (e.g., for colors). Neither you nor I have call to think I have acquired an understanding of ‘looks’ in the absence of having come by a capacity to join it with other, for example, color terms, and truthfully say things like, “That looks blue to me.”22 More generally, nothing would warrant ascribing to me the acquisition of competence in using ‘looks’ in the face of total incompetence in deploying color, shape, and other terms with which to fill the blank in ‘looks . . .’ phrases. (Although, there may be weird dissociations through a pathological deterioration of prior competence.) Thus, if the presumption that I know what I mean by ‘looks’ yields some presumption of warrant for judgments about myself that employ this term, it will yield some presumption of warrant for more specific judgments about how things look to me, even if, in a particular case, I might reasonably doubt whether I am correct in my classification of how they look to me. If this is right, we can start to account for our first-person warrant for judgments about sensory appearance. When I judge that I feel an itch, or that this looks blue to me, what makes me warranted in so judging is the fact that I could (even though normally of course, I do not) think a phenomenal-demonstrative thought— expressible using some such phrase as “the way this feels to me now” or “the color this looks to me now”—which could not fail to pick out a manner of feeling or looking—a phenomenal feature—that I have, on pain of not being intelligible to me. Now presuming, if defeasibly, that I do understand what I would mean by such phrases, in an “identifying for recognition” use, I have warrant for judgments like “the way this feels to me now is the way this feels to me now,” or simply: “I feel this way now.” And since my presumed competence in speaking thus of feeling cannot 22

For these reasons, I find it hard to accept the notion of “pure phenomenal” beliefs or concepts, detachable from a competence in “labeling.” (Compare Gertler’s and Horgan’s discussions in this volume, and Chalmers 2003.)

162 Theories of Introspection be utterly detached from my competence in further classifying some feelings more specifically as, say itches, I am also entitled to judge that the way this feels to me is itchy (or this way I feel is itchy), if I am so inclined to classify it, absent a specific reason to doubt this. (Although, I could, in some (not all) cases, given special motivation, entertain a doubt about my classification and retreat to the “way that this feels” thought.) And if I have warrant for judging the way this feels to me (this way I feel) is itchy, I have it for judging I feel an itch. Similar remarks would apply to ‘looks’ and predicates of color or shape.23 This gets us a little further. But now we may ask: what might strengthen our warrant as our self-reports become more adventuresome, and it becomes less clear that errors would reflect defects in basic semantic competence? What if I want to pronounce introspectively on some of the questions that have figured in philosophical and psychological disputes—and indeed, on the very issues that have emerged here? Consider this. It has been alleged that “naively” we are subject to some introspective illusion about the detail and richness of our visual experience. One way Dennett (2005, 45) puts this is that people think “their visual fields are approximately as detailed and fine grained all the way out.” (See also Dennett 1991, 68.) Suppose one were actually to make some such (naive) claim about one’s own experience, with no ostensible basis other than (generic) introspection. And let us grant that such a claim would be mistaken. How are we to understand its fallibility? Are we to say perhaps that we fall into error here because we venture into topics where first-person reflection simply dare not tread? Perhaps we will declare the whole issue off limits to responsible introspection, which we should confine to the blandness of “baby judgments” like “that looks orange to me” and “I feel an itch.” Such confinement would be almost as intellectually self-mutilating as behaviorism. We would more fruitfully use first-person reflection by considering how relatively general pronouncements on the character of experience have implications

23

Why all these complications? Why isn’t it enough simply to say that the fact that I feel an itch makes me warranted in judging that I do? (Compare Smithies “simple theory” of introspection in this volume.) It is not that I think this simple statement is wrong; I add the complications I do to explain why (e.g.) my feeling an itch warrants my judgment that I do. My proposal is that it does because the kind of cognitive attention that can underlie such self-descriptions, though it cannot guarantee correctness, can function at all only against a background of competence with the terms in which it is expressed that entitles me to a presumption of correctness, which may in specific cases be overturned by an examination of my understanding of what I say or would say about my experience. Moreover, only once we see this, can we see how introspection can be something other than just a detection mechanism, how it can be self-correcting, and how it can function in phenomenological investigation, where our questions go beyond ones like, “Do you feel an itch?”

On the Phenomenology of Introspection

163

for the description of more specific cases, which emerge through questioning that renders explicit distinctions that might otherwise be ignored. In this way we might ground the inquiry by moving to cases where the warrant for our judgment is closer to the sort illustrated just now with reference to color and itch experiences— where it is harder to entertain the hypothesis that we are in error, without throwing our understanding of what we were saying into doubt. And in this way introspection can check itself. This does not assume that I can isolate, once and for all, some privileged class of propositions that cannot be intelligibly expressed without being regarded as true, just that, in the context of inquiry I cannot see how to maintain an understanding of a certain distinction I have ostensibly employed, if I am misapplying it. I may still be open to the suggestion that in fact my grasp of the distinction is seriously defective, or that the prospect of misapplying it here does not really undermine my grasp of it. But I need some positive reason to think one or the other of these is the case in the instance in question, before such doubts are warranted. Let’s see how this procedure would work in the sort of case Dennett discusses. In the face of a claim to have a “nearly uniformly detailed and focused visual field all the way out,” we may say, “Granted that when you look at the words on a page, even if only briefly, in some sense, they all appear to you to be uniformly detailed. But when you look directly at, say, just one of the words, do the rest of them appear to you in nearly as much detail as the word you are looking at? When you shift from looking at one word to another does it appear to you in more detail? Is there no big difference between the way the words that you’re looking at look to you, and the way the nearby words you’re not just then looking at look to you?” By pursuing such questions, one could correct a carelessly formulated or interpreted introspective claim, partly by appeal to first-person reflection on specific cases, guided by distinctions at first neglected—like that between an area’s looking to you equally detailed throughout, and every part of it appearing to you in as much detail, a distinction whose comprehension I do not, as it stands, see how to divorce from its successful first-person application. My proposal then is that, as we get beyond questions like whether you feel an itch, a pain, or a tickle, the strength of the warrant to be had for first-person reflection will depend on assiduously seeking out, asking, and answering in good faith questions that make explicit pertinent distinctions and implications, and on our sustaining coherence in our replies under persistent, wide-ranging, and honest examination. I have been trying in this very chapter to do just that, eliciting responses that allow us to construct a reasoned view on topics of central theoretical concern—the reality of inner sense and the character of introspective attention. (So, for example, the more ambitious claims about this are evaluated in light of a distinction between something’s constant shape appearing to us and its shape appearing

164 Theories of Introspection to change.) By means of this open-ended (“dialectical,” “Socratic”) use of introspection we can responsibly address questions about the mind. And I would venture: if—whether from sloth or prejudice—we refuse such reflection, then in a sense, we are just not taking responsibility for ourselves. I realize that many issues still face my account. A big one is whether my sort of introspection is limited to knowledge of sensory appearances. I believe it is not, and my approach can be extended to elucidate our introspective knowledge of thinking as well. This is partly because, on my (Siewert 2011b) view, thinking is as phenomenal as sensing—so that something like the way phenomenal features figure in the account of introspection I am proposing can carry over to the cognitive case. But that is another story. I have tried to say enough here about introspection and sensory appearance to rear a plausible positive proposal on my critique of inner sense. This preserves traditional ideas associated with talk of introspection, insofar as it accords important roles to consciousness and to attention to experience in accounting for the special right with which we judge of our own minds. This right is rooted in a special access we have to our own conscious states—that is, instances of our own phenomenal features— since our cognitively attentive, distinctively first-person thoughts about experience would not be possible, did we not identify, without sensory mediation, such instances as we actually enjoy. However, my account does not tie the epistemic role of (phenomenal) consciousness to its allegedly self-referential or self-representational nature—and here I part with certain currents of tradition. Moreover, I do not accept the usual opposition of “inner” and “outer.” Attention to experience is not some senselike turn of a “gaze” to the “inside,” away from the “outside”—it is no withdrawal from the “outer” to the “inner.” But that is not to say we are somehow condemned to attend solely to “external” qualities and objects. Rather: with a special form of first-person thought we can attend at once both to what lies around us and to our way of experiencing it. (We need respect no demand to carve this attention up and distribute it between an “inner” and an “outer.”)24 And the form of 24 Compare Tye: “Our attention goes outside in the visual case . . . not to the experience inside our heads” (2002, 51). This seems like it is supposed to be an introspective remark. But I do not think I can tell introspectively that my visual experience is contained entirely in my head—if this is even the right way to think about its boundaries. So when I say one can tell introspectively that one attends to one’s own experience (and not just to its objects) I am not quarreling with the claim that introspection does not reveal us turning our attention away from the world to “something inside our heads.” Generally I reject the notion that we attend to our sense experience only if we shift our attention away from perceived things (“outside”) to our experience of them (“inside”). (Consider also Byrne’s remarks on introspection and attention in this volume.)

On the Phenomenology of Introspection

165

thought concerned indeed arises only through the “presence” (one’s actual possession) of what it concerns—namely, a specific manner of experience. But this is not to say that experience is presented to us for being thought about, as something sensed is presented for consideration—though what is “in the mind” (unlike “external reality”) is somehow presented in its entirety, hiding nothing. That would be confused. It is true that we cannot attend better to an experience by sensing more of it (as when we attend more closely to what we see). Still, we can—in another way—introspectively attend better to experience, so as to discover more about it: we can attend “more closely” to experience by asking more (and better) questions about it, which reflect finer, more articulate, and more integrated distinctions—that is, by practicing as well as we can what I call “phenomenology.”25

25

For their helpful questions and criticism, I would like to thank members of the audience at three conferences in which some of the ideas in this chapter were presented: Philosophy of Mind Conference at East Carolina University (2004); Maribor/Ohio State/Rijeka Conference on The Philosophical Significance of Attention, Dubrovnik (2009), and the Tucson Consciousness Conference (2010). For instructive feedback on these topics and on earlier drafts, I particularly want to thank Steve Crowell, Imogen Dickie,Terry Horgan, Uriah Kriegel, Bill Lycan, Fiona Macpherson, David Pitt, Anna Christina Soy Ribeiro, David W. Smith, Declan Smithies, Daniel Stoljar, Sebastian Watzl, and Dan Zahavi.

166 Theories of Introspection

References Armstrong, D. 1968. A Materialist Theory of Mind. London: Routledge. Burge, T. 2010. Origins of Objectivity. Oxford: Oxford University Press. Carruthers, P. 2004. “HOP over FOR, HOT Theory.” In Higher-Order Theories of Consciousness, edited by R. Gennaro. Philadelphia: John Benjamins Publishers. Chalmers, D., ed. 2002. Philosophy of Mind: Classical and Contemporary Readings. Oxford: Oxford University Press. ———. 2003. “The Content and Epistemology of Phenomenal Belief.” In Consciousness: New Philosophical Perspectives, edited by Q. Smith and A. Jokic. Oxford: Oxford University Press. Dennett, D. 1991. Consciousness Explained. Boston: Little, Brown. ———. 2005. Sweet Dreams. Cambridge, Mass.: MIT Press. Gennaro, R., ed. 2004. Higher-Order Theories of Consciousness. Philadelphia: John Benjamins Publishers. Husserl, E. [1900–1901] 2001. Logical Investigations. Translated by J. Findlay. London: Routledge. Kriegel, U. 2009. Subjective Consciousness: A Self-Representational Theory. Oxford: Oxford University Press. Loar, B. 2002. “Phenomenal States.” In Philosophy of Mind: Classical and Contemporary Readings, edited by D. Chalmers. Oxford: Oxford University Press. Lycan, W. 2004. “The Superiority of HOP to HOT.” In Higher-Order Theories of Consciousness, edited by R. Gennaro. Philadelphia: John Benjamins Publishers. Rosenthal, D. 2002. “Explaining Consciousness.” In Philosophy of Mind: Classical and Contemporary Readings, edited by D. Chalmers. Oxford: Oxford University Press. Shoemaker, S. 1996. The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press. Siewert, C. 1998. The Significance of Consciousness. Princeton: Princeton University Press. ———. 2004. “Is Experience Transparent?” Philosophical Studies 117: 15–41. ———. 2007a. “In Favor of (Plain) Phenomenology.” Phenomenology and the Cognitive Sciences, Special Issue on Dennett and Heterophenomenology,Vol. 6, nos. 1–2. ———. 2007b. “Who’s Afraid of Phenomenological Disputes?” In Supplement to Southern Journal of Philosophy,Vol. 45. ———. 2011a. “Socratic Introspection and the Abundance of Experience.” In Journal of Consciousness Studies, Special Issue on R. Hurlburt’s and E. Schwitzgebel’s Describing Inner Experience? Proponent Meets Skeptic. ———. 2011b. “Phenomenal Thought.” In Cognitive Phenomenology, edited by T. Bayne and M. Montague. Oxford: Oxford University Press. Smith, D. W. 2005. “Consciousness with Reflexive Content.” In Phenomenology and the Philosophy of Mind, edited by D. W. Smith and A. Thomasson. Oxford: Oxford University Press.

On the Phenomenology of Introspection

167

Tye, M. 2002. Consciousness, Color, and Content. Cambridge, Mass.: MIT Press. Williford, K. 2006. “The Self-Representational Structure of Consciousness.” In SelfRepresentational Approaches to Consciousness, edited by U. Kriegel and K. Williford. Cambridge, Mass.: MIT Press. Zahavi, D. 2005. Subjectivity and Selfhood. Cambridge, Mass.: MIT Press.

This page intentionally left blank

6 The Epistemology of Introspection Ernest Sosa

How if at all do introspective seemings have probative value? If any have it at all, do they all have it equally? We consider which seemings might have the epistemic standing required. If only some do, what distinguishes these?

1. The Place of Introspection A catalogue of the mental, even if only partial, will provide helpful background for our inquiry.1 1

Anscombe once remarked on the importance of a correct philosophy of mind for well-founded ethical theory. This she highlighted in the second sentence of her “Modern Moral Philosophy,” Philosophy 33, No. 124 (1958), according to which “. . . it is not profitable for us at present to do moral philosophy; that should be laid aside at any rate until we have an adequate philosophy of psychology, in which we are conspicuously lacking.” By the end of the article, it emerges that what she means by “philosophy of psychology” is really either “psychology” or “philosophy of mind,” or some combination, since it turns out to be . . . an account of human nature, human action, the type of characteristic a virtue is, and above all of human “flourishing.” Epistemology seems similarly related to philosophy of mind, though surely we must allow the two disciplines to proceed in parallel, especially if each is properly sensitive to the development of the other.

170 Theories of Introspection a. Beliefs are propositional attitudes that come in degrees of confidence. Belief is confidence above a certain threshold.2 b. Judgments are mental acts of affirmation. c. Sensory experiences (“experiences,” for short, in what follows) are sui generis propositional attitudes, in various modalities: the visual, the auditory, etc. Although often noted or expressed, experiences need not be expressible in words, or even capturable in concepts within the subject’s repertoire. d. Seemings are attractions to assent. These come in two main sorts: the prima facie and the ultima facie or resultant. A conflict among seemings can be resolved through conscious deliberation. But its resolution can also unfold subconsciously. Either way, a resultant attitude of attraction (or repulsion or neutrality) might emerge from the conflict of prima facie forces. e. Seemings can have various sources. Viewing a Müller-Lyer pattern attracts one to think the lines incongruent. Careful measuring yields an opposing attraction. If someone trusted says that the lines are congruent, this too can affect the balance of forces, until perhaps the conflict is resolved. In a normal Müller-Lyer case, for example, the balance eventually favors congruence over incongruence. f. One might try to identify seemings in general with degrees of confidence. Those below the threshold of belief would then count as mere seemings, those above the threshold as beliefs. But this cannot be right, since a prima facie seeming can retain a high degree of prima facie plausibility, above any plausible threshold, while defeated by counter-seemings, in which case the subject’s degree of confidence might fall to near-zero, even if the proposition still seems prima facie to be true. More plausibly identified with degrees of confidence are resultant seemings, rather than prima facie seemings. How confident you are of a certain proposition does seem equivalent to its degree of resultant seeming, to how plausible it seems all things considered. A degree of confidence above a certain threshold would then count as more than just a resultant seeming; it would count as outright belief. g. An inclination to believe is a resultant attraction that (i) is positive but (ii) falls below the confidence threshold of belief.3 All such inclinations are therefore seemings, but not all seemings are inclinations. A prima facie seeming can be 2 This in fact is just one conception of belief. There is also belief as a disposition to judge, which is not clearly equivalent to belief as degree of confidence. See my “Value Matters in Epistemology,” Journal of Philosophy (April 2010), for reasons to doubt any such equivalence. In any case, the threshold of confidence for something properly to count as a “belief ” is no doubt context-dependent. 3 It might be thought that one could have a positive resultant credence (of say .6 or so) without being at all inclined to believe accordingly. But I do not see what being “inclined” to believe could

The Epistemology of Introspection

171

swamped, as in the Müller-Lyer example, so that no positive resultant seeming favors the proposition that the lines are incongruent; indeed the resultant seeming is normally the opposite and graduates into confident belief. Nevertheless, most of us retain a prima facie (not resultant) seeming that favors incongruence. This is what we report when we say that one of the lines sure looks longer, even after we know better. h. Finally, we should perhaps distinguish attraction to assent from felt attraction to assent. Compare gravitational pull. We are all attracted by both the moon and the earth as we lie in bed at night, but what occasionally we feel (if only through the effort to overcome it) is the resultant attraction earthward. Nevertheless, the respective attractions are both still there, even if they are not severally felt. Something analogous may be true of intellectual attractions. Psychological research may uncover that a right-handed subject is somewhat attracted to choose an item simply because it is within the reach of his or her right hand, even if this attraction and its source never surface in the agent’s consciousness. Where do introspective attitudes belong in our picture?

2. Deliberating and Pondering This preliminary section further prepares the way for our questions. We begin by considering what it is to ponder a factual question and to deliberate on a practical choice. We ponder or deliberate by weighing reasons as we try to decide. In yes/no cases the overall balance may favor one side sufficiently that one affirms or chooses accordingly. And if the process has been adequate, that decision will be justified. What is it to weigh reasons, to assign a weight on the scales of judgment? We speak of giving a certain weight to a consideration. Switching metaphors, we might equally speak of attraction or repulsion. We give greater weight to one

mean if not a positive credence above.5 (though it may still fall short of the threshold of belief). Well, it could mean just an “attraction” to believe. These two meanings have not been widely recognized or, of course, kept apart. Once we do recognize the difference we have a choice, and we should try to avoid ambiguity. To me it seems more plausible to distinguish mere attraction from inclination in the way I am suggesting. A balance can remain somewhat attracted in the direction of the less weighty side. Gravity still pulls somewhat in that direction. But once a given side tilts below the horizontal it is now inclined (not just attracted but “inclined”) toward the weightier end.

172 Theories of Introspection consideration than to another if and only if it attracts us more. Even if a resultant attraction favors one side, moreover, this may not decide the matter. A resultant attraction may yield a decision only if its intensity rises above a certain threshold. If your weighing—your pondering or deliberation—is conscious, the constitutive attractions then take their place in your consciousness at the time. If a friend you trust tells you something, this attracts you to believe what he says, but if a second friend tells you the opposite, this attracts you to believe the opposite. Here we have conflicting attractions. If one friend is more trusted, his testimony will have more weight; it will attract more. But the resultant attraction may or may not suffice to determine belief. Even if it overcomes the opposing attraction, it may remain only an inclination, without determining outright belief. That situation is of a sort familiar to us all. There are many other instances. Even when you believe in the sincerity of an interlocutor, for example, shifty eyes and fidgety fingers may insinuate doubt. Even when you affirm the result of a calculation, doubt may remain if you have erred in the past. In these cases, various things are made to seem so. Take our first example: emphatic assertion may betoken sincerity, even while shiftiness and fidgeting suggest insincerity. In the second example, careful calculation may inspire confidence, even while your many past mistakes give you pause. In what follows, the expression: ‘C makes it seem that p’ abbreviates ‘Consideration C makes it seem to someone (to someone or other, or to someone in particular) that p’, and the like. Ostensibly, we use the expression to predicate a relation. A certain consideration, when it comes to a subject’s awareness, makes it seem to him that p. If such a subject says: “C makes it seem that p,” an interlocutor might retort: “Not to me it doesn’t.” Moreover, we do sometimes use the expression to signify rather the outcome of a causal process, partly or wholly mental, as when we note that it was someone’s careful calculation that made a certain answer seem so plausible to him. On either use, something can make the affirmative seem right (to someone) on a certain question, even while something else does the same for the negative (to the same subject or to someone else). Beyond the supportive examples already considered, there are many others. Thus, suppose you measure Müller-Lyer lines. This may make it seem to you that they are congruent, even while the visual attraction to think them incongruent still remains, at least while the lines are in view. We can thus see how opposing attractions might coexist concurrently in a single mind. Such attractions will play a large role in what follows, under the title of “seemings.” What constitutes being attracted by a consideration C to assent to the proposition that p? Answer: This may turn out to be just a belief by the subject to the effect that C

The Epistemology of Introspection

173

makes it probable that p. Or a presupposition, or perhaps a “commitment,” and these might be conscious or subconscious. For now we need not make a choice, since we all ponder and deliberate often enough. These mental phenomena involve the attractions that we posit, however these are to be analyzed. Soon we shall find reason to reject the present answer, in any case, as it becomes implausible that, whenever one is attracted to assent, some consideration C must exert its attraction in the way required for the present proposal to succeed. In particular, when it is just the sheer understanding of a propositional content that accounts for one’s attraction, the proposal cannot be retained in its entirety, though much of it will remain even so.

3. Introspective Attitudes An intellectual seeming can be an attraction to assent. (Perhaps it needs to be a felt attraction; here we table that secondary issue.) Such seemings can be introspective. They are involved in perception, too, which admits experiences as well. And here we find a striking difference between introspection and perception.4 Take the look of a bare chessboard arm’s-length away in bright light. The array that gives content to your experience has 64 alternating black or white squares. Yet in your first encounter with a chessboard, the proposition that you face an array with 64 squares may hold no attraction. In contrast, a chess player would immediately assent.5 So, the experience must be distinguished from the seeming. In an expert chess player the experience will prompt the seeming, and even the corresponding assent, but in the chess-ignorant it will not. Nor does this distinction require a sense-datum view. Sensory experience might be a propositional attitude, even when inexpressible in words or symbols with any currency. The propositional attitude of one’s experiencing as if p is hence distinct from the propositional attitude of it seeming to one as if p (of one’s being intellectually attracted to assent accordingly). The two attitudes are constitutively distinct. How similar are they epistemically?

4

With the reader’s indulgence we postpone clarification of this until section 6. Of course background knowledge contributes mightily to the expert’s attraction, but this conflicts not at all with the validity of the distinction (indeed it underwrites the distinction, since the experience receives no such contribution from the background). Alternatively, we might consider whether the difference is a matter of degree; our argumentative purposes would be served about as well either way, even if our case would require some reformulation. 5

174 Theories of Introspection Consider the visual sensory experience shared by a chess master and a beginner, when they both view a chessboard in good light. This experience is as passive as is an experience of pain. One’s visual experience of a chessboard cannot be changed at will nor much affected by direct choice. That is how it is for the visual sensory experience of 64 alternating black or white squares. Also passive is the seeming prompted in a chess master. Neither state is modifiable at will, by direct choice. Once he considers the question, the chess master has no choice but to suffer the prompted attraction or seeming. Nevertheless, there is a difference. The seeming is rationally based; the experience is not. The seeming manifests the subject’s rational agency; the experience does not, any more than does a pain. Someone with a headache is not in pain based on a reason. By contrast, his assent is attracted overwhelmingly by a reason. Is the pain a reason why but not a reason for which?6 I doubt it; when I am aware that my headache is starting, I have a reason for so believing, namely, that it is indeed starting. I am responding to my experience with an appropriate belief, and there is a reason within my consciousness why I so believe, and for which I so believe. That stands in contrast to the neurological reasons why I so believe without believing for those reasons. It is surely relevant that in one case I can make my reason explicit without special empirical inquiry, but not in the other case, that of the neurological causes. I can avow my reason for thinking that I am in pain by saying that I so believe because I am in pain. Not so in the other case. (Pains have only one foot in the space of reasons: they constitute reasons but are not themselves based on reasons.) Although introspective seemings (and beliefs) are not based on corresponding experiences, they are not baseless. They are based on mental states, either on mental states without propositional content (such as anxiety, perhaps) or on mental states with propositional contents (sensory experiences). For example, it now quite properly seems to me that I am having a visual experience as if I see a hand before my eyes. This seeming has a rational basis provided by my experience. Contentless mental states, by contrast, have no more rational basis than do pains or visual sensory experiences themselves. Sensory experiences are never based on motivating reasons, unlike intellectual seemings. Experiences thus differ from seemings not only constitutively but also functionally. The epistemological bearing of this contrast remains to be considered below.

6 This is a distinction between causal or explanatory reasons and motivational reasons more specifically.

The Epistemology of Introspection

175

4. What Is a Justified Introspective Seeming? A Latitudinarian Answer Ultimately we wish to understand knowledgeable introspection. What might render an introspective seeming epistemically justified, giving it probative standing? A latitudinarian answer gives all introspective seemings equally positive standing: Any propositional content that appears introspectively to you (that thus seems to you correct) is thereby epistemically justified. Someone could be epistemically maladroit, however, inappropriately incautious and overconfident, in which case he might be strongly but incorrectly attracted to assent to his being in a certain introspectable state, say that of being calm and unflappable when in fact he was nervous and scared. This is just one of many examples of failure to discern how it is in one’s own mind. We can fall short by being epistemically unjustified through incompetent, excessive attraction to assent. Perhaps then introspective seeming is epistemically justified only when it coincides in a certain way with a concurrent mental state. This would happen when the seeming is a seeming that p, while the mental state is either the fact that p or the truthmaker for that fact. Here is a revised principle: Any propositional content that appears introspectively to you (that thus seems to you correct) is thereby epistemically justified provided its propositional content corresponds to a mental state that you undergo at that time and that attracts your assent to that content (where this mental state is either the fact constituted by the truth of that propositional content or else is the truthmaker for that fact). Must any such introspective seeming be noninferentially justified? Would every seeming acquire such epistemic standing and probative value just through the subject’s attraction to assent to its content? This question of latitudinarianism is left open by the proposed account. Let us next take it up.

5. Justified Introspective Seemings: Are They Probative? Toward a Competence-Based Account Intellectual seemings are themselves epistemically evaluable. A consideration might attract too much or too little; introspective attractions are no exception. Someone might just be very poor at assessing his own state of mind in various

176 Theories of Introspection respects.7 We are not shocked by the revelation that people confabulate as to their motives or that wishful thinking affects our beliefs on all sorts of subject matter, including the introspectable. Such research in social psychology is of course welcome, even when it just establishes with scientific standing what we already suspected commonsensically, especially if it reveals how broad is the sway of such epistemically dubious mechanisms. What feature must an introspective seeming have, beyond having a true content, if it is to be epistemically justified? Here is a proposed answer. An introspective seeming is epistemically justified when the subject is attracted to assent just by undergoing the mental state self-attributed through the seeming, provided the attraction is owed to a competence, i.e., to an ability that reliably enough discerns the true from the false. Such a competent ability can be manifest not only in the attraction to which it gives rise but also in the attraction’s being correct, by having a true content. This picture involves degrees, including (a) the degree of attraction and (b) the degree of reliability of the competence. In order to be epistemically proper, a seeming must involve some minimum degree of attraction and some minimum degree of reliability. Two thresholds would thus need to be set. Our account hence abstracts from specifics: when one properly introspects, this is because there are respective thresholds of attraction and reliability, such that one is attracted above the first threshold, due to a competence above the second. What sets these thresholds? I do not know. Perhaps they are set contextually. But that seems a linguistic matter concerning the truth conditions for various utterances. And it is not clear how that is related to my question, asked in my present relevant context of writing a chapter on epistemology, or in yours when you hear or read this chapter. Once we set the relevant context, it might be thought, that sets the appropriate thresholds. So, in my context, the thresholds are set by the relevant determiners of the truth conditions for my utterance that someone “knows” or “is justified,” and so on. (How, exactly, are those thresholds set? Why does our language set them as it does for, say, my particular context as author, or yours as auditor or reader, when we now communicate about these philosophical questions? These are dark and cloudy matters.) 7 This is one sort of counterexample. The Speckled Hen provides another, or so I have argued elsewhere (see L. BonJour and E. Sosa, Epistemic Justification (Oxford: Blackwell, 2003), esp. ch. 7, “Does Knowledge Have Foundations?”). The cardinality of one’s experience (eleven black dots) might perfectly correspond to one’s attraction to assent (to its being eleven dots) even if the seeming is not at all epistemically justified or probative, simply because it goes far beyond one’s ability to subitize (to tell cardinality at a glance, without counting).

The Epistemology of Introspection

177

Plausibly, the quality of one’s seeming will vary with the relevant degrees. A seeming is epistemically better as a function of the reliability of the relevant competence, I would say, but also as a function of the degree of attraction. What is to be said for or against this option?

6. Introspection and Perception It is instructive to compare introspection more closely with perception. Visual perceptual knowledge involves a visual experience that gives rise to a visual seeming. Normally when a visual seeming corresponds to a visual experience, it is thereby justified, and yields in turn a corresponding belief, itself thereby justified. However, such perception is not a good model for successful introspection. Unlike visual experiences, seemings are not sources of our attraction to assent. They are rather the attractions themselves. An attraction is introspective when one is attracted in a way that involves no rational basis other than the subject’s conscious understanding of the proposition whose truth is intuited, with its specific content, except for the truth of the content, or for the mental state that functions as truthmaker for that truth. What is distinctively involved in introspective justification? What distinguishes it from perceptual, sensory justification? Here again is a proposed point of difference: What distinguishes introspective justification is that the truth of the content of the introspective seeming (or the truthmaker for that truth) exerts its attraction unaided. Introspective seemings are reason-based in a way that does not go beyond conscious grasp of the specific propositional content and the truth of that content. Introspective justification involves nothing relevantly like the visual sensory experience that in normal human perceptual knowledge accompanies one’s conscious entertaining of a visual propositional content. Introspective justification involves no pre-seeming experience that can prompt and thereby justify a corresponding seeming.

7. Parallel Epistemic Roles: Perceptual Regress Stoppers and Intuitive Regress Stoppers Introspective seeming is intellectual attraction that derives its existence, and, when proper, also its propriety, from a distinctive source, one beyond any combination of intuition, perception, testimony, memory, and inference. (The formulation here takes care to allow overdetermination of justification.)

178 Theories of Introspection Perhaps the attraction of a proposition is introspective when it concerns the thinker’s concurrent mental states and attracts his assent through its sheer truth. It may seem that foundational justification requires a basis like the sensory experience that figures in perceptual knowledge and justification. Foundational justification may seem to require a regress stopping conscious state that provides justification without requiring it in turn, a state undergone passively. Such a state must be one that cannot be entered or departed through voluntary choice. It must even lie beyond the subject’s rational agency, so that it is not entered for a motivating reason. Sensory states are of course like that. They provide justification without requiring it in turn and are undergone quite passively. You enter or depart them neither by choice nor based on reasons; they are beyond your rational agency. Even when a sensory state provides justification, we would like to know just how this mode of justification does its work. Does such a state always justify its corresponding seeming or belief ? The Speckled Hen Problem shows that this cannot be right, or so, again, I have argued elsewhere. How then does our supposed sensory state provide justification when it does so? What is distinctive of such cases? An alternative account requires competence. Take those who justifiably believe that they face a collection of three items, based immediately on a corresponding visual experience. These are distinguished by their ability to subitize (reliably) that many items. Your belief that you see eleven items in your visual field has no comparable justification, absent the ability to subitize reliably much beyond the normal four or so items. Without that ability there can be no such justification, not even when the seen items in your visual field do number eleven.8 The exercise of competence is necessary for an epistemic status of interest to us, even should that status fall short of “knowledge,” as we ordinarily and literally conceive of it. When we say that apt belief is “animal knowledge,” is this to speak metaphorically? Perhaps aptness of belief is necessary but not sufficient for knowledge. Perhaps it must be supplemented by a second-order, reflective dimension, 8

I here set aside difficult and important issues taken up elsewhere. Suppose what accounts for your belief is just that you have been hypnotized, or temporarily wired, so that you will respond to eleven items in your experience with a corresponding seeming. In that case, your belief is related to the fact believed in a way that is more than just a sheer accident. One might hence attribute a kind of “knowledge” to you, if only metaphorically. But your state falls short of a fuller state of human knowledge, and justification. Through your incautious overconfidence, you think you can subitize right up to eleven-membered arrays and even beyond. However, unbeknownst to you that is just rash bravado, except for the case of eleven, where you have been outfitted with a real competence. Plausibly, you still do not “know” even so, which highlights the importance of the second-order perspective, and the closer approach to literal “knowledge” that it enables. Or so I argue in “How Competence Matters in Epistemology,” in Philosophical Perspectives (2010), and in “Knowledge and Competence,” forthcoming.

The Epistemology of Introspection

179

before it can reach the fuller status. In any case, the exercise of competence is necessary even for that preliminary status. Plausibly, it is also sufficient, moreover, even in the absence of any regress-stopping state, whether of sensory experience or any other. There is no apparent reason to require that our basic noninferential mechanisms of proper belief fixation must always operate through an ulterior mental basis. This is not only implausible abstractly. It is implausible also when we consider basic a priori beliefs with elementary content, such as those in arithmetic, geometry, and logic. No ulterior mental basis seems discernible for such beliefs. As for contingent beliefs, consider also our competences to discern cardinality, or shape, or color. Some such competences are reason-based: the subject acquires his belief or seeming based on some guiding ulterior mental state. Even perceptual competences, however, are not plausibly required to operate (always, necessarily) through the guidance of ulterior states. Take the phenomenon of blindsight. No personal-level experience guides blindsight beliefs about the orientation of certain lines. Yet the subject’s seemings and beliefs about the faced lines are still formed with above-random reliability. Perhaps in the actual world visual experiences do after all guide such blindsight beliefs. Even so, in a nearby world they derive from purely subpersonal mechanisms with much higher reliability. How are these beliefs to be assessed epistemically? Suppose some are as “justified” as are simple arithmetic and geometric beliefs, and thereby constitute knowledge. If so, then perceptual knowledge of one’s environment requires no foundational regress stoppers. What really matters is rather that the belief manifests a relevant competence. This is so whether or not it is guided by regress-stopping states beyond justification. Even if there are such states, they must still guide the subject through competence. Absent such basis states, moreover, as with our enhanced blindsight, belief still needs the guidance of competence if it is to be justified and knowledgeable. But this competence now operates subpersonally, not through rational basing. If sensory experiences have propositional content, are they not epistemically evaluable after all, in terms of their accuracy, adroitness, and aptness? Yes, so they are, granted. However, there is still an important difference. Unlike judgments and beliefs, unlike seemings, experiences are not conceptually constituted. Take a squiggle that to your eyes is formless. Someone brought up in a culture for which that shape has great significance may well have a concept, a phenomenal, discriminative concept of that squiggle. This may be so even if the religion prohibits any name or predicate—any symbol whatever—corresponding to it, while forbidding also its artificial reproduction. Only through indexicals and demonstratives are children imbued with proper respect for that squiggly pattern. They do eventually acquire a visual concept of that shape, however, so that they can properly venerate its instances, etc., so that the shape can acquire and sustain its proper place in their

180 Theories of Introspection form of life, through shared intentions and beliefs. If two identical twins are separated at birth and one is brought up in that culture and one not, the two can still have identical visual experiences when they see the same squiggle with proper attention and illumination. Still, one of them will have a concept of the form in question, the other not. So, the visual experience itself is not conceptually constituted, by being the exercise of a concept. It is a more basic mental state derived from the brute operation of our eyes, nervous system, etc.9 In contrast, attractions to assent, intellectual seemings, are conceptually constituted. You cannot be attracted to assent to a propositional content that you do not understand. This is not to say that you cannot in a way endorse an experience as veridical, even while deprived of concepts for its constitutive properties. You can still endorse that experience, if only by picking it out demonstratively. But you would then do so indirectly. Analogously, you can point to a sentence and endorse it as true without understanding it. Again, there is a way in which experiences are epistemically evaluable: first, as accurate or veridical; second, as adroit or competent; and even, third, as apt, as veridical because competent, in a way that manifests the subject’s visual competence. Only with seemings, however, do we ascend to the space of the conceptual, to rational deployment of concepts. Only with seemings do we enter the realm of the rational, the space of conceptual reasons.

8. Introspection and Consciousness Mental states need not be conscious. Clearly, then, we can enjoy a kind of introspective blindsight. This would be a competence reliably to discern one’s own subconscious mental states, including even dispositional states. One might try to stipulatively restrict the term ‘introspection’ to the discerning (would-be or actual) of conscious mental states of one’s own. But this seems ill motivated. One can of course just explicitly distinguish introspection of a subject’s conscious states from introspection of his unconscious states. But these are both cases in which the subject discerns his own contingent concurrent mental states directly enough, with relevant independence from inference. So they both count as introspection in a natural sense. Along with blindsight perception, accordingly, such “blindsight” introspection is then knowledgeable if and when it manifests the subject’s reliable epistemic competence. 9

Its presence does enable subjects to demonstratively distinguish items so shaped from many otherwise shaped and viewed concurrently, even if these subjects lack the recognitional abilities required for mastery of a concept of the shape.

The Epistemology of Introspection

181

9. Epistemic Competences Can Be of Use Even Absent Full Understanding of Their Specific Nature We can appeal to such competences in epistemology even with limited understanding of their modus operandi.10 This in fact applies not only to justified introspection but also to rational intuition and even to perception and memory. Up to a point, people could know how they knew things even before we gained our fuller understanding of how perception actually works, in its various modalities. Our knowledge through various perceptual sources is enhanced with improved understanding of their nature and operation, but this boost of meta-sophistication is not essentially required. It is not required for animal knowledge through such competences. It is not even required for some measure of reflective knowledge, despite our limited understanding of the operative competences.

APPENDIX. Knowledge First and Factive Perceptual States In a knowledge-first theory of knowledge, factive perceptual states are considered forms of knowledge. Seeing that p, for example, is thought to be a special case of knowing that p. It is to know that p visually. This gives rise to the view of knowledge as the most general factive mental state. Intuiting that p might then take its place as another such factive state. To intuit that p would then be to know that p intuitively.11 Our account enables a different view of factive mental states. We can see the potential for this alternative through an example involving vision. Suppose someone sees a red surface in good light. The surface presents to him a visual experience as of a red surface, and this in turn attracts him to take the surface to be red. However, the subject suspects the light, perhaps for good (though misleading) reason. Accordingly, he does not take the surface to be red, despite his strong, vision-derived attraction to do so. Some are willing to say that our subject sees that the surface is red, although he does not believe and hence does not know that it is red.

10

This is defended in my “Minimal Intuition,” in Rethinking Intuition:The Psychology of Intuition and Its Role in Philosophical Inquiry, edited by Michael R. DePaul and William Ramsey (Lanham, Md.: Rowman and Littlefield, 1999). 11 Timothy Williamson, Knowledge and Its Limits (New York: Oxford University Press, 2002).

182 Theories of Introspection Our framework has room for such a state of factive seeing that is not knowledge. Such factive seeing would be visual seeming that gets it right in a way that manifests the subject’s visual competence. Factive seeing is thus apt visual seeming, through the proper operation of the subject’s visual system. This would involve the subject’s deriving a visual experience as of a red surface through the use of his eyes in good light, involving rods and cones, the optic nerve, etc. The subject would be attracted on that basis to consider the seen surface red. This attraction, this seeming that the surface is red, does then get it right, and its doing so manifests the subject’s visual competence. It is hence an apt visual seeming, which on our present account is factively to see that the surface is red.12 On this model we can also understand a corresponding concept of factive introspection. An introspective seeming on our view is a kind of direct (noninferential) attraction to assent to

, a seeming that p, where

appropriately concerns a mental state hosted by the subject. Such a seeming can get it right. If it does so in a way that manifests the subject’s introspective competence, then it is an apt introspective seeming, which makes it factive. Such factive introspective seemings are not forms of knowledge. Factive seemings are rather in general pre-knowledge states on the basis of which subjects can acquire corresponding knowledge. On this view, accordingly, knowledge is not the most general factive mental state. Rather, apt seeming is a more general factive mental state than knowledge. Every case of knowledge will be a case of corresponding apt seeming, but not every case of apt seeming will be a case of corresponding knowledge.

12

We could of course distinguish further between core factive seeing, or apt experiencing, and full factive seeing, or apt seeming (apt attraction to assent). Factive intuiting reduces to apt understanding-based seeming, since in that case there is no core experience that mediates between the fact known and the subject’s intuitive attraction to assent.

7 Knowing What I See Alex Byrne

If I descry a hawk, I find the hawk but I do not find my seeing of the hawk. My seeing of the hawk seems to be a queerly transparent sort of process, transparent in that while a hawk is detected, nothing else is detected answering to the verb in ‘see a hawk’. —Ryle, The Concept of Mind

1. Introduction By using my eyes, I can come to know that there is a hawk perching on the fence post. Cognitive science has made enormous progress in understanding how we have this sort of perceptual knowledge. Any textbook on perception will go into detail about the receptors in the eye, the detection of low-level features such as edges, the recovery of 3D shape, and theories of object recognition. Much is controversial, and much is unknown, but the broad shape of a satisfying account, together with some of the intricate small parts, is clear enough. When I am in a position to know, by using my eyes, that there is a hawk on the fence post, I am usually in a position to know something else, namely that I see a hawk. This second item of knowledge is of course not entailed by the first: idealism aside, the hawk has no essential connection to me or my perceptual state. Equally obviously, this second item of knowledge is not evidentially

184 Theories of Introspection probable given the first: the mere fact that there is a hawk on the fence post is hardly good evidence that it is seen, let alone that I am the one who sees it. Which is to say that an account of how I know that there is a hawk on the fence post by vision cannot easily be converted into an account of how I know I see a hawk. Hence, the question this chapter will attempt to answer: How do I know that I see a hawk?1 Textbooks are of no help, and the issue has received very little discussion by philosophers. The obsession of contemporary epistemology has been my knowledge of the hawk, not my knowledge that I see a hawk. The explanation of this curious state of affairs is an interesting topic in its own right; it will not be examined here, however.

2. Dretske on Zombies One philosopher who has addressed our question is Dretske, most extensively in “How Do You Know You Are Not a Zombie?” (2003).2 The eponymous question of Dretske’s paper covers perception and awareness in general; it has our question about seeing as a special case. Dretske’s discussion highlights a serious problem with one ostensibly attractive answer to our question, so this is a good place to start. Dretske’s mention of “zombies” might mislead. In more-or-less standard usage, “zombies” are creatures who are physically exactly like awake and alert human beings, but who are not “phenomenally conscious”—there is “nothing it is like” to be a zombie. Zombies are frequently presumed to have a typical package of intentional mental states. So zombies believe that it is raining, and see hawks, although of course their perceptual states are devoid of any “qualia.” For those who think that this conventional sort of zombie could have existed, the question “How do you know you are not a zombie?” can seem pressing. After all, zombies are (arguably) firmly convinced that they are not zombies—just like us.3 Importantly, Dretskean zombies are not the standard sort, and epistemological issues about qualia are only of peripheral relevance to Dretske’s concerns. In Dretske’s usage, “zombies [are] human-like creatures who are not conscious and, 1 ‘I see a hawk’ is to be read in the everyday sense in which it entails that there exists a hawk to be seen. If there is a sense of ‘see’ in which I can be truly said to see a phoenix (as is claimed in Anscombe 1965), it does not occur in this chapter. 2 See also Shoemaker 1963, especially 83–84. 3 For extensive discussion, see Chalmers 1996, ch. 5. See also note 18 below.

Knowing What I See

185

therefore, not conscious of anything” (2003, 9 n. 1). A Dretskean zombie is simply a superficial human look-alike who behaves in humanlike ways and who lacks intentional states; in particular, a Dretskean zombie sees nothing. One day the Sony Corporation will produce mindless robots to help around the house, so sophisticated that the casual observer will take them to be normal humans—Dretskean zombies are rather like that. The possibility of standard zombies is controversial; in contrast, only a hard-line behaviorist would deny that Dretskean zombies could have existed.4

2.1. Dretske’s Statement of the Problem Dretske writes: In normal (i.e. veridical) perception, then, the objects you are aware of are objective, mind-independent objects. They exist whether or not you experience them. . . . Everything you are aware of would be the same if you were a zombie. In having perceptual experience, then, nothing distinguishes your world, the world you experience, from a zombie’s. This being so, what is it about this world that tells you that, unlike a zombie, you experience it? What is it that you are aware of that indicates that you are aware of it? (2003, 1, note omitted) Dretske’s point—applied to our running example—was alluded to in this chapter’s second paragraph. The hawk has no special connection to visual experience— provided it does not notice my presence, the bird does not “indicate that I am aware of it.” Hawks can and frequently do happily perch unseen on fence posts; put more generally, the world as revealed by vision does not have vision in it. Thus, the presence of the hawk does not favor the hypothesis that I see it over the “skeptical hypothesis” that I am a (Dretskean) zombie, and hence do not see it. The evidence (facts) provided by vision would be exactly the same even if I were a zombie.5

4 Dretske briefly alludes to “some readers who doubt that [Dretskean zombies] are possible” (2003, 10 n. 1), so to be on the safe side the explanation of Dretskean zombies in the text should be viewed as a friendly elaboration or amendment. 5 Cf. Wittgenstein [1921] 2001, Tractatus 5.633, “nothing in the visual field allows you to infer that it is seen by an eye” (referred to in Dretske’s opening paragraph).Wittgenstein’s concern is the self (in fact the “metaphysical subject”), not vision. And the self is another problem: the world as revealed by vision also does not have me in it. (The fact that I can usually see my nose is not the key to this difficulty.) The point is somewhat obscured by concentrating on a Dretskean zombie scenario, which does have me in it (or at any rate a Dretskean zombie who resembles me).

186 Theories of Introspection One (vaguely stated) answer to our question is that I know that I see a hawk merely by attending to the scene before my eyes, and in particular to the hawk. Dretske was formerly sympathetic to this idea; with a nod to the epigraph from Ryle, call it the transparency proposal.6 Despite the proposal’s attractions (of which more shortly), the considerations just rehearsed seem decisively to refute it. For the facts revealed by attending to the scene before my eyes are at best very weak evidence that I see a hawk. Call this the evidential objection to the transparency proposal. This chapter will elaborate and defend the transparency proposal. Let us begin by confirming that the evidential objection has no quick solution.

2.2. Supplementary Environmental Evidence According to the transparency proposal, I know that I see a hawk merely by attending to the hawk. And the evidential objection is that the evidence about the hawk gathered by this procedure is not good evidence that I see it. Notice, though, that vision also gives me information concerning the spatial relation between the hawk and myself, namely that I am facing the hawk.7 Now this does not help much on its own—I can easily face the hawk and not see it for the simple reason that I might have my eyes closed. But what if we add in evidence (provided by proprioception or kinesthesia) about the disposition of my eyelids, and other relevant bodily parts? Can’t I then know that I see a hawk? Admittedly, vision is not then the only source of my evidence, but this revision preserves the basic idea that one knows what one sees by attending to one’s environment, broadly construed. Dretske in effect considers the revision and dismisses it in a few sentences:“Zombies, after all, have bodies too. . . . A zombie’s arms and legs, just like ours, occupy positions. Their muscles get fatigued . . .” (2003, 2). In the skeptical scenario, the zombie’s body also faces the hawk, the zombie’s eyes are open, etc. This additional evidence does not discriminate, then, between the scenario in which I see a hawk and the scenario in which I am a (Dretskean) zombie.

Likewise, a video taken by a camera does not (usually) have the camera in it. Of course, one could work out the location and motion of the camera from a video containing enough perspectival information (cf. Gibson 1979, ch. 7), but the information the camera records about the scene does not itself imply that it was recorded. Although the emphasis will be on vision, the problem about the self will also be addressed in what follows, albeit inexplicitly. 6 For Dretske’s earlier view, see Dretske 1995, ch. 2. The now numerous discussions of “transparency” in the literature are of course not sourced to Ryle [1949] 1963, but to Moore 1903. 7 That vision allows me to know something about myself is itself puzzling (see note 5). But let us grant it for the moment.

Knowing What I See

187

But even by Dretske’s lights this is too quick. His question is: “What is it about this world [of “objective, mind-independent objects”] that tells you that, unlike a zombie, you experience it?” And his dismissal of the present proposal gives the impression that an answer needs to be absolutely skeptic-proof, displaying a body of evidence gained through perception that entails that I see a hawk. In fact, Dretske is not setting the bar so high: the challenge he poses is to explain how I know that I see a hawk by observing the environment (including, perhaps, my body). And the suggestion about proprioception and my relation to the hawk is, in effect, the idea that I find out that I see a hawk on the basis of the sort of evidence that would support the claim that someone else sees a hawk. I can come to know that someone else sees a hawk by noting that there is a suitably placed and salient hawk, that the person’s eyes are open and converge on the hawk, and the like.8 Or so we may assume—skepticism about other minds is not the issue. Hence, the problem with the present suggestion is not that it fails to supply a way of knowing that I see a hawk, and so that I am not a zombie. The problem, rather, is that I plainly do not rely on supplementary proprioceptive evidence in order to know that I see a hawk. Suppose that unbeknownst to me, I am suddenly stuck by a bizarre medical condition that renders my eyelids transparent. I turn to face the hawk and close my eyes. I know these latter two facts. Since I know that my eyes are closed, on the present suggestion I lack sufficient evidence to know that I see a hawk. But if Ryle asks me “Do you see a hawk?” I would hardly claim that I do not, or that the question was a difficult one to answer! I know that I see a hawk, just as I do in the normal case.9

3. Evans on Knowledge of Perceptual Experiences The evidential objection would be of little interest if the transparency proposal lacked plausibility anyway. However, there is much to be said in its favor. As Ryle puts it, my seeing of the hawk “seems to be a queerly transparent sort of 8 Uncontroversially, I might sometimes have good evidence from vision that I see a hawk (perhaps I see myself in a mirror, staring at a hawk). But this is not a typical case. 9 Since I know that my eyes are closed, I do have some evidence that I do not see anything (and so must be hallucinating a hawk). But this contrary evidence surely does not prevent me from knowing that I see a hawk. The point can be reinforced by considering other modalities, which one would expect to have the same basic epistemology as vision. Suppose I hear the distinctive scream of a red-tailed hawk and cannot identify the direction of the sound. I can know that I hear the scream (or the hawk) without checking that my ears are not blocked, or gathering further evidence about the location of the hawk and the orientation of my ears.

188 Theories of Introspection process.”10 There is the hawk, sitting on the fence post. There is Gilbert Ryle, out for a stroll, pausing to descry the hawk. Here am I, looking at Ryle and the hawk. To me, Ryle’s seeing of the hawk is a perceptually manifest fact, as is the fact that the hawk is on the fence post. My own seeing of the hawk, on the other hand, is quite a different matter. I see Ryle and note that his gaze is hawkwards; I do not see myself, or my eyes. Moreover, it does not ring true to say that I discover that I see the hawk by some special introspective sense. There is no switch in attention—say to myself or to a “visual experience”—when Ryle asks me “Do you see the hawk?” I answer by attending to the hawk. (Indeed, if I attend to something else, I might well give the wrong answer.11) The transparency proposal can be extracted from Evans’s influential but brief discussion of the “self-ascription of perceptual experiences” (1982, 226): [A] subject can gain knowledge of his internal informational states [his “perceptual experiences”] in a very simple way: by re-using precisely those skills of conceptualization that he uses to make judgements about the world. Here is how he can do it. He goes through exactly the same procedure as he would go through if he were trying to make a judgement about how it is at this place now . . . he may prefix this result with the operator ‘It seems to me as though . . .’ (ibid., 227–228) Here Evans is concerned with knowledge of how things perceptually appear. But the point is evidently supposed to apply to knowledge of what one sees. The subject—Evans might have added—may, after looking at the scene before his eyes, prefix a phrase encapsulating the result (‘a hawk’) with ‘I see’. Although the quotation has the subject attaching a sentential operator to a sentence, presumably Evans did not mean to tie knowledge of one’s perceptual states to language. Recast in nonlinguistic terms and restricted to the case of seeing an object, the procedure suggested by the quotation is that one can come to know that one sees an object by an inference whose sole premise concerns one’s (typically non-mental) environment,“how it is at this place now,” as Evans puts it. (In fact, this is not Evans’s view. This will become clear later, in section 8, when the elision in the quoted passage from Evans is filled in.) If we remain similarly coy for the moment about the exact nature of the premise, this inference can be set out as follows: 10 Unfortunately, Ryle then goes on to claim that “the mystery dissolves when we realize that ‘see’, ‘descry’, and ‘find’ are not process words, experience words, or activity words. . . . The reason why I cannot catch myself seeing . . . is that [this verb is] of the wrong type to complete the phrase ‘catch myself . . . ’ ([1949] 1963, 285). Since the mystery can be stated without falsely assuming that ‘see’ is a “task verb” (Ryle’s phrase) like ‘run’ and ‘aim’, Ryle’s proposed solvent does not work. 11 With the defensible assumption that one may see an object without attending to it.

Knowing What I See

189

p: It is thus-and-so at this place now c: I see a hawk With this more explicit statement of the transparency proposal in hand, Dretske’s evidential objection can be put as follows: this inference is not knowledge-conducive because p is not good evidence for c.

3.1. Epistemic Rules For the discussion to come, some terminology will be useful. Let us say that an epistemic rule is a conditional of the form, ‘If conditions C obtain, believe that p’, for example: woodpecker: If x is a bird with a red head, believe that x is a woodpecker. And let us stipulate, not unnaturally, that one follows this rule on a particular occasion iff one believes that x is a woodpecker because one recognizes that x is a redheaded bird, where the ‘because’ marks the kind of reason-giving causal connection that is discussed under the rubric of ‘the basing relation’.12 In general, then, S follows the rule ‘If conditions C obtain, believe that p’ on a particular occasion iff on that occasion: 1. 2. 3. 4.

S believes that p because she recognizes that conditions C obtain; which implies: S recognizes (hence knows) that conditions C obtain; conditions C obtain; S believes that p.

Following woodpecker (in certain circumstances that can be left uninvestigated) tends to produce knowledge, and hence is a good rule. Following dodo, ‘If x is a quacking bird, believe that x is a dodo’, produces beliefs that are not knowledge, and hence is a bad rule. The Evans-inspired transparency proposal and Dretske’s evidential objection can be put in terms of this apparatus of epistemic rules. On the transparency proposal, I come to know that I see a hawk by following this rule: hawk: If it is thus-and-so at this place now, believe that you see a hawk.13 And Dretske’s evidential objection is that if ‘it is thus-and-so’ is spelled out as intended, as concerning the hawk before my eyes, the rule must be a bad one, because the antecedent is very poor evidence that I see a hawk. 12

In the terminology of Byrne 2005, 94, because of the presence of the schematic letter ‘x’, woodpecker is a schematic rule; one follows a schematic rule iff one follows an instance of it. 13 ‘You’ refers to the rule-follower; ‘now’ and ‘this place’ refer to the time and place the rule is followed.

190 Theories of Introspection In fact, Dretske’s objection can arguably be overcome, but only if another objection is dispatched first. Explaining this is the burden of the next two sections.

4. Knowing What One Knows and the Evidential Objection Often one knows what one knows. I know that I know that there is a hawk on the fence post, for example. How do I know that? Adapting a famous passage from Evans on the epistemology of belief: if someone asks me, “Do you know that there is a hawk on the fence post?” I must attend, in answering him, to precisely the same outward phenomena, as I would attend to if I were answering the question “Is there a hawk on the fence post?”14 This suggests that I know that I know that there is a hawk on the fence post by following the rule: know: If p, believe that you know that p. Whether or not this suggestion is correct, it should not be in dispute that we at least have the capacity to follow know: often we know that p, and are capable of performing elementary inferences. Now Dretske’s evidential objection arises for know. The fact that there is a hawk on the fence post is hardly good evidence that I know that there is a hawk on the fence post. Hence, if I follow know and infer that I know that there is hawk on the fence post from the premise that there is hawk on the fence post, then my conclusion is based on inadequate evidence, and so is not known. Notice, though, that know has the interesting property that, necessarily, if one follows it, one’s resulting belief about one’s state of knowledge is true. This is because one can only follow know if one knows that p. know, as we can put it, is self-verifying. In this respect, know is unlike typical good rules that we follow in ordinary life, such as woodpecker. woodpecker is not self-verifying because one may know, of a certain non-woodpecker (a red-crested cardinal, for instance), that it is a red-headed bird. That know is self-verifying is enough to blunt the force of the evidential objection. Usually, if one reasons from inadequate evidence, then one’s conclusions will

14 For the passage (about “a third world war”), see Evans 1982, 225. The passage overreaches slightly: Evans says that I can answer “the question whether I believe that p” (my emphasis) by this procedure, which is incorrect in cases where I have no opinion either way. The point also applies to knowledge: if I am clueless about the location of nearby hawks, Evans’s procedure does not apply.Yet I can easily know that I do not know that there is a hawk on the fence post.

Knowing What I See

191

be false. And if they are true, that will be by accident, and hence they will not amount to knowledge. But the beliefs produced by following know are true, and nonaccidentally so.15 Admittedly, this is not sufficient for these beliefs to amount to knowledge, but why think they do not? If in these special cases reasoning from inadequate evidence results in beliefs with one of the characteristic signatures of knowledge, then the status of the evidential objection is moot. Return to Dretske’s evidential objection to the goodness of: hawk: If it is thus-and-so at this place now, believe that you see a hawk. The upshot of the comparison with know is that if hawk, like know, yields nonaccidentally true beliefs, then the evidential objection is muted, if not entirely silenced. But does following hawk lead to nonaccidentally true beliefs? To answer that question the template ‘thus-and-so at this place now’ needs to be filled in. And as soon as we try to do that, another—potentially more serious—objection is apparent.

5. The Amodal Problem A first thought is to fill in the ‘thus-and-so’ along these lines: hawk†: If there is a hawk over there, believe that you see a hawk. On second thought, hawk† is not a good rule, even waiving Dretske’s evidential objection. Suppose I follow hawk†, and so know that there is a hawk there. Since there are numerous ways of knowing that there is a hawk there that do not involve currently perceiving the hawk, let alone seeing it, the probability that I see a hawk, given that I know that there is a hawk there, is low. To conclude that I see a hawk is to take a stab in the dark. Suppose we try inserting the subject into the antecedent: hawk†: If there is a hawk right in front of you, believe that you see a hawk. This sort of maneuver certainly helps increase the probability that I see a hawk, conditional on my knowing the antecedent. But again, there are many other nonvisual ways in which I might know the antecedent. If following a rule like hawk† were my chief strategy for finding out that I see a hawk, then I would be prone to all sorts of errors that I actually never make. The apparent root of the difficulty is that information does not wear its provenance from a particular sensory modality on its face—information is amodal. Perhaps 15

Cf. Byrne 2005, 96–98.

192 Theories of Introspection the idea that one knows what one believes and knows by directing attention “outward—on the world” (Evans 1982, 225) has something going for it. But the amodal nature of information, it might be thought, shows that perception is where this idea irretrievably runs into sand. What are the alternatives?

6. Alternatives to Transparency According to the transparency proposal, I know that I see a hawk by an inference from a single premise about the hawk-infested landscape beyond. There two main alternative options. Option 1 is that no premise about my environment is needed: I know that I see a hawk without appealing to evidence concerning the scene before my eyes—I know non-observationally that I see a hawk. (‘Observation’ is meant to cover only observation of the normal visual sort; on option 1, perhaps I know that I see a hawk by introspection, conceived of as a kind of inner perception.) Option 2 is that although a premise about my environment is needed, it is not enough: additional mental evidence is required. Let us take these in turn.

6.1. Option 1: Non-Observational Knowledge Option 1, that I know non-observationally that I see a hawk, requires immediate amendment. First, note that this does not apply to every case of knowing that I see a hawk, because sometimes an environmental premise is plainly needed: I know that I see that bird (pointing to a hawk perching atop a distant tall tree), but I am not in a position to know that I see a hawk. Ryle is passing by and informs me that the bird is a hawk; with this environmental premise in hand, I conclude that I see a hawk. Second, extending this first point, perhaps one can never know nonobservationally that one sees a hawk—all such knowledge is based on evidence that one sees such-and-such, and that such-and-such is a hawk, with the latter item of evidence being known observationally. So a more careful and general statement of option 1 is as follows: knowledge that one sees an F/this F is either non-observational, or else based on evidence that includes the fact that one sees a G/this G, known non-observationally. If there is any non-observational knowledge of this sort, knowledge that one sees this red spot (pointing to a clearly visible red spot) is an example, or so we may suppose. Since the fact that one sees this red spot entails that this spot is red, one may come to know that this spot is red by inference from the fact that one sees this

Knowing What I See

193

red spot. Now one may also know that this spot is red simply by looking at it—an animal with no conception of seeing could use its eyes to know that this spot is red. So no knowledge that one sees this red spot is necessary. Thus, on this view, there are two routes to the same conclusion: one may know that the spot is red twice over, by inference from a non-observationally known fact about what one sees, and by the more familiar method of simply using one’s eyes. This result is more than strange. First, note that one may see what is, in fact, a red spot, even though the spot does not look red (perhaps one is viewing the spot in very dim light). One is not able to tell by looking that this spot is red, but one might have various backup routes to that conclusion—perhaps one painted the spot oneself from a can of red paint. However, the alleged non-observational backup route is clearly inoperative: although it is true that one sees this red spot, no amount of introspection will reveal this fact. The obvious explanation is that in cases where the spot is clearly visible, the information one obtains by vision about the spot is somehow used to derive the conclusion that one sees this red spot, but if that is right then option 1 must be rejected. Second, note that when one sees a red spot and believes both that this spot is red and that one sees this red spot, it is not a possibility that two spots are in play. Could this red spot be a different spot from this red spot that one sees? That is not a serious question, but if one knew non-observationally that one sees this red spot it apparently would be. Return to the situation in which one views this red spot in dim light. Suppose one remembers that one painted this spot red; on occasion, one might reasonably wonder whether one’s memory was quite accurate—perhaps one painted another spot red, not this very spot. As before: the obvious explanation of why the identity of the spot is never in question is to say that the information about this spot is used to derive the conclusion that one sees it. Finally, if I know non-observationally that I see this red spot, then certain dissociations are to be expected. In particular, one’s vision and reasoning capacities might be working perfectly normally, while the mechanism that yields nonobservational knowledge that one sees this red spot is broken or absent. One’s only means of finding out that one sees a red spot would then be similar to thirdperson cases: one knows that one sees a red spot because one knows that there is a red spot right there, that the light is good, that one’s eyes are open, and so forth. Often one knows through vision about an object’s location and other features, but is unsure whether someone else sees it (perhaps one does not know that the person’s gaze is in the right direction). Similarly, someone who only had thirdperson access to her states of seeing would sometimes be in a state of uncertainty about whether she saw an object, while quite certain (via her excellent vision) about the nature of the object itself. It is safe to say that this bizarre condition

194 Theories of Introspection never occurs.16 Pending some explanation of why the non-observational mechanism never fails in this way, this is a reason for thinking that option 1 is incorrect.17

6.2. Option 2, First Pass:Visual Sensations Since option 1 faces some serious objections, let us turn to option 2, that additional evidence is required. And from a more traditional position in the philosophy of perception, the need for such evidence is palpable. Seeing an object is a matter of the object causing distinctive sorts of affectations of the mind, “visual sensations.” It is thus natural to think of knowledge that one sees an object as resting on evidence about both ends of this causal transaction—evidence about the object coming from observation, and evidence about the sensation coming from some other source. So, 16 As Ned Block pointed out to me, the closest approximation in the literature appears to be the case of “reverse Anton’s syndrome” described by Hartmann et al. (1991; see also Block 1997, 159). The patient was initially diagnosed as blind due to a stroke. Two years later he was found to have spared vision in a 30º wedge in both fields. Anton’s syndrome patients deny that they are blind; this patient denied that he could see. At one point he remarked that “you (the examiners) told me that I can see it, so I must be able to see it” (Hartmann et al. 1991, 33). However, the patient’s vision was far from excellent. He could read words, but with limited accuracy (51% correct on a standard test). Strikingly, he was “unable to discriminate light from dark” (37). The patient’s cognition was also impaired, with mild language and memory deficits. Further, sometimes he used perceptual verbs in describing his condition: on a color-naming task, “he maintained that he could “feel” or “hear” the color” (34). The correct description of the patient’s predicament is unobvious. As Hilbert notes, “a certain amount of scepticism about the case is in order” (1994, 449). And, as Hartmann et al. say, reverse Anton’s syndrome is not clearly documented in any other published case. 17 The bizarre condition is what Shoemaker (1994) calls “self-blindness” with respect to seeing. It is worth emphasizing that someone who is self-blind with respect to vision is not the “super-duper blindsighter” of Block, who has “blindsight that is every bit as good, functionally speaking, as [normal] sight” (1997, 409). There are two differences. The first is that Block’s superduper-blindsighter denies that he sees anything, whereas the self-blind person knows by thirdperson means that she sees things.This first difference probably just reflects unimportant differences of detail between these two science-fiction stories; Block would not deny that a super-duperblindsighter could investigate his own states of seeing third-personally. The second difference is the important one: the super-duper blindsighter has a faculty “that is every bit as good, functionally speaking,” as normal vision, except that the resulting perceptual states lack “phenomenal consciousness.” The super-duper-blindsighter is thus, as Block says, a “quasi-zombie” (1997, 409), or a “visual-zombie,” in something close to the usual sense of ‘zombie’ (not Dretske’s sense: see section 2).The self-blind person has normal vision, at least in the sense that she sees what we see, and has the sort of perceptual knowledge that we have, but lacks the “peculiar access” (Byrne 2005) that we have to our states of seeing. Nothing is being assumed, one way or the other, about whether the self-blind person’s visual states have phenomenal consciousness in Block’s sense. Dissociation problems also afflict option 2, but this will not be discussed further.

Knowing What I See

195

in order to know that I see a hawk, I need to know, inter alia, that I am having a visual sensation. Such a sensation is an occurrence in my mind, not on the fence post beyond, so no wonder peering at the hawk is not sufficient. An analogy can clarify the traditional position further. I am holding a nettle, and feel a stinging pain in my hand. How do I know the additional fact that the nettle is stinging me (i.e., causing the pain)? It would be a mistake to investigate the issue by concentrating solely on the nettle; rather, I need to attend to something else entirely, namely the pain in my hand. Putting these two items of evidence together— that I am holding a nettle and that I have a pain in my hand—I can conclude that nettle is stinging my hand. That conclusion is not entailed by my evidence, but in the circumstances my evidence strongly supports it. Likewise, on the present suggestion, I can conclude that I see a hawk on the basis of two items of evidence: the external nonpsychological fact that a hawk is present and the internal psychological fact that a visual sensation is occurring. (Note that placing a substantive restriction on the type of visual sensation would not be advisable, since almost any kind of visual sensation could accompany seeing a hawk—it could look blue, or cubical, or whatever.) Knowledge that one sees an F, then, is obtained by following this rule: seei: If an F is present and you are having a visual sensation, believe that you see an F. Evidently seei is hopeless. Taking the existence and epistemology of “visual sensations” for granted, on many occasions one knows that an F is present and that one is having a visual sensation, yet one does not see an F. seei is thus a bad rule. Moreover, we do not follow it. Suppose I see a sheep in a field; although no hawk is in sight, I know that there is a hawk in the vicinity. I have no inclination to follow seei and conclude that I see a hawk. Although rules are generally defeasible—despite knowing that x is a red-headed bird, one might have additional evidence that prevents one following woodpecker—it is unclear what the defeater might be in this case.18 We can pass over attempts to add epicycles to seei, because the nettle analogy is fundamentally defective. When I see a hawk I do not have a spectacular kind of migraine headache whose only connection to the hawk is that it is caused by the hawk. This is basically Ryle’s point when he observes that in the “unsophisticated use of ‘sensation’” a typical case of seeing does not involve any sensations ([1949] 18

Another problem is due to the word ‘present’ in the antecedent.This prevents me from always believing that I see a hawk, since I always believe that there are hawks somewhere. But ‘present’ excludes too much—in principle, I can see a hawk at any distance (cf. seeing a supernova) and also readily know that I see it.

196 Theories of Introspection 1963, 228). One can know what stinging sensations are without knowing anything about nettles, but insofar as the philosophical notion of a “visual sensation” is intelligible, it is not likewise only externally related to its causes. Visual sensations or, better, visual experiences, are specified in terms of the region of the external world that they purportedly reveal. That is, when I look at the hawk and recognize it as such my visual experience is an experience of a hawk. Does this reconception of visual sensations as visual experiences help rescue Option 2?

6.3. Option 2, Second Pass:Visual Experiences of an F Start by applying the reconception to seei: Seeii: If an F is present and you are having a visual experience of an F, believe that you see an F. This straightforwardly copes with the case where I see a sheep in a field and know that there is a hawk in the vicinity, which I do not see. I do not have an experience of a hawk, and so am not in a position to follow seeii. But what is it for a visual experience to be “of ” a hawk? An influential discussion of this question is in Searle’s 1983 book Intentionality. Searle writes: I can no more separate this visual experience from the fact that it is an experience of a yellow station wagon that I can separate this belief from the fact that it is a belief that it is raining; the “of ” of “experience of ” is in short the “of ” of Intentionality. (Searle 1983, 39) An experience of a hawk may be said to be “of ” a hawk in the same way that a belief about a hawk is “of ” a hawk. Experience, then, like belief, has intentionality: my experience of a hawk and the belief that there is a hawk on the fence post are both “of ” or “about” a hawk. But the parallel, Searle thinks, is even closer. The belief that there is a hawk on the fence post has propositional content, namely the proposition that there is a hawk on the fence post. And likewise for visual experiences: The content of the visual experience, like the content of the belief, is always equivalent to a whole proposition. Visual experience is never simply of an object but rather it must always be that such and such is the case. (40) In the case of an experience of a yellow station wagon, “a first step in making the content explicit,” Searle says, “would be, for example, I have a visual experience (that there is a yellow station wagon there)” (41).19 19

Searle’s considered view is that the content is the proposition that “there is a yellow station wagon there and that there is a yellow station wagon there is causing this visual experience” (1983, 48), which has attracted a lot of criticism. See, e.g., Recanati 2007, ch. 17.

Knowing What I See

197

On this view, perceptual experiences have content, like belief, desire, and other “propositional attitudes.” To a first approximation, one may think of the content of the subject’s visual experience as the information (or misinformation) delivered to the subject by his faculty of vision (cf. Armstrong 1968: 224). When the delivery is one of misinformation, the subject suffers a visual illusion. Although this is somewhat controversial, it is at least a huge advance over the sense datum theory, and the traditional view mentioned in the previous section.20 Assume, then, that visual experiences have contents, v-propositions; true v-propositions are v-facts.21 Let ‘[ . . . F(x) . . . ]V’ be a sentence that expresses a particular V-proposition that is true at a world w only if x is F in w. Read ‘You V[ . . . F(x) . . . ]V’ as ‘You have a visual experience whose content is the proposition that [ . . . F(x) . . . ]V’. Then a more explicit version of seeii is: seeiii: If you V[ . . . F(x) . . . ]V and x is an F, believe that you see an F. What exactly are v-propositions? Searle’s example—the proposition that there is a yellow station wagon there—is at best a “first step,” as he says: it hardly begins to capture the apparent scene before the eyes when one sees a yellow station wagon. In fact, it might not even be a first step. Does the content of visual experience concern station wagons, hawks, and the like, as such? If the ostensible yellow station wagon is actually white, vision is surely to blame for delivering misinformation to the subject. But what if the ostensible station wagon is a sedan? Here there is a temptation to exonerate vision, and instead to point the finger at the subject’s judgment that the car is a station wagon. The issue is less than clear, and in any event disputed. Granted that visual experiences have contents, it is not disputed that the content at least concerns what falls under the rubric of “mid-level vision” in vision science: shape, orientation, depth, color, shading, texture, movement, and so forth: call these sensible qualities. In fact, without begging any important questions we can restrict v-propositions so that they just concern sensible qualities. (This restriction is imprecise, but that will not matter here.) With this restriction, and letting ‘[ . . . x . . . ]V’ express a v-proposition that is true at a world w only if x has a certain sensible quality in w (i.e., if x is red, or square, . . .), we get: seeiv: If you V[ . . . x . . . ]V and x is an F, believe that you see an F.

20 For skepticism about the existence of “visual experiences” as Searle and many other philosophers conceive of them, see Byrne 2009. That skepticism will be suspended here. 21 Some disjunctivists will deny that illusory cases involve having a visual experience whose content is a false proposition (see the introduction to Byrne and Logue 2009). However, they have no special reason to dispute this chapter’s account of the epistemology of veridical cases.

198 Theories of Introspection Notice that because ‘F’ does not appear in the scope of ‘V’, this is an improvement on seeiii. Return to an example given in section 6.1: I see a bird atop a tall tree, too far away to make out its shape or color, which Ryle tells me is a hawk. I am presumably not having an experience “of a hawk,” since the information available to my visual system is too impoverished. I therefore cannot follow seeiii. But I can follow seeiv, since I am having a visual experience with a content that concerns the hawk, albeit a content that does not identify it as such. Although seeiv is the best attempt so far, it is not good enough. Recall that ‘[ . . . x . . . ]V’ expresses an object-dependent proposition—one whose truth at world w depends on how a certain object (namely x) is in w. Further, it is very plausible that one can only enjoy a visual experience with such an object-dependent content in a world in which the object exists (at some time or another). That is,‘You V[ . . . x . . . ]V’ entails ‘x exists’. But if that is right, then we are back in the same bind that afflicted the suggestion that I can know that I see a hawk non-observationally (see section 6.1). Since the existence of x is entailed by the proposition that I V[ . . . x . . . ]V, I have two routes to the conclusion that that object (the hawk) exists. And, as before, the account leaves open a possibility that should be closed, namely that there are two objects, one known about through vision, and the other known about through non-observational means. Can these problems be avoided by denying that v-propositions are objectdependent? The view is not well motivated. By perceiving, in particular by seeing, one may come to know things about individual objects in one’s environment—that that is a hawk, for example. It is thus natural to think that the information delivered by vision is object-dependent: the testimony of one’s visual system concerns this very hawk. Still, this alternative needs examining further. Suppose, then, that when I see the hawk, it is not pinned down by a v-proposition with the hawk as a constituent, but rather by a proposition that identifies the hawk by description. (For the sake of the argument, we can ignore the difficult question of what this description exactly is.) Here is the descriptive counterpart of seeiv: seev: If you V[ . . . (the G) . . . ]V and the G is an F, believe that you see an F. Apart from paucity of motivation, is there anything wrong with it? Consider a case where I think or suspect that I am suffering from an illusion. I see a hawk, but I doubt that the hawk is the way it looks. Perhaps the hawk looks like a penguin right in front of me, and I have reason to believe that this is the product of a devious arrangement of distorting mirrors, with the ordinary-looking hawk being positioned behind my back. However ‘the G’ is filled in, we may safely suppose that I do not know or believe that the hawk is the G. seev is thus of no help. Nonetheless, nothing prevents me from knowing in an ordinary sort of way that I

Knowing What I See

199

see a hawk. For instance, I know that I see a hawk because Ryle tells me that this (clearly referring to the penguin look-alike before me) is a hawk. Since there is nothing epistemologically special about this case, if seev does not explain my knowledge here, it does not explain it elsewhere. Even taking the ontology and epistemology of “visual experiences” for granted, there are no easy alternatives to the transparency proposal. So let us revisit it.

7. Back to Transparency For the moment, shelve illusions and concentrate on veridical cases, where one sees an object and it is as it looks. Return to the object-dependent suggestion: seeiv: If you V[ . . . x . . . ]V and x is an F, believe that you see an F The problems just rehearsed are all in effect traceable to the ‘V’, which suggests the experiment of dropping it. And removing the ‘V’ yields a version of the transparency proposal: see: If [ . . . x . . . ]V and x is an F, believe that you see an F The amodal problem of section 5 seemed to doom the transparency proposal. Could v-facts rescue it? Recall that a v-fact concerns the sensible qualities of objects in the scene before the eyes. In one way this notion is perfectly familiar. When I see the hawk on the fence post, a segment of the visible world is revealed: an array of colored, textured, three-dimensional objects, casting shadows, some occluding others, at varying distances from my body, with various illumination gradients, and so forth. A certain v-fact just specifies that array, the scene before my eyes. If Ryle and I strike up a conversation about the spectacular view of the North York Moors, v-facts are our topic. On the other hand, giving a theoretically satisfying characterization of v-facts is difficult. Armstrong, for instance, speaks of perceptual content as comprising “certain very complex and idiosyncratic patterns of information about the current state of the world” (1968, 212), while declining to be much more specific.22 Even vision science often in effect dodges the issue with placeholders like ‘visual representation’. Complexity or informational richness is no doubt part of the story, but even in the case of viewing a very simple scene—say, a red spot against a grey background—it is unclear how to proceed. Just concentrating on one feature of the 22

A rare example of a more detailed account is in Peacocke 1992.

200 Theories of Introspection spot, its hue, the predicate ‘is red’ (or even some made-up predicate like ‘is red29’) does not quite do it justice. The particular red hue of the spot might be a little yellowish, or alternatively a little bluish; how exactly information about the hue is packaged by vision is not at all obvious.23 Even though the familiar may resist theory, fortunately for our purposes not much theory is required. Vision, we may say, reveals the visual world: the world of v-facts. In the visual world things are colored, illuminated, moving, and so on, but not smelly or noisy.24 Likewise, olfaction reveals the olfactory world: the world of o-facts. The olfactory world—at least, our olfactory world—is a relatively impoverished place, consisting of odors located around the perceiver’s body. The auditory world, the world of a-facts, is considerably more complicated, consisting, inter alia, of sounds of varying loudness and pitch at a variety of locations. One may base one’s actions and inferences on how things are in the visual world—this just requires a sensitivity to different aspects of one’s environment. (In particular, it does not presuppose self-knowledge.) Suppose one investigates one’s environment and finds that a certain v-fact, the fact that [ . . . x . . . ]V, obtains. Vision is, at least in creatures like ourselves, an exclusive conduit for v-facts. Hence, one’s information source must be vision, not audition, olfaction, testimony, or anything else. Although information is amodal in principle, for us v-facts do indicate their provenance—(visual) information is practically modal. Thus, see apparently solves the amodal problem. What about the evidential problem? That has not gone away, because the fact that [ . . . x . . . ]V remains stubbornly devoid of vision. That is, the hawk before my eyes, with its rich variety of visual sensible qualities, offers no indication at all that it is seen. Still, see takes the sting out of the evidential objection much as know did. Recall that the latter rule is: know: If p, believe that you know that p. Section 4 noted that know is self-verifying: if one follows it, then one’s belief that one knows that p is true. know typically produces reliably true beliefs, and there is no clear barrier to supposing that it also typically produces knowledge. To that extent, the evidential objection is rebutted.

23

For a sketchy proposal about the visual representation of hue, see Byrne and Hilbert 2003, 14. The last two sentences should be qualified. First, cross-modal effects show vision does not reveal the visual world unaided—other modalities sometimes help too. Second, the negation takes wide-scope: it is not the case that there are smells or noises in the visual world. The visual world is silent on such matters. 24

Knowing What I See

201

see, in contrast, is not self-verifying: perhaps one could in principle learn that [ . . . x . . . ]V by reading it in the—as-yet-unwritten—language of vision; one would not thereby see x. But it is practically self-verifying: in all ordinary situations, one knows that [ . . . x . . . ]V only if one sees x. As section 4 also noted, self-verification is not sufficient for a rule to be knowledge-conducive; a fortiori, practical selfverification is not either. But the dialectical imperative is not to prove that following see yields knowledge; rather it is to reply to the evidential objection. And for that, practical self-verification will do.

7.1. The Memory Objection The claim that see is practically self-verifying might be thought to be too strong. Surely, if v-facts can be known, they can be remembered. Shouldn’t we then have said: in all ordinary situations, one knows that [ . . . x . . . ]V only if one sees or saw x? And if so, there is the following difficulty. Suppose I see a red spot (x) at time t1.Write the relevant v-fact as ‘the fact that [. . . . Red(x, t1) . . . ]V’, and further suppose that I remember it. Shortly after, at t2, a piece of cardboard is placed in front of the spot, completely occluding it; I am quite confident that the spot itself has not changed color: the distinctive visual way the spot was is the way the spot now is. I know (we may assume) that [ . . . Red(x, t2) . . . ]V. Granted all this, I am in a position to follow see and conclude that I see a red spot. But obviously I do not. Why not? Either something blocks in the inference in this case, or I do not follow see in any circumstances. Once this disjunction is conceded, it is hard to avoid the second disjunct.25 However, the memory objection should not be allowed to get started in the first place. Consider recalling something one has seen—say, recalling the red spot one saw. Such an exercise of one’s episodic memory is akin to visual imagery: it is somehow visual in character, but easily distinguished from actually seeing a red spot. In a Humean framework, this is because, in episodic recollection, one is aware of a faded copy (an “idea”) of one’s past visual “impression” (or sense datum). The similarity is explained by the copying, the difference is explained by the fading. Granting the existence of impressions and ideas, this purported explanation is perhaps the best that can be found. 25 Could the fact that the cardboard “occludes” the spot block the inference? No. If ‘occludes the spot’ means ‘prevents me from seeing the spot’, this just raises the question how I know the cardboard occludes the spot. On the other hand, if it means ‘is opaque and in front of the spot’ then my knowing this fact does not explain why I do not follow see. Suppose I can in fact see the spot, due to some devious arrangement of mirrors, or because I have suddenly gained Superman’s ability to see through walls. Despite knowing that the cardboard is opaque and in front of the spot, I would follow see and conclude that I see it.

202 Theories of Introspection The Humean attempt at an explanation has a (superior) counterpart in the present information-based framework. In episodically recollecting the spot, one is aware of a segment of the past visual world, but although the information is packaged visually, it is a transformed and degraded version of the visual information that characterizes successful seeing. The similarity is explained by the visual packaging, the difference is explained by the transforming and degrading.26 If that is right, then it was too hasty to say that I remember that [. . . . Red(x, t1) . . . ]V. What I do remember could be written as ‘the fact that {[. . . . Red(x, t1) . . . ]V}’, where the curly brackets represent the episodic-memory transformation and degradation of the true v-proposition that [. . . . Red(x, t1) . . . ]V. And if we not unreasonably assume that memorially transformed and degraded v-facts are disjoint from v-facts, then remembering that {[. . . . Red(x, t1) . . . ]V} will not thereby put me in a position to follow see, and the memory objection fails.27 On this account, one might expect that in some cases, where other cues are absent, episodic visual recollection (or visualizing more generally) will indeed be mistaken for seeing. Since this is basically the converse of the Perky effect (where one mistakes seeing for visualizing), it can be called the converse-Perky effect. And there is some evidence that the converse-Perky effect occurs.28

8. Evans Again, and the Known-Illusion Problem So far we have concentrated on the veridical case: I see the hawk and it is as it looks. Let us now return to illusions. To give some examples more realistic than the one mentioned at the end of section 6: the hawk looks closer than it really is, or a shadow appears as a patch of darkened green on the field beyond, or the hawk is perching on a wall that generates Richard Gregory’s “café wall illusion.”29 In such cases, the fact I seem to apprise, that [ . . . x . . . ]V, where x = the hawk, is no fact 26

For a discussion of this, see Byrne 2011. Our visual memories are very impressive, at least under some conditions (Brady et al. 2008). But as far as I know, the evidence supports the transformation and degradation hypothesis. Some examples: Burnham and Clark 1955 (memory for hue), Uchikawa and Ikeda 1986 (memory for brightness). Clearly much more could be said here, though. 28 See Goldenberg, Müllbacher, and Nowak 1995 (blindness denial arguably explained by the patient’s spared visual imagery; see also Byrne 2010, 117–118). A more common example might be this. Suppose one is in bed, in almost total darkness. One opens one’s eyes and looks in the direction of a familiar object—a desk, say. One episodically recollects its distinctive visual qualities as they appear from this angle. Does one see the desk? When in this sort of situation myself, I sometimes wonder whether I am merely visually recollecting the object rather than seeing it. 29 See . 27

Knowing What I See

203

at all. Still, I can easily discover that I see a hawk, just as I did in the original veridical example. If I do not know that I am illuded, this case presents no difficulty. Say that one tries to follow the rule ‘If conditions C obtain, believe that p’ iff one believes that p because one believes that conditions C obtain. If one follows a rule one tries to follow it, but not conversely.30 One cannot follow see if the relevant v-proposition is false, but one can try to follow it. And in the illusory example of the previous paragraph, if I try to follow see, then I will end up with a nonaccidentally true belief that I see a hawk, for essentially the same reasons as before.31 The problem, rather, is similar to the one faced by seev at the end of section 6, and concerns the case when I know (or believe) that I am illuded. The method I use to discover what I see does not obviously alter when I know (or believe) that the hawk is not the way it looks: I can still know that I see it by attending to the hawk. If the transparency proposal applies at all, it must apply unmodified across the board. But if I do not believe the relevant v-propositions, I cannot even try to follow see. Hence, cases of known-illusion threaten to blow the transparency proposal entirely out of the water.

8.1. Evans’s Proposal Recall the quotation from section 3, where Evans is explaining how someone may gain knowledge of how things perceptually appear by “re-using precisely those skills of conceptualization that he uses to make judgements about the world.” The quoted passage contained an elision, and it is time to restore it. Here are the crucial sentences. [The subject] goes through exactly the same procedure as he would go through if he were trying to make a judgement about how it is at this place now, but excluding any knowledge he has of an extraneous kind. (That is, he seeks to determine what he would judge if he did not have such extraneous information.) (1982, 227–228) Consider the following case. I am staring at what I know to be a gray patch on a green background. Due to a color contrast effect, the patch will look slightly 30

See Byrne 2005, 97. An objection at this point is that one cannot come to know something by inference from a false premise, a moral commonly drawn from Gettier cases. But an arguably better diagnosis of the Gettier cases is that safety (in the sense of Sosa 1999 and Williamson 2000, ch. 5) is a necessary condition for knowledge, not that no reasoning through false steps is a necessary condition for knowledge. And beliefs produced by trying to follow see will often be safe (cf. Byrne 2005, 96–98). See also Silins, this volume. 31

204 Theories of Introspection reddish. Since I am aware of the effect, I do not believe the relevant v-proposition, that [ . . . Reddish(x) . . . ]V, where x = the patch. I know that I see a grey patch, but cannot know this by trying to follow see. Evans’s remarks suggest the following two-step alternative. First, I verify a certain counterfactual truth: if I had not known extraneous facts, I would have judged that [ . . . Reddish(x) . . . ]V. That tells me that x looks reddish, and so that I see x. I then add in the fact that x is a grey patch, and conclude from this that I see a grey patch. One immediate problem with this suggestion turns on the notion of “knowledge of an extraneous kind.”32 The effect of excluding extraneous knowledge is intended to make me rely exclusively on the testimony of vision, but it cannot be characterized as “facts I know other than by current vision” on pain of circularity. Could an extraneous piece of knowledge be characterized simply as something that I previously knew about the patch? Then the counterfactual to be verified is ‘If I had not known anything about the patch beforehand, I would have judged that [ . . . Reddish(x) . . . ]V’. This suggestion has a number of problems. First, it is quite implausible that a counterfactual of this sort will always be true in every case, or that I will judge that such a counterfactual is true.33 Second, intuitively it gets things back to front. If I do know that the counterfactual is true, then isn’t this because I know the patch looks reddish? Finally, in bringing in sophisticated counterfactual judgments about my own mind, the attractive idea that I can know that I see a hawk merely by attending to the hawk has been thrown overboard.

8.2. Belief-Independence The known-illusion problem is entirely generated by the widespread assumption that, as Evans puts it, there is a fundamental (almost defining) property of states of the informational system,34 which I shall call their ‘belief-independence’: the subject’s being in an informational state is independent of whether or not he believes that the state is veridical. It is a well-known fact about perceptual illusions that it will continue to

32

More generally, it should be belief, not just knowledge. For example, suppose one has a known-illusion of motion by viewing Kitaoka’s “rotating snakes” figure (). Assume, with Evans, that one does not believe that anything in the figure is moving. If one had not known anything about the figure beforehand, would one have judged that anything in the figure were moving? That depends. The figure and the motion both look so unusual, that a sensible person might well smell a rat. (Cf. Jackson 1977, 40–41.) 34 Which subserves “perception, communication, and memory” and “constitutes the substratum of our cognitive lives” (1982, 122). 33

Knowing What I See

205

appear to us as though, say, one line is longer than the other (in the Müller-Lyer illusion), even though we are quite sure that it is not. (1982, 123)35 Put in the present notation: even though one’s visual system may (mis-)inform one that [ . . . x . . . ]V, one may nonetheless resist its testimony and not believe this v-proposition. But is it true that perception is belief-independent? Evans’s correct observation about the Müller-Lyer illusion does not immediately establish this conclusion. He notes that it may appear to one that the lines are unequal even though one believes they are equal. For belief-independence to follow, it must also be assumed that if one believes that the lines are equal, one does not also believe that they are unequal. And since having contradictory beliefs is a familiar phenomenon, this assumption needs to be backed up with an argument. Let us call the view that vision constitutively involves belief in the relevant v-proposition, belief-dependence. (Belief-dependence is, more-or-less, the “judgemental theory of perception” defended in Craig 1976.) Belief-dependence is not, it should be emphasized, the view that to enjoy visual appearances is simply to have beliefs of a certain sort. (For a reductive theory along these lines, see Armstrong 1961, ch.9.) Neither it is the view that perception can be analyzed partly in terms of belief. In these respects, belief-dependence is analogous to the view that knowledge constitutively involves belief: that does not imply that knowledge is belief, or that it can be analyzed partly in terms of belief. Although the passage from Evans does not conclusively establish that belief-dependence is false, it might be thought that the idea that one has contradictory beliefs in cases of known illusion is implausible. So can anything positive be said in favor of belief-dependence? Here, briefly, are three considerations.36 First, presumably some animals with visual systems very similar to ours (some other primates, say), cannot cognitively override visual illusions: in this sense, for them, seeing is always believing. Belief is thus built into their visual systems. And since we have basically the same visual systems, seeing is believing for us too. Second, perception is clearly belief-like—which is why Armstrong-style attempts to reduce perception to belief were certainly worth trying. Perception compels belief: the visual appearance of unequal lines is accompanied by the belief that the lines are unequal, absent (apparent) evidence to the contrary. And perception has 35 There is a slight infelicity in this passage. Evans has ‘believes that the state is veridical’ where it would have been better to write ‘believes that p’, where the proposition that p is (on his view) the “conceptualized” version of the content of the experience. (The vexed issue of conceptual versus nonconceptual content is not addressed here.) 36 See also Byrne 2009, 450–451 and 2010, 120–121.

206 Theories of Introspection the same “direct of fit” as belief: false beliefs and illusory perceptions are mental states that are both failures, in some (admittedly obscure) sense. Belief-dependence explains both these features. Compulsion is explained simply because the visual appearance of unequal lines is always accompanied by the belief that the lines are unequal. Sometimes that belief will not be manifest because it is suppressed by the contrary belief that the lines are equal; remove that contrary belief, and one will have an unsuppressed belief that the lines are unequal, that will manifest itself in the usual way. And direction of fit is explained because the failure of a constitutive component of a perceptual state presumably implies the failure (or less than complete success) of the state as a whole. Finally, consider the really quite remarkable phenomenon that numerous notlong-dead philosophers claimed to believe the deliverances of vision even in cases of illusion.37 “When I see a tomato,” H. H. Price famously declared, “there is much I can doubt. I can doubt whether it is a tomato that I am seeing, and not a cleverly painted piece of wax. I can doubt whether there is any material thing there at all. Perhaps what I took for a tomato was really a reflection, perhaps I am even the victim of some hallucination. One thing however I cannot doubt; that there exists a red patch of a round and somewhat bulgy shape, standing out from a background of other color-patches, and having a certain visual depth” (Price 1932, 3). On the orthodox view, the plain man does not believe that there exists a bulgy red patch when he knows that the devious color-illusion has been explained. So why, on the orthodox view, do distinguished philosophers like Price believe the contrary after careful phenomenological study? Are they insane? On some accounts of delusions (e.g., the Capgras delusion), they involve beliefs that are “modular” in something like the sense of Fodor 1983: delusory beliefs are largely inferentially isolated and persist despite evidence to the contrary (see, e.g., Jones, Delespaul, and van Os 2003). Belief-dependence offers a similar model of cases of known illusion. Since one believes the relevant v-proposition in a case of known illusion, one is in a position to (try to) follow see. Therefore, the knownillusion problem does not arise.38 37 More exactly, the deliverances of (roughly) mid-level vision. There are also many long-dead examples. 38 If an ideal of rationality is avoidance of inconsistency, then belief-dependence implies that someone who suffers a perceptual illusion thereby falls short of the rational ideal. (As Craig 1976, 15–16 points out; see also Glüer 2009, 303 n. 10.) Can this be turned into a convincing objection? No. It will not do simply to claim that the illuded subject is not, or need not be, irrational. Taken as a claim about a rational ideal, its truth is not evident.Taken as an ordinary sort of remark, on the other hand, it is true but not in conflict with belief-dependence.The belief that the subject knows to be false (e.g., a certain v-proposition that is true only if the lines are unequal) does not influence her verbal reports about the lengths of the lines, or any plans for action based on the

Knowing What I See

207

To close, three features of the transparency proposal should be highlighted. First, the transparency proposal is an inferentialist account of knowledge of what one sees; an odd inference, to be sure, but an inference nonetheless. Second, the transparency account is economical: it explains self-knowledge in terms of epistemic capacities and abilities that are needed for knowledge of other subject matters. And third, the account is detectivist: broadly causal mechanisms play an essential role in the acquisition of knowledge of what one sees, as they do generally in knowledge of contingent matters. Economy is needed to explain why intelligent subjects with normal vision inevitably know what they see in the usual distinctive first-personal way.39 Detectivism is just common sense. Inferentialism, on the other hand, runs counter to the usual characterization of self-knowledge as “direct” (that is, not inferential). Does this mean that the transparency proposal faces yet another serious objection? That is left as an exercise.40

lengths of the lines. She is not therefore ‘irrational’ in the practical sense of an ordinary accusation of irrationality. The subject’s belief that the lines are unequal does little harm—at worst, it would make her a sense datum theorist. Indeed, given the epistemological account of this chapter, it actually does some good, by allowing the illuded subject to know what she sees. Glüer, who thinks that belief-dependence founders on this sort of consideration, asserts that “there is nothing ‘irrational’ about the lines looking of different length” (2009, 303 n. 9). But she does not explain why this is true on the required reading of ‘irrational’. 39 That is, it is needed to explain why dissociations of the sort mentioned at the end of section 6.1 do not occur. 40 Many thanks to audiences at Stockholm University,Victoria University of Wellington, ANU, IJN, and NYU. I am especially grateful to Ned Block, Jesse Prinz, François Recanati, Nico Silins, Declan Smithies, Robert Stalnaker, and Daniel Stoljar.

208 Theories of Introspection

References Anscombe, G. E. M. 1965. “The Intentionality of Sensation: A Grammatical Feature.” In Analytical Philosophy, second series, edited by R. J. Butler. Oxford: Blackwell. Armstrong, D. M. 1961. Perception and the Physical World. London: Routledge & Kegan Paul. ———. 1968. A Materialist Theory of the Mind. London: Routledge & Kegan Paul. Block, N. 1997. “Biology Versus Computation in the Study of Consciousness.” Behavioral and Brain Sciences 20: 159–165. Brady, T. F., T. Konkle, G. A. Alvarez, and A. Oliva. 2008. “Visual Long-Term Memory Has a Massive Storage Capacity for Object Details.” Proceedings of the National Academy of Sciences 105: 14325–14329. Burnham, R., and J. Clark. 1955. “A Test of Hue Memory.” Journal of Applied Psychology 39: 164–172. Byrne, A. 2005. “Introspection.” Philosophical Topics 33: 79–104. ———. 2009. “Experience and Content.” Philosophical Quarterly 59: 429–451. ———. 2010. “Knowing That I am Thinking.” In Self-Knowledge, edited by A. Hatzimoysis. Oxford: Oxford University Press. ———. 2011. “Recollection, Perception, Imagination.” Philosophical Studies 148: 15–26. Byrne, A., and D. R. Hilbert. 2003. “Color Realism and Color Science.” Behavioral and Brain Sciences 26: 3–21. Byrne, A., and H. Logue. 2009. Disjunctivism: Contemporary Readings. Cambridge, Mass.: MIT Press. Chalmers, D. 1996. The Conscious Mind. Oxford: Oxford University Press. Craig, E. 1976. “Sensory Experience and the Foundations of Knowledge.” Synthese 33: 1–24. Dretske, F. 1995. Naturalizing the Mind. Cambridge, Mass.: MIT Press. ———. 2003. “How Do You Know You Are Not a Zombie?” In Privileged Access: Philosophical Accounts of Self-Knowledge, edited by B. Gertler. Aldershot: Ashgate. Evans, G. 1982. The Varieties of Reference. Oxford: Oxford University Press. Fodor, J. A. 1983. The Modularity of Mind. Cambridge, Mass.: MIT Press. Gibson, J. J. 1979. The Ecological Approach to Visual Perception. Boston: Houghton Mifflin. Glüer, K. 2009. “In Defence of a Doxastic Account of Experience.” Mind and Language 24: 297–327. Goldenberg, G., W. Müllbacher, and A. Nowak. 1995. “Imagery Without Perception—A Case Study of Anosognosia for Cortical Blindness.” Neuropsychologia 33: 1373–1382. Hartmann, J., W. Wolz, D. Roeltgen, and F. Loverso. 1991. “Denial of Visual Perception.” Brain and Cognition 16: 29–40. Hilbert, D. 1994. “Is Seeing Believing?” PSA: Proceedings of the Biennial Meeting of the Philosophy of Science Association 1: 446–453.

Knowing What I See

209

Jackson, F. 1977. Perception: A Representative Theory. Cambridge: Cambridge University Press. Jones, H., P. Delespaul, and J. van Os. 2003. “Jaspers Was Right After All—Delusions Are Distinct from Normal Beliefs.” British Journal of Psychiatry 183: 285–286. Moore, G. E. 1903. “The Refutation of Idealism.” Mind 7: 1–30. Peacocke, C. 1992. A Study of Concepts. Cambridge, Mass.: MIT Press. Price, H. H. 1932. Perception. London: Methuen. Recanati, F. 2007. Perspectival Thought. Oxford: Oxford University Press. Ryle, G. [1949] 1963. The Concept of Mind. London: Hutchinson. (Page reference to Peregrine Books 1963.) Searle, J. R. 1983. Intentionality. Cambridge: Cambridge University Press. Shoemaker, S. 1963. Self-Knowledge and Self-Identity. Ithaca, NY: Cornell University Press. ——— 1994. “Self-Knowledge and ‘Inner Sense’.” Philosophy and Phenomenological Research 54: 249–314. Sosa, E. 1999. “How to Defeat Opposition to Moore.” Philosophical Perspectives 13: 141–153. Uchikawa, K., and M. Ikeda. 1986. “Accuracy of Memory for Brightness of Colored Lights Measured with Successive Comparison Method.” Journal of the Optical Society of America A 3: 34–39. Williamson, T. 2000. Knowledge and Its Limits. Oxford: Oxford University Press. Wittgenstein, L. [1921] 2001. Tractatus Logico-Philosophicus. Translated by D. F. Pears and B. F. McGuiness. London: Routledge and Kegan Paul.

This page intentionally left blank

8 Self-Knowledge, ‘Transparency’, and the Forms of Activity Richard Moran

1. Introduction Traditionally, the discussion of self-knowledge and self-consciousness in philosophy has given it a place central to the understanding of rationality and agency themselves, and in doing so it treats the ability to know one’s own mind as something more than a useful capacity we enjoy as human beings. In a contemporary context, this is a thought more often in the background of philosophical discussion than something defended explicitly. It is nonetheless an assumption recognizable enough to be found debatable, or difficult to make sense of in the context of contemporary understandings of the nature of mentality and agency themselves. This chapter begins by setting out in very general terms some considerations that would link self-knowledge to a certain form of agency, and then considers two recent studies that seek in their different ways to show that a certain form of account of self-knowledge (involving appeal to the “transparency” of belief) must be divorced from any appeal to rational agency. I will be arguing that in both cases the account that emerges from this divorce ends up with a kind of agency in the picture after all, only of the wrong kind, so that the ordinary exercise of coming to know what one believes

212 Theories of Introspection requires the person to exert a kind of external control over her own attitudes. In seeking to develop the relevant notion of rational agency in contrast to this, I cannot claim to be doing more than pointing to the place where I think we need some such notion. In the course of doing so, I try to characterize the sense of ‘activity’ or ‘agency’ that is relevant to a central class of cases of self-knowledge, and distinguish this sense of activity from the sense of activity indicating a process of production, or acting upon oneself so as to produce a belief. In thinking about self-knowledge and rational agency there are two broad directions from which we may begin to ask how they may be related to each other. We may ask, first of all, how self-knowledge matters to agency itself, that is, whether the specifically human forms of rational agency can be understood apart from the capacity for self-knowledge of the mental life that is expressed in that agency. Is our capacity to act for reasons, to be self-guided in that sense, dependent on our ability to know our mental life ‘immediately’? Can the ordinary ability to respond to reasons in one’s thinking, to consider reasons for and against some belief and respond accordingly, be understood apart from our capacity for immediate self-knowledge? And would the absence of the ordinary capacity for self-knowledge make no essential difference to our rational agency?1 And from the other direction we can ask how rational agency itself may matter to the understanding of self-knowledge; that is, whether the ordinary capacity to know what one thinks about something is part of the same capacity to determine one’s thought about that thing. Is our ability to know what we believe ‘immediately’, and with a kind of authority not shared by what we say about the beliefs of others, tied to the fact that our beliefs and other attitudes are expressions of our rational agency, and is there a notion of responsibility applying to a person’s relation to her attitudes that is related to the capacity for first-personal knowledge of them? One recent way of relating both sets of questions begins with the example of belief, and appeals to a notion of ‘transparency’ between a question about one’s belief and a corresponding question about the object of one’s belief. Thus, it has seemed to several philosophers that a distinctive feature of first-person discourse is that a person can answer a question about her own belief by addressing herself to the corresponding question about the topic of that very belief.2 Hence, if asked do I, R.M., think it will rain today, I can answer this question by giving my answer

1 Recent philosophical work has drawn connections between self-knowledge and agency in a variety of different ways. Here I will just mention Burge 1998, Bilgrami 2006, and O’Brien 2007. Particularly helpful to me in thinking about the issues of this paper has been Matthew Boyle’s paper ‘On ‘Making up your Mind and the Activity of Reason’ (forthcoming). 2 See Evans 1982, Edgley 1969, and Moran 2001. More recently Byrne 2005 and Shah and Velleman 2005 appeal to a notion of ‘transparency’ for belief, but offer very different accounts of it.

Self-Knowledge, ‘Transparency’, and the Forms of Activity

213

to the corresponding question about the rain, and not by inquiring in to the state of mind of a particular person. The fact that the answer is given by a particular person, the very person whose state of mind the first question directed itself to, must surely be part of the answer to how it is possible or legitimate to answer the question about one’s belief by reference to the question about rain. For ‘transparency’ of this sort surely does not apply to a question I may ask about the beliefs of another person. If the topic of my question is the beliefs of some other person, then my efforts to answer that question must address themselves to the facts concerning the state of mind of that particular person. That’s how it is, after all, with my efforts to answer questions about other topics. In seeking to answer a question about the inflation rate in China, I must direct my attention to China. Why, then, if the topic is some fact (attitudes of a certain kind) about oneself is it legitimate to answer the question in a way that seems to neglect the fact that it is about a particular person, and instead treat it as a question concerning the topic of the attitude itself (e.g., the weather)? Again, the fact that in such a case the answer is delivered by the very person whose state of mind is in question must be central to accounting for this, for if transparency is ever legitimate it must represent a systematic difference between relations to oneself and relations to others. The identity of the person whose state of mind is inquired into and the person answering the question must matter here.3 The form of account I give appeals to a form of agency that is part of a person’s being a creature with beliefs, and I claim that the transparency, which various philosophers have found attractive, cannot be accounted for without appeal to this agency. It would not, in general, make sense to answer a question about my state of mind (e.g., my belief about the weather) by attending to a logically independent matter (the weather itself) unless it were legitimate for me to see myself as playing a role in the determination of what I believe generally, not in the sense that beliefs typically owe their existence to acts of deliberation but that the responsiveness to reasons that belongs to beliefs is an expression of the person’s rational agency. However, while this form of account takes the topics of self-knowledge and agency to be closely related, the agency in question does not involve any kind of voluntarism about belief, and indeed the form of rational agency I have in mind has as a consequence that such voluntarism is false. The sense in which I see belief and other attitudes as forms of activity is deeply related to the fact that they are not matters of choice for the person, and hence the agency involved here is not that which is exercised when, say, a person chooses to raise her arm and then does so. 3

Elsewhere, I have presented an account of self-knowledge that seeks to vindicate the applicability of transparency to the first-person case, and with it the claim that in delivering an answer in this way, the person is indeed speaking from knowledge (Moran 2001).

214 Theories of Introspection

2. Knowing What and Knowing Why We can take the case of belief as representative of the attitudes generally, though it will be important to recognize that there will be differences between, for instance, believing, wanting, fearing, intending, hoping, and caring about. What they share as attitudes, however, is their involvement in forms of normative assessment, such as that of a belief ’s being justified, or a fear’s being unwarranted, or an activity’s being worth caring about. In this they differ from brute sensations, which I take to be aspects of our passive or receptive nature.4 Part of what is meant by this is that different forms of the question ‘Why?’ will apply to items in these two categories, and the relevance of asking the question of the person herself will also be different. There may be no special reason to ask me why I am experiencing pain in my lower back, although I might know something about it. And the answer to that question ‘why?’ will refer to such things as a previous injury, bad work habits, or compression on a disk. That is, the answers will contribute to the explanation of the coming to be of the sensation I am experiencing, but do not seek to say anything about the apparent point, or the good, or the intelligibility of the state I am in. Naturally, such causal versions of the question ‘why?’ will also apply to items in the other category as well (e.g., one’s beliefs, hopes, desires, and intentions), but it is internal to them that a different form of the question ‘why?’ also applies to them. Hence, we can ask why someone believes it will rain when we are asking for that person’s reasons for the belief, and we can ask why someone wants a saucer of mud when we want to know what could seem good or worth having in such a thing. This form of the question ‘why?’ thus seeks a certain normative characterization of the attitude itself, seeing it as reasonable or not, worthwhile or not. However, just as with the question seeking the origins of some condition, the answer to the question ‘why?’ is meant to tell us something relevant not just about the character of the attitude itself, but something relevant to the fact of the person’s having that attitude. That is, the way in which the attitude is found reasonable or intelligible by the person is assumed to be relevant to the question why she has that attitude, how it came to be part of her mental life, or what maintains it there. So this is something that this question shares with the kind of question ‘why?’ that applies to the person’s sensations, in that in both cases we are inquiring into how something comes to be or maintains itself. The difference is that in the case of the question ‘why?’ as applied to the person’s beliefs and intentions, the question of how it comes to be is tied to this normative 4

Ernest Sosa (this volume) makes a similar point with respect to the difference between experiences and “seemings”: “The seeming is rationally based; the experience is not. The seeming manifests the subject’s rational agency; the experience does not, any more than does a pain.”

Self-Knowledge, ‘Transparency’, and the Forms of Activity

215

question involving notions like the reasonable and the worthwhile. The second broad type of ‘why?’ question assumes a kind of dependence between these two sets of considerations: how something about a person comes to be or maintains itself, and what could be seen to be reasonable, intelligible, or worthwhile in it. And finally, there is a further difference in how these two types of question ‘why?’ are treated, and that is that for the broadly normative type of ‘why?’ question as applied to someone’s attitudes, we typically do ask that question of the person herself. Unlike the question ‘why?’ concerning a person’s pain or sensation of vertigo, we take the person herself to be uniquely relevant as the person to ask regarding what is reasonable or worthwhile in some attitude she holds. ‘Uniquely relevant’ or ‘indispensible’ here does not mean infallible or incorrigible, but rather that if she cannot tell us why, then we may begin to doubt whether there is a good answer to that ‘why’ question. This is not an assumption we make about the question ‘why?’ concerning sensations or other bodily conditions, when the person has no answer of her own.5 When philosophers claim that the notions of reason and justification are internal to the notion of belief, this is not meant to deny that children and animals can have beliefs, even though they do not themselves have these notions of reason and justification. To say this much is so far just to say that certain norms of rational assessment apply to beliefs. This is not yet different from the idea that, e.g., it is internal to something’s being a heart that it pumps blood and can be assessed as healthy or malfunctioning, even when the creature with the heart has no conception of such things. Controversy begins with the thought that, in mature humans, it belongs to the notion of belief that reason and justification not only apply to it, as a form of normative assessment, but that the believer play some role with respect to justification, a role that a creature does not typically play in the good functioning of its heart. The believer can be asked for her reasons for a certain belief, and the believer typically recognizes the applicability of that question, even when unable to give any convincing reasons on that occasion. The rational relations among beliefs (entailment, consistency, etc.) are recognized to be relevant to one’s entitlement to maintain one’s belief. In these ways we hold the believer responsible for her beliefs in ways that we do not typically hold the person responsible for the condition of her heart. We do not ask very young children and nonhuman animals for their reasons for believing something, and yet it does seem to many philosophers to be central to the very notion of belief in mature humans that believing something opens one to the norms of justification, and the responsibility for conforming to those norms. We ask 5 Compare Daniel Stoljar’s discussion (this volume) of the distinction between “explanationseeking” and “evidence-seeking” versions of the question “How do you know?”

216 Theories of Introspection the believer herself for her reasons, and the believer recognizes that she is indeed the person to ask, that the request for reasons is properly addressed to her. We do not address similar questions to the person about the condition of her heart. We do not have the same expectation that she will be the person in a special position to know about its condition and address the question of its good functioning. If this much is true, it raises several questions. One, what are we presupposing about the believer’s knowledge of her belief and its justification when we address such questions to her specifically? Two, what are we presupposing about her agency with respect to her belief when we hold her responsible in these ways, different from any responsibility she may have for the condition of her heart? And three, how does this system of reason-asking and reason-giving among mature believers relate to the capacities for belief among young children and nonhuman animals? That is, if we grant that such creatures do have beliefs, and yet do not have the concepts and capacities that would enable them to engage in the system of reason-asking and reason-giving proper to mature believers, then what is left to the claim that it is somehow internal to belief among mature believers that the system of reason-asking and reason-giving belongs to it, and that the role of justification is recognized as such by the believers themselves? In favor of retaining some version of the idea that the asking and giving of reasons belongs to the nature of belief and other attitudes themselves, I will just say the following. While some philosophers have gone so far as to doubt that the concept of belief can apply to animals and to children before they are language users, no one would want to deny that they are both capable of action in a perfectly ordinary sense. But at the same time, it seems we also do not want to say that having reasons for what one does is something only added on at a later stage and does not belong to the idea of action itself. Actions are purposive and goal-oriented. Both the child and the adult may have reasons for reaching across the table toward a glass of milk. The ability of the person to tell us what she is doing and why she is doing it is something that develops later, as part of the growth of various capacities and the initiation into various forms of responsibility. But the fact that rational assessment applies to the action, and that later these forms of assessment can be posed as questions that we direct to the person herself, asking her just what she is up to and why, is all a development of the same idea of action. These first-person capacities and responsibilities are no more extraneous to the idea of action than is the related capacity of a speaker to tell us what she means by something she said. At an early stage of the ability to talk, the child will not be expected to tell us what she means by her words. That is also something that develops later. But it is surely internal to the development of the child’s very capacity as a speaker that eventually she is understood, by herself and others, to be in a special position to tell us what she

Self-Knowledge, ‘Transparency’, and the Forms of Activity 217 means, what she is talking about, and will not count as a speaker absent any such capacity. Here again, we need not suppose that a speaker has unbounded authority over the question of what she is saying, or that allowance cannot be made for one form of semantic externalism or another, but the fact that we can and do ask the speaker to clarify what she is saying is surely part of the very notion of saying something. In all these cases then, it will only in very special circumstances makes sense to say “Why ask me?” in response to a question about what one is doing, or what one is saying, or what one believes and why.6 If the second, normative version of the ‘why?’ question is also, like the first one, meant to shed light on how something comes to be, then it may be asked why it is that we take the person herself to be particularly relevant as the person to ask here. Why isn’t another reasonably well-informed person an equally good or better source of information on this topic? And this question suggests the possibility that the reason we ask the person herself is closely related to the fact that the normative ‘why?’ question contains within it both a question about a form of normative assessment and a question about how something comes to be or maintains itself (the belief, the action). The condition of a person’s heart can be normatively assessed by a doctor, but that condition is what it is quite independently of the person’s cognitive relation to it, and it is not the person herself who is the locus of that assessment. By contrast, a person who believes it will rain, or who hopes it will not, is not just in a condition that can be normatively assessed, but is herself engaged in her own forms of normative assessment, and the believing or the hoping themselves are what they are in virtue of the person’s overall sense of what would support or undermine them as attitudes. When the belief or the action is judged or found wanting, this is an estimation of the person, and not simply of some condition she is in. If attitudes such as these can themselves be seen as forms of normative engagement on the part of the person, then the question ‘why?’ is applied to them in a particular form. Part of what is meant in calling the believing itself a form of normative assessment is simply that to believe p is to take p to be reasonable, believable,

6 This part of the story clearly presumes a creature not only capable of attitudes such as belief, but with the concepts of these attitudes. At an earlier stage of development, the child may be what Fred Dretske (this volume) calls an “unwitting authority” with respect to her beliefs.That is, she is aware of what she thinks, and she is the source for any claims about what she thinks, but, lacking the concept of belief, she cannot be said to know that she believes what she believes. There is undoubtedly a complex story to tell about the difference made to the character of the child’s belief, and the possibilities for self-knowledge of her belief, when she comes to acquire the concept of belief itself. I take it that acquiring the concept will go hand in hand with such things as being asked “When you ran to the door, did you think Daddy was home?” “Why did you think so?” etc., and hence with the primitive forms of reason-asking and reason-giving.

218 Theories of Introspection and in more articulate contexts, defensible or justifiable. And although we do not choose our beliefs and do not perform them like actions, this relation to forms of normative commitment is a matter of common form between beliefs and actions: to believe p is to take p to be believable and open oneself to the question “why believe that?” and to do something is to take the action to be worth doing in some way, and thus to open oneself to the question “why are you doing that?” And in both cases the person takes the answer to the normative ‘why?’ question to be directly relevant to the existence or continuation of the belief or action in question. With this much in place, we can see it as making a certain sense to see the person herself as the one to ask when we want to know why she believes this or hopes for that, and why she will indeed recognize herself as the right person to ask, and might indeed insist that treating her relation to this question as something in principle dispensable, or at best one among indifferently many equally good sources of information, would be to fail to take her seriously in a fundamental way. We have special reason to ask the person herself both what she is doing and why, or what she believes about something and why, because it is possible to see both actions and attitudes as themselves responses to questions of one form or another,7 or as ways of resolving oneself. A belief is the answer to the question about what the evidence points to, or what best explains some happening; and an action expresses one’s resolution with respect to the question what is to be done in a certain situation, or in response to a certain problem. As answers or rational responses, beliefs and actions invite the normative question ‘why?’ and assume the responsibility of the person to be able to speak to that question. By contrast, another internal condition of mine, like the condition of my heart or my sensation of pain, belongs to a different category and is not a possible answer to a question, or the possible conclusion of some line of reasoning (practical or theoretical). I may be responsible in one way or another for either the condition of my heart or my sensation of pain, but that is something purely external to the heart or the sensation itself, whereas on this view actions and attitudes are modes of resolving oneself, and hence involve forms of responsibility. I take these considerations to amount to a reason to think that the ordinary person’s ability to say what she is doing or what she thinks about something, and to know this without having to make the kind of observations of herself that she would if the question were about someone else’s belief or action, is related to the fact that she bears a certain responsibility for her belief and action, fundamentally different from the responsibility a person may have for the sensations she finds herself with. In the case of action, we ask the person herself what she is doing because 7 Pamela Hieronymi has stressed this formulation in recent work, although I suspect my understanding of the connection differs from hers.

Self-Knowledge, ‘Transparency’, and the Forms of Activity 219 we ordinarily take her to know, and we take her to know what she is doing because we take her to know why she is doing it, and we take her to know what she is doing and why because we expect this of her, it is her business to know. This brings a certain notion of agency into the picture of self-knowledge in that on this picture the non-observational character of self-knowledge with respect to actions and attitudes is tied to their being expressions of the rational, active side of one’s nature. Hence, the relevant notion of ‘activity’ belongs to the category (attitudes, or actions themselves) as distinguished from another category (sensation or bodily condition), rather than to one’s relation to a particular item in that category. Both a sensation of pain or a belief about my chances of winning the next hand of blackjack can be controlled or manipulated by me in various ways, and with various degrees of success. This notion of ‘control’ applies just as well to my relation to the perfectly inert objects in my immediate environment and is not relevant to the notion of agency being appealed to with respect to my doing and believing. Rather, if there is anything to this difference in category, then believing, intending, and hoping are themselves forms of activity, or expressions of the person’s active nature. A person’s beliefs are not chosen by her, nor are they typically “controlled” by her. Rather, what we call a person’s beliefs are the precipitate of her ongoing rational activity. It is only derivatively that a person is ‘active’ with respect to a particular attitude itself. In the normal case, I find myself wanting to learn Russian, or suspecting that my relatives will not be visiting for the summer after all. Such attitudes do not emerge out of nowhere, of course, but rather become mine in the course of my ongoing thinking and acting, and are not aimed at as states to put myself in. I do not aim at acquiring some particular attitude, and its rationality is not expressed by my singling it out for control or manipulation. But for all that, my wanting, suspecting, and caring about something are expressions of my active nature, to which some form of the normative ‘why?’ question naturally applies, along with my taking myself to be the person who is answerable for why I do the things or believe the things that I do.8

3. Self-Knowledge and Settled Beliefs Various recent writers have taken the ‘transparency’ of belief to be part of the explanation for the person’s ordinary ability to say what she believes about something without having to base what she says on empirical observation of herself, but have found the appeal to rational agency to be misplaced. For some, this appeal has been 8 For more on the distinction between exerting external control over one’s attitudes and assuming rational responsibility for them, see Moran 2002.

220 Theories of Introspection thought to falter on an ambiguity between how I know what I already believe about something and how I know what I believe about it now, upon considering the question. Thus in a recent article, Nishi Shah and David Velleman say the following: The question “Do I believe that p?” can mean either “Do I already believe that p (i.e., antecedently to considering this question)?” or “Do I now believe that P (i.e., now that I am answering the question)?” . . .. . . Now, either of these questions can give way to the question whether p. If the question is whether I already believe that p, one can assay the relevant state of mind by posing the question whether p and seeing what one is spontaneously inclined to answer. In this procedure, the question whether p serves as a stimulus applied to oneself for the empirical purpose of eliciting a response. One comes to know what one already thinks by seeing what one says—that is, what one says in response to the question whether p. But the procedure requires one to refrain from any reasoning as to whether p, since that reasoning might alter the state of mind that one is trying to assay. Hence asking oneself whether p must be a brute stimulus in this case rather than an invitation to reasoning.9 This is said in the context of seeking to vindicate a notion of the transparency of belief, but it is not immediately clear just what sort of problem they see here, or

In “When We Are Ourselves: The Active and the Passive,” Joseph Raz makes the case for seeing belief and related attitudes as part of our active nature in terms that are congruent with the account given here. I encountered his paper after completing this chapter, but the clarity of his account is worth quoting at length, particularly for the distinction between agency as responsiveness to reasons and the agency involved in explicitly arriving at a new belief. “We are active when our mental life displays sensitivity to reasons, and we are passive when such mental events occur in a way which is not sensitive to reasons; or at least this is part of what accounts for the distinction. In these terms beliefs are—pathological cases excepted—on the active side of our mental life.This does not mean that we form beliefs only as a result of deliberation. We may form them because— with our senses—we perceive how things are, or because through subconscious processes we come to have or to form them. All of this is consistent with the active character of believing, or having beliefs. . . . Even when we form perceptual beliefs, or when we come to have certain beliefs without being aware of the fact, the beliefs are responsive to reason. This responsiveness is manifested in two ways. First, in that unconscious processes of belief formation, just like explicit deliberation, depend on absence of awareness of reasons against the belief, and—normally—on reasons for it.When it seems to me that I see a cat I—without deliberation—believe that there is a cat there. But if I believe that I am in a magic show, then I do not form that belief. Second, when I deliberate and come to the view that the evidence is that a proposition that I believe is false the very process of coming to that conclusion is also a process of ceasing to believe it. By their responsiveness to reasons believing and beliefs belong to the active side of the active/passive divide” (1997, 218). 9 Shah and Velleman 2005, 16. A related objection is made in the article by Alex Byrne (2005) discussed in the next section, as well as in Gertler (forthcoming).

Self-Knowledge, ‘Transparency’, and the Forms of Activity

221

how the kind of insulation from rational agency they have in mind could still deliver an answer about what I already believe about something. Does engaging my rational capacities corrupt the process of reporting on what I already believe about something? As a first approach, we could adapt an example from Sydney Shoemaker and suppose that I am asked who I believe was the President of the Confederacy during the American Civil War (Shoemaker 2003). It may happen that in responding to the question I start to say “Robert E. Lee” and then correct myself and say “Jefferson Davis.” In this way, I failed to treat the question as a “brute stimulus,” and in correcting myself I engaged my rational capacities, but none of this provides a reason for thinking that I have thereby produced a new belief rather than reported what I believed all along. Surely it may be the case in an example like this that I did in fact believe all along that it was Jefferson Davis, but blurted out the wrong answer and corrected myself, all the while being faithful to what I already believed. The engagement of my rational capacities in delivering the answer need not be seen as substituting a new belief for what I already believed. On the other hand, there are difficulties in seeing how my response to such a question about what I believe could be a response to a “brute stimulus” as imagined here, and still be seen as reporting on a belief of any kind. To begin with, let us recall that with respect to any belief of mine, it counts as a belief insofar as I take it to be true (this, of course, is what makes some sort of appeal to ‘transparency’ seem attractive). If I relate to my “stored” belief as something I take to be true, it will be hard to see how I can see my relation to it, however spontaneous, as insulated from the engagement of my rational capacities for determining what is true or false. It cannot, for instance, be seen by me (or my auditors, if they are the ones applying the stimulus) as simply some name that is produced upon receiving the stimulus, for it has to represent what I take to be true, as an answer to the question asked. Hence, there is a considerable background involving my rational agency that has to be assumed for my response to the stimulus to count as my spontaneous answer to the question. Consider the fact that there are countless words and names that may be floating around in one’s mind, and any one of them might be what comes out when the stimulus is applied. None of them will count as indicating one’s belief about the matter unless, minimally, one can recognize the word as a name, and not something else, and recognize it as the name of a person, and recognized the name of this person to be relevant somehow to the stimulus such that it can serve as indicating what one’s belief is about this matter. Simply hearing oneself coming out with something in response to a brute stimulus will provide no more reason for thinking this represents one’s belief about something than if one were to sneeze in response to the stimulus. Rather, for my response to the stimulus to be seen as telling us what I already believe about the question, I have to relate myself in various ways to the

222 Theories of Introspection name I come out with, and not just hear myself say it. I must, at a minimum, understand the words I am saying, and understand them as responding to a question whose meaning I understand. And ‘response’ here must mean something like ‘replying to the question’, for in a broader, more neutral sense of ‘responding’ we would need an additional reason to think that this ‘response’ bears any relation to my beliefs about anything (which is what the stimulus is supposed to deliver). For any mental content, word association, or exclamation that may be produced by the stimulus, it will count as relevant to the question of my belief only if I am relating to that content or word as representing my belief about the matter. That however, will mean engaging with my rational capacities in a way that involves my reflecting on the facts of the matter, which was supposed to be excluded from the process because it was thought to “contaminate” it.10 But if that were so, then it is hard to see how a person could ever perform the ordinary task of telling us what they already believe about something. What is meant by the “activity of reason” here need not be explicit, nor need it involve the production of a new belief. I may hear the question “Where was Balzac married?” and come out spontaneously with the name “Berditchev,” and when asked why, perhaps I cannot say much more than “I must have read this somewhere” (see Dennett 1981, ch. 16). But to know that much about my response is already to know a great deal, is already for me to have classified my ‘response’ as the name of a place, in answer to a question about a person, and in connection with an event that left some written record I could have encountered. The confidence with which I spontaneously come out with the answer is surely dependent on such things as that I understand the statement proposed to me, that the proposition in question makes sense to me, and that the possibility it presents seems perfectly plausible to me, even if I can presently see no special reason to think it true. (Compare: the answer I come out with is “on the moon.”) All of this and more is part of my apprehension of the rational environment of this proposition. There is no isolating my response to the question from all of this, and still have me responding to a question. It is only when we take for granted that this background is in place that we then confront the scenario of my replying to the question, with perhaps nothing more to go on than a feeling of familiarity, or the sense that I must have heard this before, but by then the work of reason has already prepared the place for my answer. We can, of course, isolate my response from reason if I do not understand the language in which the question is posed, or if the response I come out with makes no sense to me, something I just find myself saying, or if I come out with it as a sheer guess, but 10

“As we pointed out, one cannot engage in reasoning aimed at answering the question whether p if one wants to find out what one already believes, because such reasoning would contaminate the result by possibly altering the state that one is trying to assay” (Shah and Velleman 2005, 17).

Self-Knowledge, ‘Transparency’, and the Forms of Activity

223

in none of those cases will we think that this response expresses my belief about anything. And with respect to other things I already believe, which are more integrated in the rest of my life and thought, things are more complex, and it is even harder to imagine what could be meant by isolating my response from the influence of reason. That is, if the question about what I already believe is something like: “Do you believe there are people living in Phoenix?” or “Do you believe that you can buy food with money?” then my spontaneous answer will be ‘yes’, not because I have reasoned my way to this conclusion, but because so much else of what I believe would have to be upended if this were not true, and I would have no idea how to begin such a revision of my beliefs. I do not have to think about it, but not because my answer is insulated from the influence of reason, but precisely because my answer is so fully integrated with the rest of my beliefs and rational capacities.11 What seems to be imagined in the scenario from Shah and Velleman is a situation of wanting to know what someone thinks about something, but prior to having that person consider the question itself. If I am playing poker with someone, I may want to know whether he thinks I am bluffing, but without actually raising that question with him, for that would risk alerting him to a possibility that I do not want him to consider if he is not doing so already. In this sort of case, I might do various indirect things to elicit information on this point, without raising the question itself. This might be thought of as a “brute stimulus” in the relevant sense, but it will importantly not involve my poker partner responding to a question as a “brute stimulus.” Now, in my own case, it is much harder to imagine what would count as my “wanting to know what I think about some possibility, but without my considering the possibility itself.” In the two-person case, I can raise the question of what this other person believes about p, without that person considering the question of the truth of p itself. In my own reflections about him, I can speculate about what may or may not be going on there in this other mind, knowing that my reflections are strictly mine, and not part of what constitutes the state of mind I am speculating about. But with only one person on the scene, this is not a real possibility. I cannot pose the question to myself of whether I believe that p without raising the question of the truth of p, for there is only one mind under consideration here, inquiring about itself. Naturally, this does not mean that I cannot, in certain circumstances, seek to “assay” what I really believe about something, in a way that 11

See Wittgenstein 1956, § 478: What kind of reason have I to assume that my finger will feel a resistance when it touches the table? What kind of reason to believe that it will hurt if this pencil pierces my hand?—When I ask this, a hundred reasons present themselves, each drowning the voice of the others. “But I have experienced it myself innumerable times, and as often heard of similar experiences; if it were not so, it would. . . . ”; etc.

224 Theories of Introspection brackets the question of the truth of what I seem to believe. In various circumstances, a person can indeed take such an ‘outsider’s’ perspective on her own belief, even though the result may be an inherently unstable one (“Well, I know that the plane is safe, but clearly I am also in the grip of a fear that it is not safe.”) But even this possibility is quite different from the idea that I could raise the question of whether I really think the plane is safe without considering the question of whether the plane is safe. I have to understand what question I am asking myself (the way my poker partner does not have to understand my question about him); or if I am applying a ‘stimulus’ to myself I have to know how I understand the relation between the stimulus I am applying and the question I am seeking to answer. And I cannot understand either of these things without considering the content of the state of mind I am inquiring into, which orients me with respect to the question of its truth, reasonability, or comprehensibility. Inquiring into one’s own mind about what one already believes about something cannot be insulated from one’s rational agency in the way that it can be with respect to the mind of another person, but nor does this involvement mean that what I already believe is elusively out of my reach, because always threatened with replacement by something else once my reason is engaged. Unlike my relation to my partner in poker, in the first-person case, inquiry into the mind has to be intelligible to the very mind being inquired into, and hence involves the engagement of reason on the part of both the inquirer and the mind inquired into, for they are one.

4. Transparency and Rules for Belief In a recent paper, Alex Byrne (2005) seeks in a different way to account for Transparency, while avoiding appeal to notions of activity or rational agency in the explanation of what makes appeal to Transparency possible. The paper presents a rethinking of several issues concerning self-knowledge, and its general aim is to vindicate what Shoemaker criticizes under the name of the Broad Perceptual Model of self-knowledge. This is to be distinguished from any appeal to “inner sense” (although Byrne has his sympathies here12) and comes down to the two claims that (1) our detection of our mental states is based on some sort of causal mechanism, and (2) that our mental states themselves obtain independently of our access to them (Byrne 2005, 86). Shoemaker’s case against the “broad perceptual 12

To be sure, much of the paper is devoted to showing how the case against “inner sense” turns out to be unconvincing, and hence counts as a defense of that view. But at the same time “inner sense” is described as an “extravagant” rather than “economical” account of self-knowledge, in that

Self-Knowledge, ‘Transparency’, and the Forms of Activity

225

model” depends on challenging (2) rather than (1), since Shoemaker (this volume) defends the idea of a “constitutive relation between believing something and believing that one believes it.” The constitutive relation between one’s mental life and one’s first-person access to it is defended by Shoemaker as a consequence of being in a certain belief state, combined with ordinary rationality.13 In this sense, then, although we do not infer or reason our way to knowledge of our beliefs, there is a constitutive connection between being a subject of belief in the first place and having one’s beliefs available to one, without the need for the kind of evidence one would need for knowledge of the beliefs of others. Byrne’s alternative claim is that the phenomena of self-knowledge can be understood as the result of our following a rule for belief formation, which he calls BEL. The phenomena to be accounted for include both privileged access, described as the idea that our “beliefs about our mental states acquired through the usual route are more likely to amount to knowledge than beliefs about others’ mental states” (2005, 80), and what he calls “peculiar access”; that is, the idea that one has a special method for learning of one’s own beliefs that cannot be applied to the beliefs of others (ibid., 81). The BEL rule is presented as a reconstruction of the Transparency condition, and an explanation of why it is legitimate for a believer to answer a question about her belief about something, by reflection not on herself, but on the topic of the belief in question. Hence, he presents the BEL rule as follows: BEL: If p, believe that you believe that p. This rule for belief is introduced by comparison with another rule, called DOORBELL, which states: DOORBELL: If the doorbell rings, believe that there is someone at the door. DOORBELL is what Byrne calls a “good rule” insofar as it “tends to produce knowledge about one’s visitors” (ibid., 94), and this goodness will depend, naturally, on various empirical conditions obtaining in the environment of the person following this rule. It can lead one astray, and hence is not failsafe, but may be reliable enough in practice to count as a rule the following of which results in knowledge. BEL, however, has epistemic virtues superior to DOORBELL in that it is “self-verifying”: if it is followed,

it requires appeal to an additional capacity or mechanism, beyond what is already needed for our general capacity for rationality (Byrne 2005, 92). The account Byrne later goes on to defend, in terms of the BEL rule, is recommended for being economical rather than extravagant (ibid., 99), hence I take it that Byrne should be understood as defending a version of the “broad perceptual model,” but not “inner sense.” “The account is not a version of the inner-sense theory” (ibid., 9). 13 “[B]elieving that one believes that p can be just believing that p plus having a certain level of rationality, intelligence, and so on” (Shoemaker 1994, 244, quoted by Byrne on 89).

226 Theories of Introspection the resulting second-order belief will be true (p. 96). This has the nice feature of also capturing “peculiar access,” since BEL will only be self-verifying when applied to one’s own beliefs. The variant of BEL “If p, believe that Fred believes that p” is not a good rule. Finally, BEL is an especially “safe” rule in that the resulting secondorder belief will be true, even if one tries but does not succeed in following it. As the rule is formulated, actually applying it requires the obtaining of some fact ‘p’, hence in the absence of that fact the BEL rule is not in fact being followed. However, BEL has the special virtue that even trying to follow it will result in a true second-order belief, since if I mistakenly take ‘p’ to obtain, and then seek to follow BEL on the basis of this mistake, I still end up with the true second-order belief that I believe that p. BEL is described as a rule that we follow, and the following of this rule is described as form of reasoning (ibid., 94). And indeed, the following of this rule is described by Byrne as no more problematic than the following of less exotic epistemic rules, such as DOORBELL.14 At the same time, Byrne does acknowledge BEL as an unusual epistemic rule since, unlike DOORBELL, it recommends a transition from the apprehension of a fact to a belief concerning a logically independent fact, without the first fact being evidence for what is believed in the resulting belief. One way to display the difference between the two rules is by noting that, despite the imperative form in which it is presented, DOORBELL can also be formulated as simply a relation between two contents, rather than as instruction for belief formation: DOORBELL: If the doorbell rings, there is probably someone at the door. And indeed, upon hearing the recommendation of the original DOORBELL rule, it would be natural to assume that the only reason for following it was that there was just such a relation between the two contents, such that the fact of the ringing supported or was evidence for the fact of someone’s being at the door. Absent such a connection between the contents in the two parts of the rule, it would be difficult to embark upon complying with it, unless one could somehow install beliefs in oneself, exerting a kind of purely pragmatic agency with respect to one’s beliefs. As mentioned, Byrne is well aware of this difference between BEL and DOORBELL, since in a sense this difference simply comes to the original puzzle of Transparency: How can it be legitimate to answer a question about a particular person’s beliefs, by appeal not to facts about that person, but to facts relating to the content of the belief? The parallel with the revised DOORBELL rule, in terms of a relation between two contents, would be:

14 “Given that we follow rules like DOORBELL, it should not be in dispute that we can follow BEL” (ibid., 96).

Self-Knowledge, ‘Transparency’, and the Forms of Activity

227

BEL: If p, then I probably believe that p. As Byrne points out, however, “this is a bad rule: that p is the case does not even make it likely that one believes that it is the case” (ibid., 95). How then are we to understand following the original BEL rule as a form of reasoning when it does not present a rational relation of support between the two contents? If BEL is a rule we can be said to follow, and the result of following it is the formation of a particular belief, then there must be some answer to the question of the person’s entitlement to make the transition in thought that is being recommended by the rule.15 Whatever “following the rule” comes to, it cannot be the exercise of a kind of agency unrelated to the truth-centered demands of belief (as it would be for instance with the recommendation, “Believe that p, because then you will stop worrying so much that not-p”), for the account is meant to be an account of self-knowledge. The fact that BEL is “safe” in the sense of being unlikely to deliver false beliefs would be a recommendation of it as a description of a good mechanism of belief formation, but the person seeking to follow the rule, the way she follows DOORBELL, will still be in need of some reason relating the two contents. Otherwise the rule is reduced to saying, as it were, “Act upon yourself, do whatever it takes to produce the one belief on the basis of the other.” On externalist epistemic assumptions, one may be entitled to some belief without having reasoned one’s way to it, and without now being able to provide justification for it, so long as there is in fact a reliable connection between one’s belief and the fact in question. A reliable mechanism for the production of true beliefs need not represent itself in terms of a rational connection between belief contents, but may instead operate without any understanding of it on the part of the person herself. BEL, on the other hand, represents itself as a rule for belief, expressed to the believer as an imperative or a recommendation. As such, BEL is not just the description of a good disposition for someone to have, but rather requires the believer to do something in response to it, and in accordance with it. This much is needed by the parallel with DOORBELL, and the understanding of BEL as itself a form of reasoning. Following a rule for belief, however, is not simply undertaking to produce a belief in oneself by whatever means necessary, but requires from the rule follower some understanding of, and an endorsement of, the rational connection between the contents mentioned in the rule. DOORBELL does provide this, given the relation between the contents as represented in DOORBELL, but BEL itself does not. 15

I speak of “transitions” of thought here so as not to pre-judge the question whether we should understand BEL as an inference, as a rational connection of another sort, or as a kind of “blind” rule that nonetheless produces “good” results (“safe” beliefs). Shoemaker (this volume) questions whether following the BEL rule can be understood as a form of reasoning.

228 Theories of Introspection Why should there be such a difference in apparent “goodness” between the original BEL rule, an imperative or recommendation addressed to some “you,” and the revised rule BEL relating two contents, especially when the corresponding two versions of DOORBELL do not display this difference in goodness, and indeed seem to support each other? And can the imperative form retain any force when divorced from the relations of support among belief contents described in rules like DOORBELL? Looking at BEL as an account of Transparency, and as a proposed answer to the problem of “two topics” (e.g., the weather, my belief), we might look at the issue in terms of three possible candidates for belief and their requirements. 1. Considering p as a candidate for belief, I require evidence for p or truth-centered reasons of some sort if I am to believe p. 2. Considering the candidate “Jones believes p,” I likewise require evidence or other truth-centered reasons concerning Jones, since he is the subject of the content in question. 3. However, considering “I believe that p” as a candidate for my own belief; that is, the question of attributing the belief p to myself, I do not appeal to evidence for the content “I believe that P,” where that is taken to refer to the beliefs of a particular person, as in (2). Rather, I appeal to the sorts of reasons mentioned in (1), reasons in favor of that content, the one that is embedded in (3), and which does not mention any person. Statement (3) is a reconstruction of Transparency as I have been understanding it. As such it raises the question of what could make legitimate the appeal to reasons relating to the embedded content (P), rather than to reasons relating to the ostensible content of the attribution itself (‘I believe that P’). On this way of looking at it, the beliefs governed by Transparency are not altogether independent of the appeal to evidence or other truth-centered reasons, but the question is what makes legitimate the exclusive appeal to reasons relating to the embedded content (1) rather than reasons concerning the apparent content of a psychological attribution (2). What breaks the apparent parallel between (2) and (3), such that (3) is answered as though it were a version of (1)? As with the original Transparency condition, the appeal to the BEL rule also relates the question of the content p (1) to the question of the self-attribution (3), and hence incurs a similar burden to explain how it can be legitimate to make the transition from ‘p’ to ‘I believe that p’. The explanation adverts to the self-verifying character of following BEL, but that will only be a good explanation if we have a better understanding of what “following” it comes to, and the conditions under which doing so is possible. In particular it bears explaining how the imperative form of BEL could make sense to someone seeking to follow it when it is understood to

Self-Knowledge, ‘Transparency’, and the Forms of Activity

229

lack the support of the corresponding version of the rule (BEL), which describes relations among the contents of beliefs (“If p, then X believes that p”). If we distance ourselves from the imperative form of the rule for a moment, we can describe the believer who is guided by the BEL rule as someone who makes a transition from the apprehension of some fact p (“It’s raining out”) to a belief about herself. This “belief about herself ” is the product of following the BEL rule, which the imperative form presents as addressed to “you” (“Believe that you believe that p”). In taking it out of the imperative form and representing the transition in thought that the rule recommends, we need some substitute for “you,” and it will be crucial that the substitution retains the status of BEL as a “good rule.” If, for instance, the description of the transition takes us from “p” to “Jones believes that p,” this will be a bad rule, even if the person making this transition is indeed Jones. For he may not recognize himself as “Jones,” in which case the BEL rule would be taking him from the apprehension of some fact “p” to a belief about the beliefs of some person named Jones. The problem of “two topics” would re-emerge for the understanding of Transparency, the problem of understanding what could be legitimate in the transition from a thought about, e.g., the weather, to a thought about some person’s beliefs. Hence, the transition described in the BEL rule is subject to certain conditions, familiar from the philosophical discussion of the first person. The transition described in BEL will only be legitimate when the person following it is identical to the person whose beliefs are mentioned in the resulting second-order belief. And this identity must be recognized by the person making this transition, and hence the name we substitute for “you” when we describe this transition outside the imperative form of the rule must be one under which the person in question could not fail to recognize herself. If there were a possibility of misidentification or failure to identify oneself with the person whose beliefs are the topic of the resultant second-order belief, the application of the BEL rule would be as illegitimate as the transition from some proposition p to the beliefs of some arbitrary person. The ‘I’, then, is what lies behind the ‘you’ who is the implicit addressee of BEL in its imperative form. And in particular the “subject use” of ‘I’ is crucial to understanding how it is that, while the attribution of belief to a person under some name, description, or demonstrative must appeal to evidence concerning that person (as in (2)), the first-person discourse of belief appeals to reasons concerning the content of the belief rather than evidence or identifying information about the believer.16 When the BEL rule instructs someone to “believe that you believe that p,” this “you” must not be in need of identifying information anymore that the subject use of “I” 16 For the initiating discussions of the idea of this use of ‘I’ proceeding independently of “identifying information” about the person, see Wittgenstein 1958, Shoemaker 1968, and Evans 1982, ch. 7.

230 Theories of Introspection is, if the following of the rule is to be safe or self-verifying. If identifying information were required to pick out the right person named in the instruction “believe that you believe that p,” or if evidence about that person were needed to determine what that person’s beliefs were, then forming the belief about the beliefs of this person “you” would be in need of the same kind of support, and subject to the same semantic and epistemic risks, as would the corresponding belief about the beliefs of Jones. Under these conditions BEL would be “bad” in the same way noted by Byrne in connection with the “neutral schematic” version of BEL: the obtaining of some fact p would be treated as a reason for concluding that some person believes it (ibid., 95). So a fuller picture of the conditions under which BEL would be a good rule are that (a) it is only knowledge of one’s own beliefs, and not those of another person, that following BEL will make possible, and (b) that in this knowledge I recognize myself as the person whose beliefs are the topic of this knowledge, and (c) that my recognition of myself as the person whose belief is known in this way not be based on identification information of any sort. One situation in which a person may indeed need to avail herself of identification information regarding some attitude is when she takes a belief or a fear of hers to be evident in her behavior but otherwise inaccessible to her because it is not responsive to her sense of the reasons that would support the belief or fear. If I have reason to believe that what sustains my actual attitude regarding someone’s competence or trustworthiness is some fear or prejudice irrelevant to the question of his actual competence or trustworthiness, then I will not take myself to be entitled to make the transition described in the Transparency condition or the BEL rule. If I think that my actual belief about him (or people “like him”) is controlled by prejudices whose operations are beyond my awareness and which persist independently of my grasp of the reasons for or against p, then I will not take myself to be entitled to answer the question “Do you believe that p?” by reflection on the question ‘p?’ Likewise, I will not be entitled to follow the BEL rule, which tells me that if “He is just as competent as the others,” I may help myself to the belief that I believe he is just as competent. The difference between the topics of the two questions will re-emerge. I will recognize that I may need to give different answers to the two questions. I may know about this tendency of mine, and know that the beliefs that are its product are resistant to change, and may not be reflected in my overt, sincere declarations when considering the content itself.17

17 See Moran 2003, 407–408, for more on different ways of imagining such failures of transparency.

Self-Knowledge, ‘Transparency’, and the Forms of Activity

231

Thus, the appeal to either Transparency or the BEL rule has various conditions, which matter to the understanding of the kind of rationality that is assumed in making the transition from ‘p’ to “I believe that p.” For a person to be entitled to make this transition, he must recognize the identity between himself and the person whose belief is the topic of “I believe that p,” and he must take himself to be able to speak for the beliefs of that person, in the sense that he presents himself as the very person whose sense of the reasons in favor of ‘p’ are expressed in the statement “I believe that p.” Transparency fails if the statement I arrive at, “I believe that p,” is delivered in a purely attributive mode, for that would require evidence about me, which Transparency does not provide.18 What Transparency does purport to provide is a connection between my reflection on a particular content ‘p’ and the reasons in favor of it, and the answer to the question of what my actual belief is. That connection presumes that my belief about some matter ‘p’ is determined by my sense of the reasons in favor of ‘p’, and not by forces independent of those reasons. This is not a matter of constructing a new belief, but of seeing my past settled beliefs as available to me through Transparency insofar as they make sense to me in the light of reflection on their contents. This requirement does not mean that the declaration “I believe that p” cannot be the expression of what one already believes,19 but that the statement of belief must be delivered in the mode of endorsing the content of the belief, and not as the attribution of belief (to oneself, to another person) in a way that leaves open the question of endorsement. The idea of endorsement points to a form of rational agency that both Byrne, and Shah and Velleman, find misplaced in the account of Transparency, but in distancing themselves from it, they each in their different ways picture the self as exercising the “wrong kind of agency” with regard to its own beliefs, either by seeking to elicit one’s own beliefs through a kind of stimulus insulated from the exercise of reason, or by proposing the application of a rule of belief formation that could only be applied “externally,” since we lack a rational connection between the two belief contents. The centrality of the first person to the understanding of transparency points to the fact that conforming to it should not be pictured as producing a new state in oneself so much as committing oneself on a certain matter.20 The transition described in Transparency is not an inference from evidence about a particular person, but rather something more like a general

18

On “attributive” see Moran 2001, ch. 3.3. Byrne (2005, 85). Gertler (2011) makes a similar criticism, but unlike Byrne, she makes it in the course of denying the appeal to transparency, rather than as a reconstruction of it. 20 Boyle (forthcoming) argues for seeing the notion of agency relevant to self-knowledge as a form of “self-activity” and against what he calls the “process conception” of deliberation. 19

232 Theories of Introspection presupposition of rational thought, to the effect that, from the first-person point of view, I must take what I believe about something to be the expression of my sense of the reasons relating to the content of that belief. As we have seen, it is a presupposition that may lapse in cases of compromised rationality, but at the same time, it is hard to see what could be more basic to rationality than the idea that I take the question of what I believe about X to be determinable by my reflection on X itself. This is why Transparency is more like a requirement than like a permission or an optional means for determining one’s attitudes, for if I could not learn what I (really) think about X by reflection on X itself, that would mean that my belief about X was out of reach of my reflection on the facts concerning X itself. If we assume a subject with mastery of the first-person pronoun, and its independence of identifying information, it will only be in conditions of compromised rationality that a person could believe that it is raining, possess the concept of belief, and yet be unable to know her belief through reflection on the weather.21

5. The Activity and Passivity of Belief Part of the sense of ‘rational agency’ in this domain lies in the fact that, with respect to something I already believe, if I am considering it as something I am in fact committed to, then at a minimum I should be able to see my having this belief as a potential reason for someone else to believe the same thing. My confidence in some belief of mine may vary, but this will vary in accordance with whether what I take myself to discover is a belief rather than a hunch or a mere guess. Hence, I may not be prepared to assert it, but if it is a belief at all I must take the fact that I think so to be something that could be part of someone else’s reason for believing the same thing. Hence, while I may not be ‘active’ with respect to it in the sense that it is a conclusion I have arrived at just now, seeing it as a belief of mine means such things as knowing what would counter it, knowing what sorts of reasons are relevant to it, my being prepared to increase or decrease my credence in it depending on other things I learn.

21 Compare Byrne, in summarizing his account as a version of the Broad Perceptual Model: “Thus there is an appropriate causal mechanism.The state detected is independent of its detection. The subject might not have followed BEL, in which case the first-order belief would have been present without the second-order belief. What’s more (we may fairly supposed) someone might believe that it’s raining, possess the concept of belief, and yet not even have the capacity to follow BEL” (2005, 98).

Self-Knowledge, ‘Transparency’, and the Forms of Activity 233 More importantly, even for perceptual beliefs that are in some sense forced on me, these still differ from mere feelings (sensations) in that a perceptual belief is still something I am prepared to assert and defend.22 Normally, I take my perceptual belief to put me in a position to claim something, something of potential epistemic import for others, and not just myself. Here we already see something of my agency involved in a way that we do not in considering my headache. In some sense I may be said to be passive with respect to both of these, but in the case of my perceptual belief the sense of my having no choice in the matter is precisely a matter of my finding the reasons in favor of the belief to be unassailable. This is not a matter of something merely befalling me, like a headache. Rather, my ‘passivity’ here stems from the fact that, for me, to change my belief would require a reason, and a reason of a special sort (viz., a reason connected with truth), and hence I do not have what Pamela Hieronymi calls ‘discretion’ with respect to this change, as I do with respect to the question of changing my shoes (Hieronymi 2006). So, in that sense, my ‘agency’ is quite restricted here. But that restriction is just the reflection of the dominance of another kind of agency, that is, my responsiveness to epistemic reasons for my beliefs, taking them to be the basis for what I can assert. And in the case as we are imagining it here, within this restricted range of admissible reasons there is dominance with respect to a single conclusion: I have every reason to trust the visual evidence here and no reason to doubt it, hence I cannot help but believe there is, e.g., a tomato in front of me.This is not like being assailed by a stabbing pain (something not subject to justification in the first place), nor is it like being carried away by a compulsion to believe that I will win the next round of black jack because the stakes are so high. Rather, it is an instance of what is called being compelled by the weight of reasons in favor of some conclusion. I have no choice because change of belief requires reasons, because beliefs are things I am prepared to defend, and in this case the reasons are overwhelming in a particular direction. This ‘passivity’, then, is an expression of my rational agency and is utterly unlike my passivity or lack of choice with respect to how fast my hair grows. In this way, then, the ‘passivity of belief ’ is the reverse side of a person’s rational agency as a believer, for it is because one’s beliefs are the expression of one’s rational relation to the world that they cannot be simply “chosen.” If what I believe were not answerable to the ways the world is, then I could indeed treat my beliefs as states which I could seek to produce in myself for reasons unrelated to their truth. I could “control” them in a way that enjoys “discretion” over the kinds of considerations that will count for me as reasons in favor of bringing it about. Likewise, when I am

22

Here again, see Sosa (this volume), as well as the quotation from Raz in note 8.

234 Theories of Introspection relating to the beliefs of another person, I may also see them as states to be produced or removed for purely practical reasons of my own. In this sense I can be ‘active’ with respect to them the way I am when I seek to manipulate the objects in my environment. The fact that I am not active in this sense with respect to my own beliefs does not, however, mean that I am passive with respect to them the way I am passive in relation to a sudden pain or something falling on top of me. The very limitation in my manipulator’s or producer’s relation to my own attitudes is an expression of the fact that they are my cognitive and affective relation to the world, and hence demand from me reasons connected with the kinds of attitudes they are. A belief or a hope represents the world in a certain way, is something to which justification is internal, and which stands in logical relations to other attitudes. While there may be reasons for having a sensation, the sensation is not itself something reasonable or unreasonable. By contrast, the reasons that are primarily relevant to believing are not reasons for having the belief, but rather reasons in favor of the content of the thought, that is, reasons relating to its truth. The fact that in the case of attitudes there is a contrast between reasons in favor of having the attitude and reasons in favor of the content of the attitude itself is something that distinguishes them as a category from the category of sensations, for there is no corresponding contrast in the case of sensations. When it is said that we cannot adopt a belief “at will,” part of what this means is that one cannot adopt a belief on a whim, in response to a simple request from another person, or otherwise for reasons unrelated to its truth. Discretion of this sort represents a kind of freedom or liberty we are familiar with in other contexts, a form of freedom that appears to characterize our relation to our ordinary actions (e.g., changing one’s shoes), but does not appear to apply to our relation to our beliefs and other attitudes. Hence, it may be natural to conclude from this that the notions of agency and activity have no bearing on the nature of belief and other attitudes, that they are “forced upon us,” in much the same way as are the facts of gravity or the inert objects in one’s environment. Or it may suggest that if a form of agency did characterize a person’s relation to her beliefs, that could only be as a form of managing or manipulating a set of items within one’s purview. This is, of course, not only unrealistic as a picture of psychological life, but neglects the fact that the great majority of one’s beliefs and other attitudes manage to maintain themselves in a rough rational systematicity without any need for monitoring or intervention from the person as such. And hence it may seem natural to conclude from this that self-knowledge of attitudes like belief has no particular bearing on their rationality or on epistemic responsibility generally. I have tried to suggest room for a different picture of the relation between self-knowledge and agency, which shifts the picture of self-knowledge from that of monitoring an internal condition to the ordinary ability to say what I am doing and

Self-Knowledge, ‘Transparency’, and the Forms of Activity

235

why, or what I think about something and why. This also involves a difference in the kind of agency invoked here, a shift from agency as production, to agency as responsiveness to reason. In one sense raising my arm voluntarily is an expression of my active nature, and in a quite different sense what I count as a reason is an expression of my active nature, as well as how I take myself to be answerable for what I believe. With respect to the beliefs and attitudes of another person, I may exercise a kind of productive agency, aiming at specific results. Here my reasons for favoring the production of the belief as state may be fully separate from reasons in favor of what is believed, and may be as diverse as my various reasons for wanting something to happen. In the first-person perspective on belief, however, my primary relation is not to the fact of having some belief but rather the commitment to its truth and what that requires of me. Detaching my relation to a state of belief (mine or another’s) from the commitment to its truth is precisely what would allow for discretionary reasons in relation to its production. The first-person point of view presumes the absence of such separation, presumes the identity of the considerations in favor of the thing believed with the fact of one’s believing it. The absence of such separation characterizes the kind of rational agency involved, specifically that it is not a matter of acting upon oneself or taking oneself as an object, but rather of being resolved about some matter. It is this same absence of separation that characterizes the kind of self-knowledge in question, namely that it makes possible a form of ‘transparency’, the fact that a person can normally tell us what she thinks about some possibility by reflecting on that possibility itself.23

References Bilgrami,Akeel. 2006. Self-Knowledge and Resentment. Cambridge, Mass.: Harvard University Press. Boyle, Matthew. 2009. “Two Kinds of Self-Knowledge.” Philosophy and Phenomenological Research 78(1): 133–164. ———. Forthcoming. “On Making Up Your Mind and the Activity of Reason.”

23

In writing this chapter I have benefitted profoundly from conversations with Matthew Boyle, and from his seminar on “Self-Consciousness and Self-Knowledge” in Fall 2009. I had very helpful comments from and conversation with Endre Begby, as well as other participants at the conference “Self-Knowledge and Rational Agency,” sponsored by the Center for the Study of Mind in Nature, in Oslo, June 2010. I am grateful as well to Alex Byrne, Jonathan Vogel, Pamela Hieronymi, and Doug Lavin for conversations about these issues, and to the editors of this volume for very helpful comments on the final version, as well as the comments of two anonymous reviewers.

236 Theories of Introspection Burge, Tyler. 1998. “Reason and the First-Person.” In Knowing Our Own Minds, edited by C. Wright, B. C. Smith, and C. Macdonald. Oxford: Oxford University Press. Byrne, Alex. 2005. “Introspection.” Philosophical Topics 33(1): 79–104. Dennett, Daniel. 1981. Brainstorms. Cambridge, Mass.: MIT Press. Edgley, Roy. 1969. Reason in Theory and Practice. London: Hutchinson. Evans, G. 1982. The Varieties of Reference. Oxford: Oxford University Press. Gertler, Brie. 2011. “Self-Knowledge and the Transparency of Belief.” In Self-Knowledge, edited by A. Hatzimoysis. Oxford: Oxford University Press. Hieronymi, Pamela. 2006.“Controlling Attitudes.” Pacific Philosophical Quarterly 87(1): 45–74. Moran, Richard. 2001. Authority and Estrangement: An Essay on Self-Knowledge. Princeton: Princeton University Press. ———. 2002. “Frankfurt on Identification: Ambiguities of Activity in Mental Life.” In Contours of Agency: Essays for Harry Frankfurt, edited by Sarah Buss and Lee Overton. Cambridge, Mass.: MIT Press. ———. 2003. “Responses to O’Brien and Shoemaker.” European Journal of Philosophy 11(3): 402–419. ———. 2007. Essay Review of The Reasons of Love by Harry Frankfurt, Philosophy and Phenomenological Research 74(2): 463–475. O’Brien, Lucy. 2007. Self-Knowing Agents. Oxford: Oxford University Press. Peacocke, Christopher. 1998. “Conscious Attitudes and Self-Knowledge.” In Knowing Our Own Minds, edited by C.Wright, B. C. Smith, and C. Macdonald. Oxford: Oxford University Press. Raz, Joseph. 1997. “When We Are Ourselves: The Active and the Passive.” Supplement to Proceedings of the Aristotelian Society 71: 211–227. Shah, Nishi, and David Velleman. 2005. “Doxastic Deliberation.” Philosophical Review 114(4): 497–534. Shoemaker, Sydney. 1968. “Self-Reference and Self-Awareness.” Journal of Philosophy 65(19): 555–567. ———. 1988.“On Knowing One’s Own Mind.” In Philosophical Perspectives: Epistemology, edited by J. Tomberlin. Ascadero, Calif.: Ridgeview Publishing. ———. 1994. “Self-Knowledge and ‘Inner Sense’.” Philosophy and Phenomenological Research 54: 249–314. Page reference to the reprinting in Shoemaker (1996). ———. 1996.“Moore’s Paradox and Self-Knowledge.” In The First-Person Perspective and Other Essays, by Sydney Shoemaker. Cambridge: Cambridge University Press. ———. 2003. “Moran on Self-Knowledge.” European Journal of Philosophy 3(3): 391–401. Wittgenstein, Ludwig. 1956. Philosophical Investigations. Oxford: Blackwell. ———. 1958. The Blue and the Brown Books. Oxford: Blackwell. Wright, Crispin, Barry C. Smith, and Cynthia Macdonald, eds. 1998. Knowing Our Own Minds. Oxford: Oxford University Press.

Part III Constitutivism

This page intentionally left blank

9 Self-Intimation and Second-Order Belief Sydney Shoemaker

This chapter defends the view that there is a constitutive relation between believing something and believing that one believes it. This view is supported by the incoherence of affirming something while denying that one believes it, and by the role awareness of the contents one’s belief system plays in the rational regulation of that system. Not all standing beliefs are accompanied by higher-order beliefs that self-ascribe them; those that are so accompanied are ones that are “available” in the sense that their subjects are poised to assent to their contents, to use them as premises in reasoning, and to be guided by them in their behavior. The account is compatible with the possibility of negative self-deception—mistakenly believing that one does not believe something—but the closest thing to positive self-deception it allows is believing falsely that a belief with a certain content is one’s dominant belief on a certain matter through failure to realize that one has a stronger belief that contradicts it. The view has implications about Moore’s paradox that contradict widely held views. On this view self-ascriptions of beliefs can be warranted and grounded on reasons—but the reasons are not phenomenally conscious mental states (as held by Christopher Peacocke) but rather available beliefs.

240

Constitutivism

1 How is it that we go from merely believing something to believing and knowing that we do? One answer, criticized by many (including me), is that we do so by employing an inner sense. In a recent paper Alex Byrne suggests a different answer: that we do so by reasoning (Byrne 2005). More specifically, we do so by following, or trying to follow, a rule he calls bel, which says “If p believe that you believe that p.” So as I write this, following bel will take me from the proposition that the sun is shining, which I believe, to the proposition that I believe that the sun is shining. Because bel is a good rule, one such that, necessarily, following it will lead to believing truths, the belief it takes me to counts as knowledge. Byrne regards this as a version of what I have called the “broad perceptual model,” since it represents belief self-ascriptions as caused by the beliefs they self-ascribe, but he says that it is not a version of inner sense theory, since the only mental faculty it involves is that of reasoning, a faculty involved in many other cases of knowledge acquisition. (To forestall false expectations, I should say that Byrne’s view is not my primary topic in this chapter; I use it only as a launching pad.) It turns out that on Byrne’s view I need not actually succeed in following bel in order to acquire knowledge of what I believe—it is sufficient that I try to follow it. If the proposition p is false, I will not succeed in following the rule, since following it involves recognizing the premise as true. But my trying to follow the rule involves my believing that p is true, and in that case the second-order belief I get to—that I believe that p—will of course be true. It might be added that I need not even be justified in believing p. As long as I do believe it, justifiably or not, the belief that I do will be true and can count as knowledge. To my mind this undermines the claim that reasoning is involved here. This is certainly not a case of coming to know the truth of a proposition by validly inferring it from premises one knows to be true and is justified in believing. I also question whether BEL as formulated is a rule we can be said to follow. A rule tells one to do something, perform a certain act, under certain conditions. But believing something—having the standing belief that so and so is the case—is not an act. Judging, thought of as a mental occurrence rather than a standing state, is an act, so a better formulation of BEL, call it BEL, would read “If p, judge that you believe that p.” Trying to follow this rule will certainly lead to correct judgments. But one can have the standing belief that one believes something without occurrently judging, or having ever occurrently judged, that one does, so appeal to BEL does not explain how we have or are justified in having such standing beliefs. But BEL has an interest independent of Byrne’s thesis. It should remind us of the point that it seems incoherent for someone to affirm a first-order proposition while

Self-Intimation and Second-Order Belief

241

refusing to assent to the proposition that he believes it. To do so is something like asserting the Moore-paradoxical sentence “p but I do not believe it.” This suggests that if one affirms p one must, on pain of irrationality, be prepared to affirm that one believes that p. And this seems a step toward the view that beliefs are constitutively self-intimating—that it is part of being a rational subject that belief that p, together with the possession of the concept of belief and the concept of oneself, brings with it the belief that one believes that p. This apparently would not be congenial to Byrne, who seems to oppose the constitutive view. And it does not, I think, support the view that anything like reasoning is involved in the acquisition of first-person second-order beliefs.

2 But I want to look further at the idea, involved in Alex Byrne’s view, that selfascriptions of belief can be viewed as the result of reasoning from first-order propositions about the world. In several places I have suggested a different way in which reasoning from first-order propositions about the world could lead to self-ascriptions of belief (Shoemaker 1988, 1994). The reasoning involves what I will call the “zany argument,” because that it how it usually strikes people. I have used this as part of a reductio ad absurdum argument against the possibility of what I have called “selfblindness,” a condition in which someone who is perfectly rational, and suffering from no cognitive deficiency, is introspectively blind to his own beliefs, and so incapable of self-ascribing them except on the basis of third-person evidence. I argue that when the supposedly self-blind person has the first-order belief that p he will, given his rationality, have available to him an instance of the zany argument that starts from the premise p and results in a self-ascription of a belief that p. The argument, as I have formulated it in earlier work, runs as follows. p. Since p is true, it will, ceteris paribus, be in the interest of anyone to act on the assumption that p, if one is in circumstances (call these relevant circumstances) in which whether one so acts is likely to affect the satisfaction of one’s interests. To act on the assumption that p is to act as if one believes that p. And part of acting as if one believes that p is acting in ways that indicate to others that one believes that p; for given that p is true, it will be in anyone’s interest to act this way in relevant circumstances. So acting will help one enlist the aid of others who believe that p in the pursuit of one’s goals. Others who believe that p, and share one’s goals, will cooperate with one in ventures undertaken on the assumption that p, and since p is true such ventures will tend to be successful. Acting in ways that indicate to others that one believes that p will include saying, in appropriate

242

Constitutivism

circumstances, that one believes that p. Since this applies to everyone, it applies to me. And since I am in appropriate circumstances, I should say that I believe that p. Having so reasoned, the self-blind man says, “I believe that p.” And in other ways he acts as if he believes that p, when he does so believe, which seems contrary to the supposition that he is self-blind with respect to his beliefs. Why does this argument strike one as zany? Well, for one thing, it has the guy saying “I believe that p” not on the basis of considerations that support its truth but for a non-epistemic reason—because he thinks that saying it will enhance his ability to enlist the cooperation of others in his pursuit of his own aims. While his statement self-ascribes a belief that p, it does not seem to be an expression of a belief that he believes that p. It is certainly odd, to say the least, to suppose that someone might have the second-order belief that he believes that p, and say that he has it on appropriate occasions, and yet never say this as an expression of that belief. But how does the availability of the zany argument show that the guy really does have the second-order belief that he believes that p? If the claim is that he has such second-order beliefs because he is behaviorally indistinguishable from someone who has them, it might seem to commit me to behaviorism. It is a defect of the zany argument as I have presented it that it assumes that it is always a legitimate inference from the truth of p to the conclusion that it is in one’s interest to act in ways that reveal to others that one believes that p. This overlooks the fact that there are cases in which one is indifferent to whether others believe that one believes something and cases in which it is in one’s interest to deceive others about what one believes.1 What is true in all cases in which one believes that p is that one is entitled to infer from p to the conclusion that it is in one’s interest to act on the assumption that p, in relevant circumstances, where this includes using p as a premise in one’s reasoning, both practical and theoretical.2 In many cases this will involve acting in ways that involve indicating to others that one believes that p, which may include saying “I believe that p.” In other cases it may involve taking steps to mislead others about what one believes. The point is that it will lead, in 1

As was forcefully urged by members of the audience when I read this paper at Brown University. The conclusion that it is in one’s interest to act on the assumption that p is drawn from the proposition that it is in anyone’s interest to do so, this because it is in anyone’s interest to act on assumptions that are true when in relevant circumstances. So if p is the proposition that it is raining here, in Ithaca, N.Y., and that proposition is true, it is in the interest of Vladimir Putin to act on the assumption that it is raining in Ithaca if he is in relevant circumstances (as, of course, he is not, and is unlikely ever to be). But in the sense in which it is true of me that I ought to act on the assumption that p, given that I believe it, this is not true of Putin. Given that I believe that p, I ought to act on the assumption that p even if in fact p is false, and obviously this is not true of Putin, who does not believe it. And because he does not believe it, we cannot conclude about him, as we can about me, that I believe that I ought 2

Self-Intimation and Second-Order Belief

243

combination with one’s desires and one’s other beliefs, to precisely the sorts of behavior that the second-order belief that one believes that p will lead to. One might put the matter as follows. There is a normative proposition that goes with the second-order belief that one believes that p, namely that one ought to be guided by the proposition p in one’s actions and in one’s reasoning, where this includes using p as a premise in relevant circumstances. One could make a case for holding that belief in the truth of this normative proposition is identical with the belief that one believes that p, for it seems identical with it in the dispositions to act and reason that it bestows. In any case, it seems plausible both that having the belief in the truth of this normative proposition entails having the second-order belief, and that having the second-order belief entails having the belief in the normative proposition. The heart of the zany argument is the point that any first-order belief gives its possessor a reason for believing the normative proposition that he ought to be guided by the believed proposition in his thought and action. And that supports the claim that any first-order belief gives its possessor a reason for thinking that he has that belief. I should stress that I am very far from holding that we come to our self-ascriptions of belief by employing the zany argument. We do not come to them by employing any sort of reasoning. But being rational involves being disposed to act in ways that are rationalized by beliefs we have, and the zany argument shows that first-order beliefs rationalize behavior, including reasoning, that can be taken as a manifestation of second-order beliefs that self-ascribe them. And it shows that belief in a proposition provides a reason for believing a proposition—the normative proposition that one ought to be guided by that proposition in one’s thought and action— which is arguably coextensive with the proposition that one believes that proposition.3 This appears to show that there is available to any believer of a proposition an argument having the same premise and the same conclusion as that licensed by Byrne’s rule BEL (or BEL) when applied to that proposition. But instead of invoking a special rule, like BEL or BEL, this argument relies on general considerations that ought to be beyond question—that ceteris paribus it is conducive to the achievement of one’s ends that one act on assumptions that are true, and that one ought to do what one believes is conducive to one’s ends. to act on the assumption that p. But of course the proposition that I, the person giving the zany argument, believe that p does not figure as a premise in the argument—if it did, the argument would be question-begging. The role played by the fact that I believe that p consists simply in the fact that I use p as a premise; and once that premise is seen to lead to the conclusion that I ought to act on the assumption that p I will, if rational, believe that consequence of what I believe. 3 What are arguably coextensive are the propositions “I believe that p” and “I ought to be guided by acceptance of p”; the sentences “NN believes that p” and “NN ought to be guided by acceptance of p,” where “NN” has the same reference in both, can differ in truth value.

244

Constitutivism

3 A different route to the self-intimation thesis starts from a point Colin McGinn made some years ago: that it is a condition of being a rational subject that one’s belief system will regularly be revised with the aim of achieving and preserving consistency and internal coherence, and that such revision requires awareness on the part of the subject of what the contents of the system are (McGinn 1982, 20). Richard Moran has objected to this that such revision and updating of the system does not require second-order beliefs; it can be largely automatic and subpersonal (Moran 2001, 109–113). But I think that in an important class of cases the revision and updating does require that that there be second-order beliefs about what the contents of the belief system are (see Shoemaker 2003, 399). These are cases in which the revision of the belief system requires an investigation on the part of the subject, one that involves conducting experiments, collecting data relevant to certain issues, or initiating reasoning aimed at answering certain questions. Such an investigation will be an intentional activity on the part of the subject, and one motivated in part by beliefs about the current contents of the belief system. These will include the belief that there are certain apparent inconsistencies or incoherences in the system, the belief that there are gaps in how the system represents the world, and the realization that the system represents the existence of certain states of affairs for which it provides no explanation. Having full human rationality requires being such that one’s revisions and updating of one’s belief system can involve such investigations, and this requires awareness of, and so beliefs about, the contents of the system. Notice that the required awareness must go beyond awareness that certain contents are believed; it must also include awareness that certain contents are not believed. If these considerations support the claim that belief is self-intimating, they also support the claim that disbelief is self-intimating.

4 But the claim that in a rational believer belief in a first-order proposition will be accompanied by a second-order belief that self-ascribes it needs qualification. The vast majority of a person’s beliefs are not accompanied by second-order beliefs to the effect that one has them. Corresponding to each thing a person knows there will be a standing belief she has—for of course knowing that p entails believing that p. And there will be countless other standing beliefs that in one way or another fall short of being knowledge. These are beliefs one has even when asleep or lounging on a beach and giving no thought to their contents. They are, most of the time,

Self-Intimation and Second-Order Belief

245

unconscious—although not in the Freudian sense. It is implausible to suppose that each of these beliefs one has is accompanied by a second-order belief to the effect that one has it.4 Let us say that a belief is “available” if the subject is “poised” to assent to its content if the question of whether it is true arises, to use it as a premise in her reasoning, and to be guided by it in her behavior. One thing that can be meant by saying that a belief is conscious is that it is available in this sense. This would be something like what Ned Block calls access consciousness.5 Available beliefs of this sort are standing beliefs; one may have such a belief at a time at which one is giving no thought to its content. Some standing beliefs are available at any time at which the subject is conscious—for me these would include the belief that I was born in Boise, Idaho. Other standing beliefs are usually latent and become available only under certain circumstances—when they are relevant to one’s current concerns or when the question of their truth is raised. What is true of any available belief is that if its content is considered it will be affirmed, and if its content is seen to be relevant to the achievement of its subject’s goals it will influence how the subject acts. It is available beliefs that are constitutively self-intimating. While most standing beliefs are not available, and so are not self-intimating in the way available beliefs are, there is nevertheless a constitutive relation between standing beliefs generally and availability. There is a constitutive relation between believing that p and being disposed to reason and act on the assumption that p. What activates this disposition is the question of p’s truth arising or the subject’s being in circumstances in which the truth or falsity of p is seen to be relevant to the subject’s concerns. The tendency of a belief to become available in such circumstances can be blocked, as in cases of self-deception. But if an informational state’s having the content that p did not have such a tendency, it would not count as a belief at all. Imagine a creature in which a part of the brain regularly, say every hour, records the creature’s blood pressure. If these blood pressure recordings lacked appropriate connections with the creature’s cognitive system and behavior they would not be beliefs.

4

Perhaps an exception should be made for beliefs that are “tacit.” Nico Silins points out that if I can be counted as believing that I did not eat an airplane for breakfast, never having thought of the matter, perhaps in the same sense I can be counted as believing that I believe things I have never thought about. 5 See Block 1995. I say “something like” because Block apparently holds that in actual cases (as contrasted with fictional cases such as that of “super-blindsight”) what is access conscious is always phenomenally conscious, whereas beliefs that are available in my sense are frequently not phenomenally conscious. I first used “available” in this sense in Shoemaker (1994).

246

Constitutivism

5 But we need to get clearer about the nature of the awareness of beliefs that occurs when they “intimate” themselves to their possessors. I have spoken of this awareness as involving the having of second-order beliefs—beliefs to the effect that one has certain first-order beliefs. This notion needs to be examined. If a second-order belief is a belief that ascribes a belief, then we have plenty of second-order beliefs that do not have first-person content. One’s beliefs about what other people believe will be second-order beliefs. Standing beliefs of this sort will resemble other beliefs one has about what states other persons are in, and also beliefs about what states inanimate objects are in. The cognitive dynamics of such a standing belief will be to some extent independent of those of the state it is about. Like most other standing beliefs, they have a sort of inertia—once acquired they tend to persist, at least for a while. The degree of independence, and the extent of the inertia, depends on what is tacitly assumed about the permanence or stability of the property ascribed. In the case of beliefs that ascribe beliefs to others, the degree of independence will depend on the content of the ascribed belief. It will be much lower if the belief is that it is raining than if it is the belief that the United States invaded Iraq. But in all these cases the second-order belief can outlast the first-order belief it ascribes, and can do so without being in any way unreasonable; one can be faultlessly ignorant of evidence that someone has lost a belief one had reasonably come to believe he had. It may be that in some cases first-person second-order beliefs have the same sort of independence of the ascribed first-order belief as do third-person second-order beliefs. Richard Moran holds that beliefs are sometimes selfascribed from what he calls the “theoretical stance” and gives as an example a woman who is persuaded by her therapist that she believes that her brother had betrayed her (Moran 2001, 85). Her reasons for believing that she believes this are much the same as the therapist’s reasons for believing it. And it is part of the example that while the woman believes that she has this belief, she is unwilling to assent to its content proposition—that her brother did betray her. Supposing that her second-order belief was initially true, it could certainly happen that the inertia of belief results in its continuing to exist after the first-order belief had ceased to exist. There is something problematic about the idea that a person might self-ascribe a belief while being unwilling to assent to its content, as is supposed to happen in Moran’s example. I return to this later. In any case, it is clear that most cases in which beliefs are self-ascribed are not like this. In most cases the question of whether one believes a certain proposition is “transparent to” the question of whether the

Self-Intimation and Second-Order Belief

247

proposition is true.6 This will be so whenever the first-person belief is available in the sense explained earlier. And in such cases the standing second-order belief that self-ascribes the available belief will not have the independence of the ascribed belief that we have in other cases of second-order belief. If the first-order belief ceases to exist, the second-order belief will likewise cease to exist. One might, indeed, wonder whether there is any need to postulate standing second-order beliefs that self-ascribe available first-order beliefs. It goes with having the available first-order belief that p that if the question whether one believes that p arises, one will judge that one does—one will assent to the proposition that one believes that p. But this seems to be the result of one’s having the belief that p, not the result of one’s having a second-order belief whose cognitive dynamics is independent of that of the belief that p, in the way that the cognitive dynamics of one’s belief about another person’s belief is independent of that of the other person’s belief. It would seem inefficient for our psychology to involve the storage of standing second-order beliefs ascribing available first-order beliefs, if there is nothing for these second-order beliefs to do that is not done by the first-order beliefs themselves. We do need episodes of second-order judgment, assent, or affirmation, and in that sense we need second-order beliefs. But perhaps there is no need for standing first-person second-order beliefs—except, perhaps, in cases like that of Moran’s analysand. But I think that rather than deny the existence of standing second-order beliefs that self-ascribe available first-order beliefs, it is better to say that while there are such beliefs, their relation to the first-order beliefs they self-ascribe is much more intimate than what we find in the case of other standing second-order beliefs— those that ascribe beliefs to other persons, and those that self-ascribe beliefs from the “theoretical stance,” i.e., on third-person evidence. One way of expressing this claim is by saying that in such cases the standing second-order belief contains the self-ascribed first-order belief as a part.7 Supposing that the belief that one believes that p contains as a part the belief that p, what more is there to it than this? Perhaps the answer lies in what I said earlier. It is not the belief that p all by itself that accounts for the disposition to judge that one has it if the question arises; this requires in addition the possession of the concepts of belief and of oneself, and it requires a certain degree of rationality. Perhaps it is all of this, together with the

6 The transparency will not be complete. As Nico Silins pointed out to me, if the answer to the question whether p is true is “maybe,” the answer to the question whether one believes that p will normally not be that. 7 Very different versions of this view are proposed in Heal (1994) and Shoemaker (1994)—in Heal it is presented as part of an anti-functionalist view, while in mine it is part of a functionalist view.

248

Constitutivism

belief that p, that constitutes the standing second-order belief that one believes that p. It of course falls out from this that the cognitive dynamics of the standing secondorder belief is not independent of that of the first-order belief it self-ascribes. If a belief has the belief that p as an essential part, its possession cannot survive the loss of the belief that p. Here it may seem that a regress threatens. If having an available belief requires having a second-order belief that has it as a part, and that second-order belief is itself available, it would seem that the second-order belief requires a third-order belief that has it as a part—and so on ad infinitum. This threatens a regress only if the higher-order belief that self-ascribes an available belief must itself be available, and this seems to me not obvious. But even if it is true, I do not believe there is a damaging regress here. Having the second-order belief requires the possession of concepts, of belief and of oneself, beyond those required by the possession of the first-order belief, but no additional concepts are required by the possession of beliefs of higher orders. If having the second-order belief that one has an available first-order belief is just having that available belief together with the concepts of belief and of oneself and an appropriate degree of rationality, it would seem that having that second-order belief should also count as having the third-order belief that one has it, that this in turn should count as having the fourth-order belief that one has that one, and so on ad infinitum. (This has the consequence that beliefs having different contents can have the same metaphysical basis.)

6 But if we take this approach, what are we to say about cases of self-deception? I think that one can clearly be self-deceived in thinking that one does not believe something. It is common for a person to have prejudices that he is unaware of and would sincerely deny having. Is it also possible for someone to be self-deceived in thinking that he does believe something? It is this that would be threatening to my version of the self-intimation claim, for a false belief that one believes that p could not have as a part the belief that p. Christopher Peacocke has an example in which an administrator claims to believe that graduates of foreign universities are as qualified as graduates of universities in her own country, but who in making hiring decisions that are supposed to be based solely on the qualifications of the applicants systematically gives preference to her compatriots (Peacocke 1998, 90). This is supposed to be a case of someone deceiving herself into thinking that she believes something she does not. But I do not think that it is. The administrator’s hiring record is certainly evidence that she

Self-Intimation and Second-Order Belief

249

has the belief that graduates of universities in her own country are more highly qualified, so let us suppose that she does have that belief and is mistaken in thinking that she lacks it. It is compatible with this that she also has the belief that she sincerely avows, that foreign graduates are equally qualified—and so, of course, has incompatible beliefs. How could her avowal be sincere if she does not in some sense, or in some degree, believe its contents? It may be that in some cases of this sort there is no fact of the matter whether a person does have a belief that he professes to have. He avows the belief with apparent sincerity, and, what would go with this, accepts some inferences in which it figures as a premise. But he sometimes fails to accept its consequences, and much of his behavior is not in accord with it. If it is indeterminate in such a case whether the person believes the thing in question, it is also indeterminate whether he is sincere in avowing it. But suppose, for the moment, we allow that there can be a case in which a person does definitely lack a belief that he definitely believes he has, where this is the result of self-deception. Since he lacks the first-order belief, it cannot be the case here that the second-order belief has the first-order belief as a part, the rest of it being the possession of the required concepts and the required degree of rationality. What then can the second-order belief consist in? Can it be like the second-order belief that Moran’s analysand is supposed to have—that is, can it be like second-order beliefs about other persons? It will not, of course, be grounded on third-person evidence. What dispositions will it bestow, besides the disposition to assent to the proposition that its subject believes the thing in question? Will it bestow the disposition to assent to the content of that putative belief ? If it does, we will then have a case for saying that the person does believe that content, or at least that it is not determinately true that he does not. If it does not bestow that disposition, then the person will be liable to fall into a version of Moore’s paradox—saying, or thinking, “I believe that p, but not-p,” or “I believe that p, but I have no idea whether p is true.” Of course, the case of Moran’s analysand is problematic in much the same way. She is supposed to believe that she believes that her brother betrayed her. But she is not willing to affirm the proposition that he did betray her. And so she seems liable to fall into versions of Moore’s paradox. I think that the only way to save the coherence of this case is to suppose that it involves there being a divided mind. One part of a person’s mind believes something, and another does not, and the part that does not believe it ascribes the belief to the part that does. This may seem farfetched, but if one reflects on what constitutes the synchronic unity of a mind it does not seem impossible that such unity might be only partial. In the case of Moran’s analysand, the speaking part of the mind has normal introspective access to its own mental states, and finds there no belief about her brother’s betrayal, but has

250

Constitutivism

behavioral access to another part of her mind in which indeed there is such a belief. The Moore-paradoxical utterance “I believe that my brother betrayed me, but to the best of my knowledge he did not” comes out as not really paradoxical, if interpreted as saying that in one part of the subject’s mind there is such a belief and in another, the part from which the utterance comes, there is not. This is reminiscent of Wittgenstein’s remark that to make sense of my saying “It is raining and I don’t believe it,” or “It seems to me that my ego believes this but it isn’t true,” one would have to “fill out the picture with behavior indicating that two people were speaking through my mouth” (Wittgenstein 1958, 192). But can we deal in an analogous way with cases of self-deception? Consider again Peacocke’s case, on the understanding of it that has the person willing to affirm the proposition that foreign candidates are as good as local ones. There it is the same part of the mind, the speaking part, that affirms the second-order proposition and the first-order proposition it represents the person as believing. And as we saw, there is a case here for saying that the person does have the first-order belief, even though she also has a prejudicial belief that contradicts it. We might invoke a division of the mind to explain the person’s denial that she has the prejudice—we might suppose that the prejudice is located in a part of the mind other than the speaking part. But this gets us only negative self-deception—not positive self-deception involving self-ascription of a belief one does not have. It may be that the best we can do by way of allowing for positive self-deception about one’s own beliefs is to say that a person can be deceived in thinking that he has a belief with a certain content as his dominant belief about a certain matter, i.e., as the belief that normally governs his behavior and reasoning in relevant circumstances. Such a higher-order belief is not wrong in self-ascribing a belief with the content in question, but it is wrong in assigning it a dominant and stable status in his belief system. We can suppose that this is true in the Peacocke example. This is really a special case of negative self-deception about one’s beliefs, for it results in the mistaken belief that one does not have a stronger and more stable belief whose content conflicts with that of the one in question.

7 But second-order beliefs can be negative, as well as positive—one believes that one does not believe certain things, or, as I will say, that one disbelieves them. And these seem to have the same authority as positive second-order beliefs. I thought at first that one could give an account of these that is exactly parallel to the account I have given of positive second-order belief—let the second-order belief that one does not

Self-Intimation and Second-Order Belief

251

believe that p have as a part the disbelief that p, the rest of it being the possession of the concepts and the degree of rationality required for the second-order belief. But this would make impossible the sort of self-deception I have said clearly exists, namely being self-deceived in thinking that one does not believe something. We need a slightly different account. What is true when someone thinks that he does not believe that p is that (a) he does not have an available belief that p and (b) he is such that considering whether p does not give him one. It is this state of affairs plus possession of the relevant concepts and degree of rationality that constitutes the second-order belief, and this can exist when the person does in fact have a belief, although not an available one, that p. There is a notion of availability applicable to disbelief that is somewhat analogous to the notion of availability applicable to belief. But the cases are not quite parallel. There is no first-order manifestation of disbelief corresponding to affirmation or assent as a manifestation of belief—disbelieving that p need not go with affirming not-p, for one may have no opinion on the matter, and so believe neither p nor not-p. The only disposition to affirm that goes with having an available disbelief that p is the disposition to make the second-order affirmation that one does not believe that p. The disbeliefs that are self-intimating are the ones that are available; negative second-order beliefs that ascribe these disbeliefs have them as parts.

8 If, as I have suggested, available first-order beliefs are parts of the second-order beliefs that self-ascribe them, it can hardly be the case that the first-order beliefs are causes of the corresponding second-order beliefs. What the first-order belief that p does cause, or contribute to causing, are the episodes of affirmation that one believes that p that occur when the proposition that p is considered. The relation of the firstorder beliefs to the standing beliefs that self-ascribe them is one of partial constitution rather than one of causation. What then of the view, which I once held, that the relation between first-order and second-order beliefs consists in the truth of conditionals saying that if the content of the first-order belief is considered, the person will acquire the second-order belief that he has it? This would be appropriate if the first-order beliefs were causes of the beliefs that self-ascribe them. And it would be true if the second-order beliefs were episodes of affirmation or assent. But it seems wrong if the second-order beliefs are standing beliefs. If what I have suggested about the nature of these beliefs is correct, there seems no reason not to say that these come into being with the first-order beliefs they self-ascribe. The standing belief that one believes that p has as a part the

252

Constitutivism

available belief that p, so whatever causes one to have this available belief, or causes the belief to be available, causes one to have the standing second-order belief.

9 The view that available first-order beliefs and disbeliefs are parts of the standing second-order beliefs that self-ascribe them has implications about the nature of Moore’s paradox. There are at least two kinds of Moore-paradoxical sentences: ones of the form “p but I do not believe that p,” and ones of the form “I believe that p, but not-p.” The standard view is that while there would be something logically amiss in an assertive utterance of a sentence of either of these forms, and in the belief such an utterance would purport to express, the sentences are not themselves self-contradictory and no contradiction would be involved in asserting them. Contradictions do lurk in the background. If we conjoin the proposition expressed by a sentence of the form “p but I do not believe that p” with the proposition that one believes that proposition, we do get a self-contradiction. If we conjoin the proposition expressed by a sentence of the form “I believe that p, but not-p” with the proposition that one believes that proposition, we do not get a self-contradiction, but we do get something that says that the subject has beliefs that contradict one another.8 None of this, however, implies that the Moore-paradoxical sentences themselves assert self-contradictory propositions. But if first-order beliefs and disbeliefs are parts of the second-order beliefs that self-ascribe them, then it appears that the beliefs Moore-paradoxical sentences purport to express are themselves self-contradictory or impossible. Consider first a sentence of the form “I believe that p, but not-p.” Belief in its first conjunct will have as a part a belief that p, and belief in its second conjunct will be a belief that not-p; so believing both conjuncts will involve believing both p and not-p. Now consider a sentence of the form “p but I do not believe that p.” If this were believed the belief would include the belief that p, since p is its first conjunct. But it should also include belief in its second conjunct, which would include disbelief that p. So the belief would have to consist of a belief that p combined with a disbelief that p. While it may be possible for someone to believe and disbelieve something at the same time, perhaps in different parts of her mind, it seems impossible that a single belief could consist in believing and disbelieving the same proposition.

8 These points are made in Shoemaker (1996). I subsequently learned from reading Heal (1994) that they are made in Baldwin (1990).

Self-Intimation and Second-Order Belief

253

Of course, the view I am defending does not say of all self-ascriptions of beliefs and disbeliefs that they contain as parts the beliefs or disbeliefs they ascribe; it says this only of those that ascribe available beliefs or disbeliefs. And the Moore-paradoxical sentences do not say that the beliefs and disbeliefs involved are available. In the case of Moran’s analysand, on the divided-mind interpretation of it, we can have an utterance of “I believe that my brother deceived me, but he did not” that is nonparadoxical. So we certainly should not say that these sentences are self-contradictory. What we can say is that where the beliefs or disbeliefs that conjuncts of these sentences purport to express are available, beliefs in these sentences would either be selfcontradictory or impossible.

10 What does this account imply about the warrant of self-ascriptions of belief ? Some writers associate the view that there is a constitutive relation between first-order beliefs and the second-order beliefs that self-ascribe them with what Christopher Peacocke calls the “no reasons” view about such self-ascriptions—the view that someone who makes such a self-ascription neither has nor needs a reason for doing so (see Peacocke 1998). In a discussion of Peacocke’s view Michael Martin ascribes to Peacocke, and endorses, the claim that the constitutive view implies the noreasons view, and in doing so he also ascribes to Peacocke, and endorses, a rejection of both the no-reasons view and the constitutive view (Martin 1998). Let me note in passing that I question whether it is correct to interpret Peacocke as rejecting the constitutive view. He says at one point that “the explanation of the correctness of certain ways of coming to knowledgeable self-ascriptions is part of the essential nature of the mental state ascribed” (Peacocke 1998, 90). And as Martin points out, Peacocke’s account of how we can make such knowledgeable selfascriptions rests in part on his account of the possession conditions for the concept of belief, a central part of which is that for the relational concept R to be the concept of belief it must be the case that the thinker “finds the first-person content that he stands in R to p primitively compelling whenever he has the conscious belief that p, and he finds it compelling because he has that conscious belief ” (Martin 1998, 109). While Martin seems to think that this gives us only an epistemological connection between consciously believing that p and believing that one so believes, it seems to me to give us a metaphysical connection. What it makes metaphysically constitutive of having or being prone to have the second-order belief is not the first-order belief by itself, but this together with the possession of the concept of belief. This is precisely the sort of constitutive relation I argued for earlier in the case of available beliefs.

254

Constitutivism

In any case, the version of the constitutive view I am putting forward agrees with Peacocke in rejecting the no-reasons view—this despite the fact that it is passages from my work that Peacocke quotes in expounding that view. But it favors an account of the warrant of second-order beliefs that differs from his. In what Peacocke takes to be the central case, the self-ascription of the belief that p has as a reason a relevant phenomenally conscious state of judging that p. So, in his example, the belief that one believes that Dubcek was Prime Minister of Czechoslovakia at the time of the Soviet invasion might be based on an “apparent propositional memory” that he was, this leading to a judgment that he was. But Peacocke allows that we sometimes take shortcuts and can sometimes be warranted in making such self-ascriptions when no such judgment has occurred. Such shortcuts are permissible “provided that they are taken in circumstances in which the thinker could take the longer route, with each transition in the longer route made for the right sort of reason” (Peacocke 1998, 74). In such a case we have a “no intermediate conscious state,” NICS, which satisfies the “requirement of firstorder ratifiability” (ibid., 93–94). The missing intermediate conscious state in such a case appears to be the judgment affirming the content of the first-order belief. As best I can see, the requirement of first-order ratifiability comes to the same thing as the requirement that the case be one in which the question whether one believes that p is transparent to the question of whether p is true. And that is something that will be true in any case in which the first-order belief is available in the sense I have tried to explain. To say that a case satisfies the requirement of firstorder ratifiability, that it is one in which the thinker “could take the longer route,” is apparently to say that it is one in which the thinker would affirm the proposition that is the content of the belief self-ascribed by the second-order belief in question, and so could have the affirmation of it as a reason for his self-ascription. In such a case the thinker’s willingness to affirm that he has the belief depends on his willingness to affirm its content, which amounts to its satisfying the transparency condition. And in any such case the first-order belief will be one that is available. Peacocke thinks that in such cases the subject is warranted in self-ascribing the first-order belief. That amounts to the claim, with which I completely agree, that one is warranted in self-ascribing the belief that p if one has an available belief that p. But why should not that be the full account? Peacocke apparently thinks that what underlies the warrant in such cases is the fact that the circumstances are such that the subject could base his self-ascription on a reason of his preferred sort—a phenomenally conscious state. Here I am skeptical. I have no objection to the claim that self-ascriptions of beliefs are grounded on reasons. Nor do I object to the claim that the reason for such a self-ascription must be a conscious state, as long as “conscious” can here mean “available.” I do not see why an available belief cannot count

Self-Intimation and Second-Order Belief

255

as a reason for self-ascribing the belief that one has it—on the view I am putting forward here, this would amount to its being a reason for the second-order belief of which it is a part. What I question is the view that occurrent phenomenally conscious states, first-person judgments being the central case, are the canonical reasons for belief self-ascriptions. Peacocke’s view is in some ways similar to the view of Alex Byrne that I started off with in this chapter. He does not say, as Byrne does, that we get to belief selfascriptions by reasoning from first-order propositions about the world. But his central case is one in which the person affirms such a proposition, and it is his affirming it, judging it to be true, that is his reason for self-ascribing the belief in its truth. Saying that such occurrent states are reasons for self-ascriptions could mean either that they are the reasons for occurrent second-order judgments, to the effect that one believes something, or that they are reasons for standing second-order beliefs with such content. It can happen that such an occurrent second-order judgment is what initiates a standing second-order belief with the same content. And in such a case, perhaps, a phenomenally conscious state, e.g., a first-order judgment, could be one’s reason for making the second-order judgment and, derivatively, one’s reason for holding the second-order belief. But this is far from being the normal case, and I question whether it is the central case. Normally an occurrent judgment or affirmation that expresses a second-order belief comes some time after the second-order belief was acquired—for if I am right, the second-order belief was acquired when the first-order belief it self-ascribes became available. What in such case will be one’s reason for making it? If the judgment takes the form of an assertion, one will of course have a reason for making that assertion, e.g., to inform someone of its content. But that will not be an epistemic reason for affirming its content. One’s epistemic reason for affirming its content will be one’s reason for holding the second-order beliefs it expresses, and there will not be any relevant phenomenally conscious state that occurs at or near the time at which the judgment was made. The available standing belief may have been initiated by a conscious first-order judgment. But it would be strange to say that at a later time at which one has this available belief it is that earlier judgment, now probably long since forgotten, that is one’s reason for thinking that one has it. I have various available beliefs about the house I live in, and, so I think, have second-order beliefs that self-ascribe these. In most cases I have no notion of how these second-order beliefs were first acquired. No first-order judgment I am now making, and no first-order judgment I made in the past, is now my reason for holding these beliefs. Earlier, before I put forward the view that first-person second-order beliefs have as parts the first-order beliefs they self-ascribe, I flirted with the view that while there are second-order judgments that self-ascribe beliefs, there are no second-order

256 Constitutivism standing beliefs that self-ascribe available first-order beliefs. That is apparently not Peacocke’s view, for he does speak of beliefs, as well as judgments, that self-ascribe beliefs. But holding that view would protect him from the criticism just made. That criticism rested in part on the claim that one’s epistemic reason for making a judgment is one’s epistemic reason for holding the standing belief that the judgment expresses. This will not apply to second-order judgments if there are no standing second-order beliefs they express. So Peacocke could deny that there are standing second-order beliefs, and frame his view as one about the reasons for second-order judgments, holding that these reasons are, in the central case, judgments that express the first-order beliefs self-ascribed. As I indicated earlier, the denial that there are standing first-person second-order beliefs does not strike me as plausible. In any case, adopting it would not protect Peacocke from another difficulty. Beliefs that one does not believe certain things seem to have much the same epistemic status as beliefs that one does believe certain things. In whatever way the latter requires reasons in order to be warranted, it would seem that the former should do so as well. But there do not appear to be phenomenally conscious states that are capable of being reasons for self-ascriptions of disbelief in the way Peacocke thinks that first-order judgments are reasons for self-ascriptions of belief. As I indicated earlier, I am prepared to say that an available belief can count as a reason for its self-ascription. And in the same way, an available disbelief can count as a reason for its self-ascription. Here, of course, being a reason for a belief is not a matter of being evidence for it. My having the available belief that p is not evidence that I believe that p—it constitutes it. Likewise, my having the available disbelief that q is not evidence for, but rather constitutes, the fact that I do not believe that q. But it is not in general true that the existence of a fact that constitutes the truth of a proposition gives us a reason for believing that proposition. Were that so, we would have reasons for believing everything that is true! So how can it be that available beliefs and disbeliefs give us reasons for believing in their existence? The sense in which having the available belief that p gives one a reason for believing that one believes that p is just that if one has that available belief one is thereby justified, or warranted, in believing that one has it. Likewise, if one has the available disbelief that q, one is thereby justified or warranted in believing that one has it. And why is that? Here is where it helps to bring in the notion of rationality. We noted earlier that it is a requirement of full human rationality that one regularly revise one’s belief system in the direction of greater consistency and coherence, or to preserve consistency and coherence in the light of new information, and, as a condition of one’s being able to do this, that one have access to its contents and their relations to one another. Judging that one believes something when one does,

Self-Intimation and Second-Order Belief

257

and judging that one does not believe something when one does not, are manifestations of the satisfaction of this requirement of rationality. This seems a sufficient reason to say that one is warranted in doing so. And if, as argued in section 2, anyone with the available belief that p has available an argument from the proposition p to a proposition necessarily coextensive with the proposition that he believes that p, it seems reasonable to conclude that such an available belief counts as a reason for having that second-order belief.9

References Baldwin, T. 1990. G. E. Moore. London: Routledge. Block, N. 1995. “On a Confusion about a Function of Consciousness.” Behavioral and Brain Sciences 18: 237–247. Byrne, A. 2005. “Introspection.” Philosophical Topics 33(1): 79–104. Heal, J. 1994. “Moore’s Paradox: A Wittgensteinian Approach.” Mind 103: 5–24. Martin, M. 1998. “An Eye Directed Outwards.” In Knowing Our Own Minds, edited by C. Wright, B. C. Smith, and C. Macdonald. Oxford: Oxford University Press. McGinn, C. 1982. Character of Mind. Oxford: Oxford University Press. Moran, R. 2001. Authority and Estrangement. Princeton: Princeton University Press. Peacocke, C. 1998. “Conscious Attitudes, Attention, and Self-Knowledge.” In Knowing Our Own Minds, edited by C.Wright, B. C. Smith, and C. Macdonald. Oxford: Oxford University Press. Shoemaker, S. 1988. “On Knowing One’s Own Mind.” In Philosophical Perspectives: Epistemology, edited by J. Tomberlin. Ascadero, Calif.: Ridgeview Publishing Co. ———. 1994. “Self-Knowledge and ‘Inner Sense’.” Philosophy and Phenomenological Research 54: 249–314. ———. ed. 1996. “Moore’s Paradox and Self-Knowledge.” In The First Person Perspective and Other Essays. Cambridge: Cambridge University Press. ———. 2003. “Moran on Self-Knowledge.” European Journal of Philosophy 11: 391–401. Wittgenstein, L. 1958. The Philosophical Investigations. Oxford: Blackwell.

9

Earlier versions of this chapter were presented at the University of Toronto, at conferences in Dubrovnik, Croatia, Duisburg, Germany, and Canberra, Australia, at Brown University, and at Girona, Spain. Thanks to the audiences on these occasions, and to Nico Silins and an anonymous referee, for helpful comments.

This page intentionally left blank

10 A Simple Theory of Introspection Declan Smithies

1. What is Introspection? What is introspection? According to William James ([1890] 1981, 85), “The word ‘introspection’ need hardly be defined—it means, of course, the looking into our own minds and reporting of what we there discover.” However, it cannot simply be assumed, as James does, that introspection is a form of inner perception. A more neutral starting point is to use the term ‘introspection’ as a placeholder for the distinctive way (whatever it is) in which we know our own minds. This is to make the plausible assumption that there is a distinctively introspective way in which we know our own minds, but without begging important questions about its nature. Moreover, this defines the agenda for a theory of introspection—that is, to describe and explain what exactly is distinctive about the introspective way in which we know our own minds. We can begin by noting that introspection has a distinctive subject matter. We will return in due course to the question of how exactly to demarcate its subject matter, but to a first approximation, introspection is a way of knowing about one’s current mental states (events, processes, and so on), as opposed to the mental states of others, one’s past mental states, or the states of one’s body, such as one’s height or weight. However, introspection cannot be defined solely in terms of its subject matter without collapsing the distinction between knowing about one’s mental states by

260

Constitutivism

introspection and knowing about one’s mental states in some other way. So, if introspection is a distinctive way of knowing about one’s mental states, then we need to consider what sets it apart from other ways of knowing about the same subject matter. Following Alex Byrne (2005), we can separate two dimensions along which introspection may be distinguished from other ways of knowing about the world. First, introspection is peculiar in the sense that it is epistemologically different from other ways of knowing about its subject matter. And second, introspection is privileged in the sense that it is epistemologically better than other ways of knowing about its subject matter. A plausible constraint of adequacy on any theory of introspection is that it should describe and explain the sense in which introspection is both peculiar and privileged. My primary goal in this chapter is to defend what I call the simple theory of introspection. According to the simple theory, introspection is a distinctive way of knowing that one is in a certain mental state, which one has just by virtue of being in that mental state. The simple theory is motivated in part by reflection on examples. For instance, if I am in pain, then I have a distinctive way of knowing that I am in pain just by virtue of the fact that I am in pain. Similarly, if I am thinking about rhubarb, then I have a distinctive way of knowing that I am thinking about rhubarb just by virtue of the fact that I am thinking about rhubarb.1 Moreover, the simple theory explains what is distinctive about the introspective way in which we know about our own mental states. Introspection is peculiar in the sense that it is a way of knowing about one’s own mental states that is not available to anyone else. If others know that I am in pain, then they know this by testimony or by making inferences from observation of my behavior or the activity of my brain, whereas I am in a position to know that I am in pain just by virtue of being in pain. Moreover, introspection is privileged in the sense that it is a way of knowing about one’s mental states that is always available in one’s own case. Others are in a position to know that I am in pain only if they have sufficient evidence from observation, inference, or testimony, whereas I am always in a position to know that I am in pain just by virtue of being in pain. The simple theory therefore provides an elegant and plausible explanation of the way in which introspection is epistemologically different from, and better than, other ways of knowing about the world.

1 Williamson (2000, ch. 4) argues that no nontrivial conditions are luminous in the sense that if they obtain, then one is in a position to know that they obtain. However, Smithies (forthcoming, a) draws a distinction between epistemic and doxastic interpretations of luminosity and claims that Williamson’s argument fails to show that there are no epistemically luminous, as opposed to doxastically luminous, conditions. Compare the distinction between epistemic and doxastic versions of constitutivism in section 2.

A Simple Theory of Introspection

261

Any theory will agree that one has an introspective way of knowing that one is in a certain mental state only if one is in that mental state, since knowledge is factive. However, many theories, including inner sense theories and reliability theories, impose further requirements to the effect that one has an introspective way of knowing that one is in a certain mental state only if one has an inner representation of that mental state or an introspective mechanism that reliably tracks that mental state. The simple theory, by contrast, claims that if one is in a certain mental state, then one has a way of knowing that one is in that mental state just by virtue of being in that mental state. However, the mere fact that p is not in general sufficient to provide one with a way of knowing that p. So, how is the mere fact that one is in a certain mental state sufficient to provide one with a way of knowing that one is in that mental state? The simple theory explains the distinctive features of introspective knowledge in terms of introspective justification. Introspective justification, according to the simple theory, is a distinctive kind of justification that one has to believe that one is in a certain mental state, which one has just by virtue of being in that mental state.2 As such, introspective justification is fundamentally different in character from other kinds of justification, including perceptual justification. For instance, one does not have perceptual justification to believe that the world is in a certain state just virtue of the fact that the world is in that state, but rather by virtue of one’s perceptual experience, which represents that the world is in that state. The distinguishing feature of introspective justification is that its source is identical with its subject matter. Arguably, this has the further consequence that introspective justification, unlike perceptual justification, is (1) self-intimated; (2) infallible; (3) indefeasible; and (4) immune from Gettier cases. If so, then having introspective justification to believe that one is in a certain mental state is sufficient to provide one with a way of knowing that one is in that mental state. Let us briefly consider each of these features in turn. First, introspective justification is self-intimated in the sense that if one is in a certain mental state, then one has introspective justification to believe that one is in that mental state. By contrast, perceptual justification is not self-intimated, since it is not the case that if the world is in a certain state, then one has perceptual justification to believe that it is in that state. Rather, one has perceptual justification to believe that the world is in a certain state only if one has a perceptual experience, which represents that the world is in that state. This has the interesting consequence that if ideal rationality involves believing every proposition that one has justification to believe, 2

Neta (2011) makes a related proposal, which he calls the truth-sufficiency account of privileged access. According to his account, S has privileged access to the fact that p if and only if p is itself a justification for S to believe that p.

262

Constitutivism

then ideal rationality is compatible with ignorance about the external world, but not with ignorance about one’s own mental states.3 Second, introspective justification is infallible in the sense that if one has introspective justification to believe that one is in a certain mental state, then one is in that mental state. Again, perceptual justification is not infallible, since it is not the case that if one has perceptual justification to believe that the world is in a certain state, then it is in that state. After all, there are possible cases of perceptual illusion and hallucination in which one’s perceptual experience misrepresents the world and so one has justification to believe falsely that the world is in a certain state. However, there are no possible cases of introspective illusion or hallucination, since the source of one’s introspective justification is identical with its subject matter. As we shall see in section 2, there are possible cases in which one has false beliefs about one’s own mental states, but according to the simple theory, there are no such cases in which one has false beliefs that are introspectively justified. As before, this implies that ideal rationality is compatible with error about the external world, but not with error about one’s own mental states. Third, introspective justification is indefeasible in the sense that it cannot be defeated by justification to believe anything else. Perceptual justification, on the other hand, can be defeated by justification to believe that one’s perceptual experience is the result of illusion or hallucination. According to the simple theory, however, introspective justification cannot be defeated in this way, since it is a priori that there are no possible cases of introspective illusion or hallucination. Moreover, it is not clear that there is any other way in which introspective justification can be defeated. David Armstrong (1968: 109) gives the example of a brain technician, who informs me on the basis of neural evidence that I believe falsely that I seem to be seeing something green. But if my belief is false, then I have no introspective justification to be defeated; and if my belief is true, then it is more plausible to suppose that my introspective justification defeats the testimony of the brain technician, rather than vice versa.4 Fourth, introspective justification is immune from Gettier cases in the sense that if one has an introspectively justified belief that one is in a certain mental state, then

3

A prominent theme in the literature on self-knowledge is that rationality imposes limits on the possibility of ignorance and error about one’s own mental states. For instance, Shoemaker (1996) argues against the possibility of self-blindness in which one is rational and conceptually competent, but one has no introspective knowledge of one’s own mental states. Similarly, Burge (1996) argues against the possibility of brute ignorance and brute error about one’s own mental states—that is, ignorance and error that reflects no rational failure or malfunction in the subject. 4 Compare Steup’s (2009) discussion of the indefeasibility view.

A Simple Theory of Introspection

263

one knows by introspection that one is in that mental state. Perceptual justification, of course, is subject to Gettier cases in which one has a true belief that is justified by perception, but which fails to constitute knowledge. The standard recipe for constructing Gettier cases is to begin with a case in which a justified belief is false owing to bad luck and then to add an element of good luck that counteracts the bad luck and thereby makes the justified belief true, but luckily so. However, this recipe cannot be applied unless justification is fallible, which explains why it applies in the case of perceptual justification, but not introspective justification.5 Moreover, it is not clear that there is any other recipe for generating Gettier cases for introspectively justified beliefs.6 In summary, the simple theory claims that some mental states are introspectively accessible in the sense that being in a certain mental state is necessary and sufficient for having introspective justification to believe that one is in that mental state and thereby for having an introspective way of knowing that one is in that mental state. This claim is encapsulated in the following introspective accessibility thesis: The introspective accessibility thesis: for some mental states M, necessarily, one is in M if and only if one has introspective justification to believe that one is in M and one thereby has an introspective way of knowing that one is in M. However, the introspective accessibility thesis raises further questions of its own.7 First, there is the generalization question: Which mental states are introspectively accessible in the relevant sense? And second, there is the explanatory question: What explains why some mental states are introspectively accessible, rather than others, or none at all? The simple theory is motivated initially by reflecting on examples of introspective knowledge and by the way in which it explains their peculiar and privileged status. However, opponents of the simple theory may deny that any mental states are introspectively accessible in the sense articulated by the simple theory. Therefore, a full defense of the simple theory requires answering the generalization question and the explanatory question in such a way as to provide a theoretical rationale for the claim that certain mental states are introspectively accessible, rather than others, or none at all. My strategy in this chapter is to argue for the conditional claim that if

5

This recipe is proposed by Linda Zagzebski (1994, 69). She notes that the recipe applies only given some degree of modal independence between the justification condition for knowledge and the truth condition. 6 Williamson’s (2000, ch. 4) anti-luminosity argument is sometimes regarded as a source of Gettier cases for introspectively justified beliefs, but see Smithies (forthcoming, a) for arguments against this view. 7 Here I am indebted to Daniel Stoljar who suggested these labels.

264

Constitutivism

access internalism is true, then the accessibility of facts about justification is best explained by the introspective accessibility of facts about one’s mental states. Therefore, I conclude that access internalism provides a theoretical rationale for the simple theory. The overall plan for the chapter is as follows. In section 2, I elaborate the simple theory by situating it within the context of a distinction between epistemic and doxastic versions of constitutivism. In sections 3 and 4, I answer the generalization question by arguing that mental states are introspectively accessible if and only if they are phenomenally individuated. In sections 5 and 6, I answer the explanatory question by arguing that mental states are introspectively accessible if and only if they are among the determinants of justification. Finally, in section 7, I conclude by reflecting on the nature of the relationship between rationality and introspective self-knowledge.

2. Epistemic and Doxastic Constitutivism Constitutivism is the thesis that there is a constitutively necessary connection between certain mental states and our cognitive or epistemic access to those mental states. It is usually formulated in terms of an accessibility thesis of the following schematic form: The schematic version: for some mental states M, and some cognitive or epistemic relation Φ, necessarily, one is in M if and only if one Φs that one is in M. The accessibility thesis is a biconditional that conjoins a self-intimation thesis, which is the conditional that goes from left to right, and an infallibility thesis, which is the conditional that goes from right to left. The self-intimation rules out the possibility of certain kinds of ignorance, whereas the infallibility thesis rules out the possibility of certain kinds of error. The schematic version of the accessibility thesis raises various questions. Which mental states, M, are accessible? Which are the modes of presentation under which they are accessible? And what is the cognitive or epistemic relation, Φ, in terms of which accessibility is defined? According to doxastic versions of constitutivism, there is a constitutively necessary connection between certain mental states and our beliefs about those mental states. In its simplest formulation, this is captured by the following doxastic version of the accessibility thesis: The doxastic version: for some mental states M, necessarily, one is in M if and only if one believes that one is in M.

A Simple Theory of Introspection

265

Unfortunately, not even paradigm examples of phenomenally conscious mental states, such as feeling pain and feeling cold, are doxastically accessible in this sense. Even if we restrict the scope of the thesis to phenomenally conscious states, which are self-ascribed under first-person, present-tense phenomenal modes of presentation, it is easy to generate counterexamples. First, there are cases of conceptual poverty, in which one is in a phenomenal state, but one does not believe that one is in that state because one does not possess the requisite concepts. For example, an infant feels pain, but it lacks the conceptual abilities required to believe that it feels pain. Second, there are cases of inattention, in which one is in a phenomenal state, but one does not believe that one is in that state because one’s attention is distracted. For example, an athlete incurs a painful injury while playing football but does not notice the pain until the game is over.8 Third, there are cases of misclassification, in which one is in a phenomenal state, but one does not believe that one is in that state because one misclassifies it. For instance, a student in an initiation trick is threatened with a red-hot poker and then touched with an ice cube, so that he is duped into mistaking the mildly unpleasant feeling of cold for an intensely painful feeling of heat. In that case, he feels cold, but he does not believe that he feels cold; and moreover, he believes that he feels pain, but he does not in fact feel pain.9 How might we reformulate the accessibility thesis in such a way as to avoid these counterexamples? A common strategy is to retain a doxastic version of the accessibility thesis, while imposing further restrictions designed to rule out the various counterexamples. My own strategy, by contrast, is to reformulate the accessibility thesis in epistemic, rather than doxastic, terms. I do not deny that there may be some way of restricting the doxastic version of the accessibility thesis in such a way as to make it immune to counterexamples. However, I claim that any such restricted version of the doxastic accessibility thesis can be explained as a consequence of an unrestricted version of the epistemic accessibility thesis, which is more fundamental in the order of philosophical explanation. Or so I will argue in what follows. The simple theory, as we have seen, entails an epistemic version of the accessibility thesis, according to which there is a constitutively necessary connection between certain mental states and our epistemic access to those mental states:

8 This example is from Shoemaker (1996, 227). See also Armstrong’s (1968, 93–94) long distance driver and Block’s (1997, 386) unattended drill for further examples in a similar vein.The literature on inattentional blindness contains a further stock of examples. 9 A version of this example appears in Locke (1967, 86). Gilbert Ryle’s problem of the speckled hen gives rise to additional examples of misclassification; for further discussion, see Evans (1982, 228–229), Sosa (2003), and Smithies (forthcoming, a).

266

Constitutivism

The epistemic version: for some mental states M, necessarily, one is in M if and only if one has introspective justification to believe that one is in M and one thereby has an introspective way of knowing that one is in M. The epistemic version of the accessibility thesis does not entail the doxastic version, since one may have justification to believe a proposition without using it in acquiring justified belief; and likewise, one may have a way of knowing a proposition without using it in acquiring knowledge. Moreover, the epistemic version avoids the counterexamples to the doxastic version, so long as we assume that the limitations of one’s doxastic capacities do not impose corresponding limits on which propositions one has justification to believe.10 It may be that one has justification to believe a proposition even if one lacks the doxastic capacities required to use it in acquiring justified belief. Similarly, it may be that one has a way of knowing a proposition even if one lacks the doxastic capacities required to use it in acquiring knowledge. Psychological facts about one’s concepts, attention, powers of discrimination, and other rational capacities, need not constrain which propositions one has justification to believe, but only which of these justifications one has a capacity to use in forming justified beliefs. Similarly, these facts need not constrain which propositions one has a way of knowing, but only which of these ways of knowing one has a capacity to use in acquiring knowledge. On this view, one’s doxastic access to one’s mental states may be more limited than one’s epistemic access to one’s mental states in ways that reflect contingent facts about the limitations of one’s doxastic capacities. This is because the epistemic version of the accessibility thesis articulates an epistemic ideal, which abstracts away from contingent facts about the limitations of one’s doxastic capacities. If there are no doxastic limits on which propositions one has justification to believe, then the epistemic version of the accessibility thesis needs no further restriction in order to avoid counterexamples involving conceptual poverty, inattention, or misclassification.11 These are simply cases in which one lacks the capacity to use one’s introspective justification in forming introspectively justified beliefs. As a result, the epistemic version of the accessibility thesis applies more generally than doxastic versions, which do need further restriction. Moreover, as

10 For more detailed discussion and defense of this assumption, see Smithies (forthcoming, b and c). 11 Perhaps there are weak doxastic conditions to the effect that one has justification to believe a proposition only if one has some doxastic capacities to form, maintain, and revise beliefs. If so, then the epistemic version of the accessibility thesis can be restricted to subjects that satisfy those weak doxastic conditions.

A Simple Theory of Introspection

267

I will argue in what follows, the epistemic version is more fundamental in the order of philosophical explanation, since it explains the plausibility of various different strategies for imposing restrictions on the doxastic version of the accessibility thesis. One influential strategy is to abandon the self-intimation thesis altogether and to restrict the infallibility thesis to a class of demonstrative beliefs about one’s phenomenal states. Arguably, some beliefs about phenomenal states are infallible because they self-ascribe a phenomenal state under a demonstrative mode of presentation that cannot be entertained unless one is in the phenomenal state in question.12 This motivates the following restriction on the doxastic version of the infallibility thesis: The restricted infallibility version: for some mental states M, necessarily, if one believes that one is in M under a demonstrative mode of presentation, then one is in M. However, this restricted version of the doxastic infallibility thesis can be explained as a consequence of the epistemic accessibility thesis. If there are some demonstrative propositions about one’s phenomenal states that cannot be believed unless they are true, then it is equally plausible that they cannot be believed unless one has introspective justification to believe them. The assumption here is that the conditions that enable one to believe a demonstrative proposition of the relevant kind are also sufficient to provide one with introspective justification to believe it.13 But if the epistemic accessibility thesis is true, then one has introspective justification to believe a proposition only if it is true. Therefore, the infallibility of demonstrative beliefs about one’s phenomenal states may be explained as a consequence of the infallibility of introspective justification together with the existence of epistemic constraints on demonstrative concepts and demonstrative beliefs about one’s phenomenal states. Another influential strategy is to avoid counterexamples by embedding the doxastic version of the accessibility thesis within the scope of a conditional, which restricts its application to subjects that have the requisite concepts, attention, and rationality:

12

Chalmers (2003) argues that all direct phenomenal beliefs are true, where a direct phenomenal belief is partly constituted by the phenomenal state that it picks out. Similarly, Horgan and Kriegel (2007) argue that all bracketed phenomenal beliefs are true, where a bracketed phenomenal belief is one that brackets out all relational information about the phenomenal state that it picks out. 13 Compare Chalmers’s (2003, 249) justification thesis:“When a subject forms a direct phenomenal belief based on a phenomenal quality, then that belief is prima facie justified by virtue of the subject’s acquaintance with that quality.”

268

Constitutivism

The conditional version: for some mental states M, necessarily, if one is conceptually competent, attentive, and rational, then one is in M if and only if one believes that one is in M.14 However, these restrictions are not sufficient to avoid cases of misclassification unless we assume a highly idealized conception of rationality, according to which the rational subject believes that she is in a certain mental state if and only she has introspective justification to believe that she is in that mental state. For instance, the victim of the initiation trick may be rational by ordinary standards, but he is not ideally rational, since he does not believe what he has introspective justification to believe—namely, that he feels cold, rather than pain. But if the epistemic accessibility thesis is true, then one has introspective justification to believe that one is in a certain mental state if and only if one is in that mental state. Therefore, it follows that if one is ideally rational, then one believes that one is in a mental state if and only if one is in that mental state. Hence, the conditional version of the doxastic accessibility thesis can be explained as a consequence of the epistemic accessibility thesis. Another reaction is to interpret the conditional version of the doxastic accessibility thesis in such a way that it allows for exceptions.15 If we assume an ordinary conception of normal human rationality, rather than a more highly idealized conception, then it may be that rationality imposes limits on the possibility of ignorance and error about one’s own mental states without requiring absolute omniscience or infallibility. Once again, however, this weakened version of the doxastic accessibility thesis can be explained by appeal to the accessibility thesis. Arguably, one’s concepts of one’s own mental states are epistemically individuated in the sense that one possesses those concepts if and only if one is disposed to use them in forming introspectively justified beliefs about one’s mental states.16 However, these possession conditions do not require achieving the rational ideal of omniscience and infallibility about one’s own mental states, but merely some degree of approximation toward the rational ideal. Therefore, the limits on the possibility of ignorance and error about one’s mental states can be explained by 14 Compare Shoemaker, who writes, “if one has an available first-order belief, and has a certain degree of rationality, intelligence, and conceptual capacity (here including having the concept of belief and the concept of oneself), then automatically one has the corresponding second-order belief ” (1996, 288). 15 This may be what Shoemaker intends when he writes,“We should be wary of inferences from what can happen occasionally to what can happen as a matter of course: it may be true in Lake Wobegon that all of the children are above average, but it can’t be true everywhere” (1996, 227). 16 Peacocke (1992, ch. 6) claims that one possesses the concept of belief only if one finds it primitively compelling to judge that one believes a proposition because one does in fact believe it.

A Simple Theory of Introspection

269

the self-intimation and infallibility of introspective justification together with the existence of epistemic constraints on possessing concepts of one’s mental states. Epistemic and doxastic versions of constitutivism stand in need of different kinds of explanation. Doxastic versions of constitutivism demand explanation in terms of psychological claims about the nature of our mental states, the nature of our beliefs about those mental states, and the nature of concepts, attention, and normal human rationality. For instance, doxastic versions of constitutivism are sometimes explained by appeal to part–whole relations of constitution that hold between introspective beliefs and the mental states they are about.17 Epistemic versions of constitutivism, on the other hand, should be explained in terms of epistemological, rather than psychological, claims. In particular, the epistemic version of the accessibility thesis is best explained by the simple theory of introspective knowledge and introspective justification. The simple theory is a theory of the epistemology of introspection, rather than the psychology of introspection. It is an account of the nature of introspective justification, which makes introspective knowledge possible. It is not an account of the psychological mechanisms or processes that we use in taking advantage of introspective justification and thereby acquiring introspective knowledge. As such, it makes no commitments about the reliability or unreliability of our beliefs about our own mental states. As far as the simple theory is concerned, it is an open question to what extent and by what mechanisms or processes we take advantage of what we have introspective justification to believe.18 As we have seen, however, the simple theory raises questions of its own. A full defense of the simple theory requires answering the generalization question and the explanatory question: Which mental states are introspectively accessible and what explains why those mental states are introspectively accessible, rather than others, or none at all? My aim in the remaining sections of this chapter is to provide answers to these questions.

17 Shoemaker (1996, this volume) claims that second-order beliefs are partially constituted by the mental states they are about. Gertler (2001, this volume) claims that introspecting phenomenal states involves embedding the introspected state within an introspective belief. Chalmers (2003) claims that direct phenomenal beliefs are partially constituted by the phenomenal states they pick out. Horgan and Kriegel (2007) argue that phenomenal states are partially constituted by proto-beliefs, which are converted into bracketed phenomenal beliefs by the allocation of introspective attention. 18 The simple view is therefore consistent with Schwitzgebel’s (2008, this volume) claims about the unreliability of introspection and the absence of any dedicated mechanism of introspection.

270

Constitutivism

3. The Generalization Question Not all mental states are introspectively accessible. One’s mental states, by any noncontroversial criterion, include “subdoxastic” mental representations, which play a role in computational explanations in cognitive science—for instance, Noam Chomsky’s (1957) tacit knowledge of syntactic rules and David Marr’s (1982) primal and 2.5D sketches.19 Clearly, however, our justification to believe in the existence of these subdoxastic mental representations derives from scientific theory, rather than introspection. Therefore, we need some restriction on which mental states are introspectively accessible. An initially tempting criterion is that a mental state is introspectively accessible if and only if it is phenomenally conscious. However, this criterion is too restrictive, since it excludes not only subdoxastic states but also beliefs. Beliefs are standing states, which persist through time without making any ongoing contribution to one’s phenomenology. For instance, my belief that Canberra is the capital of Australia persists whether or not I am consciously considering the matter and so does my secondorder belief that I believe that Canberra is the capital of Australia. If these beliefs are justified, as we may assume, then we can ask what makes them justified at any given time. Crucially, however, there may be nothing in my current stream of phenomenal consciousness that makes them justified. In some cases, my beliefs may be justified by their inferential relations to other standing beliefs, but this is not always plausible, since I often forget the grounds on which my beliefs were originally formed.20 In that case, the only plausible candidate for what makes my second-order belief justified is the mere presence of my first-order belief. In other words, I have introspective justification to believe that I believe that Canberra is the capital of Australia just by virtue of believing it.21

19 Stich (1978, 499) defines subdoxastic states as “psychological states that play a role in the proximate causal history of beliefs, though they are not beliefs themselves.” We should add that these states are phenomenally unconscious in order to rule out perceptual experiences as examples of subdoxastic states. 20 See Conee and Feldman (2001) for a discussion of the so-called ‘problem of forgotten evidence’. 21 Compare Zimmerman (2006, 357–361) and Shoemaker (this volume). See also Peacocke’s socalled NICS cases, in which standing first-order beliefs justify occurrent second-order judgments in the absence of any intermediate conscious state: “Most of us, when it becomes conversationally appropriate to say ‘I know my name is NN’, or ‘I know my address is such-and-such’, have no need to wait upon its surfacing in consciousness what our names and addresses are.We make these utterances intentionally and knowledgeably, but not because it has just occurred to us that our names and addresses are such-and-such” (1998, 91).

A Simple Theory of Introspection

271

Any plausible answer to the generalization question must therefore be permissive enough to include beliefs, but also restrictive enough to exclude subdoxastic states. What we need, then, is a criterion that explains what beliefs, unlike subdoxastic states, have in common with phenomenally conscious states in virtue of which they are introspectively accessible. Richard Rorty (1979, 22) expresses pessimism about the prospects for any unified account of beliefs and phenomenally conscious states when he writes,“The attempt to hitch pains and beliefs together seems ad hoc—they don’t seem to have anything in common except our refusal to call them ‘physical’.” In what follows, however, I will suggest that this pessimism is unfounded. One strategy for answering the generalization question is to propose some broadly functionalist criterion, according to which a mental state is introspectively accessible if and only if it plays a certain functional role. The challenge for proponents of this strategy is to identify some relevant functional similarity between beliefs and phenomenally conscious states that does not also include subdoxastic states. For instance, one might argue that beliefs are like phenomenally conscious states and unlike subdoxastic states insofar as they satisfy Ned Block’s functional criteria for access consciousness: A state is access conscious if it is poised for direct control of thought and action. To add more detail, a representation is A-conscious if it is poised for free use in reasoning and for direct “rational” control of action and speech. (1997, 382) Since beliefs are typically access conscious, but not phenomenally conscious, one might answer the generalization question by replacing the appeal to phenomenal consciousness with access consciousness. On this proposal, a mental state is introspectively accessible if and only if it is access conscious.22 Counterexamples emerge when we consider hypothetical examples of subdoxastic states, which are access conscious, but not phenomenally conscious. In actual cases of blindsight, visual information is neither phenomenally conscious nor access conscious, since it is not poised for the direct control of speech and action except in forced choice conditions. In Block’s (1997, 385–386) hypothetical case of superblindsight, by contrast, visual information is access conscious, since it is poised for spontaneous use in the control of action and speech, although it is not phenomenally

22

Shoemaker (this volume) appeals to availability, which is closely related to access consciousness. Compare Zimmerman (2006, 357), who writes: “It is at least difficult to imagine a subject who believes that p in a fully A-conscious manner—where this belief is poised to guide her inferences and behaviour in all the customary ways—and who is also caused to believe that she believes that p by the fact that she believes that p, but who is nevertheless unjustified in believing that she has this belief. And this lends considerable prima facie support to the claim that A-consciousness is sufficient for direct accessibility.”

272

Constitutivism

conscious. Moreover, it is not introspectively accessible. Intuitively, the superblindsighter does not have introspective justification to form beliefs about what is represented in her visual system any more than the regular blindsighter does. At best, she has justification to make inferences about what is represented in her visual system from observational data about her own spontaneous verbal and nonverbal behavior. Therefore, access consciousness is not a sufficient condition for introspective accessibility. One reaction is to impose a more demanding functional criterion, according to which a mental state is introspectively accessible if and only if it is not merely access conscious, but also metacognitively conscious. A mental state M is metacognitively conscious if and only if the subject has formed (or is disposed to form) a noninferentially based higher-order thought that she is in M. According to higher-order theories of phenomenal consciousness, a mental state is phenomenally conscious if and only if it is metacognitively conscious.23 This gives rise to a dilemma for the current proposal. If higher-order theories are true, then metacognitive consciousness is not a necessary condition for introspective accessibility, since beliefs are introspectively accessible, but not metacognitively conscious. If higher-order theories are false, on the other hand, then metacognitive consciousness is not a sufficient condition for introspective accessibility, since there are hypothetical examples of subdoxastic states that are metacognitively conscious, but not introspectively accessible. Consider the case of hyper-blindsight, which is just like super-blindsight, except that the subject has a reliable mechanism that is disposed to generate noninferentially based higher-order thoughts about phenomenally unconscious visual states. Intuitively, the hyper-blindsighter does not have introspective justification to form beliefs about what is represented in her visual system any more than the superblindsighter or the regular blindsighter does. Certainly, she has a reliable disposition to form true beliefs about what is represented in her visual system, but this is not sufficient to make her beliefs introspectively justified. By analogy, the super-blindsighter has a reliable disposition to form true beliefs about stimuli in the blind hemi-field, but this is not sufficient to make them perceptually justified.24 Why suppose that the hyper-blindsighter’s noninferential beliefs about what is represented in her visual system are any more justified than the super-blindsighter’s noninferential beliefs about what is located in her blind hemi-field? In my view, there is no functional criterion for introspective accessibility that captures the distinction between beliefs and subdoxastic states except one that explicitly 23

Higher-order theories of consciousness are proposed in various forms by Armstrong (1968), Lycan (1997), Rosenthal (1997), and Carruthers (2000). 24 See Smithies (2011) for further discussion of the role of consciousness in perceptual justification.

A Simple Theory of Introspection

273

invokes relations to phenomenology. Beliefs are not phenomenally conscious states, but they are disposed to cause phenomenally conscious states of judgment. Moreover, I claim that beliefs are individuated wholly by their disposition to cause phenomenally conscious states of judgment. Subdoxastic states may also be disposed to cause phenomenally conscious states, but they are not individuated wholly by such dispositions. On the contrary, they are individuated in part by their disposition to play a role in computational processes that occur below the level of phenomenal consciousness. Let us say that a mental state is phenomenally individuated if and only if it is wholly individuated by its relations to phenomenally conscious states.25 My proposal is that a mental state is introspectively accessible if and only if it is phenomenally individuated. In other words, my proposal is that phenomenal individuation is the criterion that explains why beliefs are introspectively accessible, whereas subdoxastic states are not. Subdoxastic states are not individuated wholly by dispositions to cause phenomenally conscious states. Consider Martin Davies’s (1989) hypothetical example in which subdoxastic states that embody tacit knowledge of syntax are disposed to cause phenomenally conscious itches or tickles. These states are not individuated by their disposition to cause itches and tickles, but rather by their roles in syntactic processing. Similarly, mental representations in the computational theory of vision are not individuated solely by their disposition to cause phenomenally conscious visual experiences, but rather by their roles in visual processing. In principle, there could be subjects with exactly the same visual experiences, which are produced by different computational processes defined over different sets of subdoxastic visual states. Similarly, there could be subjects with exactly the same linguistic performance, and the same conscious states of linguistic understanding, but which are produced by different kinds of semantic and syntactic processing.26 Beliefs, on the other hand, are individuated wholly by their disposition to cause phenomenally conscious states of judgment. Judgment, as I use the term, is a phenomenally conscious state: there is something it is like to judge that a proposition is true. Moreover, the phenomenology of judgment is, in David Pitt’s (2004) terminology, proprietary, distinctive, and individuative.27 It is proprietary in the sense what

25 Individuation, as I am using the term, is closely tied to the notion of essence: what individuates a thing is what makes it what it is, but the essence of a thing is what makes it what it is, so the essence of a thing is what individuates it. For example, a set is individuated by its members in the sense that the essence of a set is exhausted by its members. So, to say that a mental state is phenomenally individuated is to say that its essence is exhausted by its relations to phenomenally conscious states. 26 The point is familiar from the literature on Quine’s (1970) challenge to Chomsky’s (1957) notion of tacit knowledge; see, for instance, Evans (1981) and Davies (1987). 27 Compare Strawson (1994, ch. 1), Siewert (1998, ch. 8), Horgan and Tienson (2002), and Pitt (2004).

274

Constitutivism

it’s like to judge a proposition is different from what it’s like to adopt a different attitude toward the same proposition; in other words, its phenomenology is attitudespecific. It is distinctive in the sense that what it’s like to judge a proposition is different from what it’s like to adopt the same attitude toward a different proposition; in other words, its phenomenology is content-specific. Moreover, it is individuative in the sense that judgment is individuated wholly by its content-specific and attitude-specific phenomenology.28 Beliefs, unlike judgments, are not phenomenally conscious states, but they are individuated wholly by their disposition to cause the proprietary, distinctive, and individuative phenomenology of judgment. A mental state counts as a belief by virtue of its disposition to cause phenomenally conscious states with the proprietary phenomenology that is specific to judgment. Moreover, a mental state counts as a belief in some specific propositional content by virtue of its disposition to cause phenomenally conscious states with the distinctive phenomenology that is specific to judging that content. The essence of belief is exhausted by its disposition to cause one to judge its propositional content: in other words, one believes a proposition if and only if (and in virtue of the fact that) one is disposed to judge the proposition in question.29 If belief is a disposition to judge, then not all judgments express beliefs and not all beliefs express themselves in judgment. A judgment does not express belief unless it manifests a stable disposition and a belief does not express itself in judgment unless a stable disposition is manifested. Nevertheless, it may be objected that a stable disposition to judge a proposition is neither necessary nor sufficient for believing it, since one’s dispositions toward judgment sometimes fail to reflect what one really believes. Christopher Peacocke (1998, 90) gives the example of an academic on a hiring committee who is disposed to judge on good evidence that foreign degrees are equal in standard to domestic degrees, although her votes in hiring decisions reveal that she does not really believe this; indeed, what she really believes is that

28

The phenomenology of judgment, like the phenomenology of perceptual experience, is specific with respect to narrow content, which is determined wholly by phenomenology, but not wide content, which is determined in part by relations to the environment. In my view, judgments are individuated by their narrow contents, but not their wide contents. See Horgan and Tienson (2002) and Chalmers (2004) for further discussion of the relationship between phenomenology and intentionality. 29 Searle (1990) argues that beliefs are accessible to consciousness in the sense that they are potentially conscious. However, beliefs are not potentially conscious: they are dispositions to cause conscious states, which are distinct from their potentially conscious manifestations. The contents of beliefs are accessible to consciousness as the contents of judgments, but this is a consequence of the fact that beliefs are individuated by the disposition to cause judgments with the very same contents.

A Simple Theory of Introspection

275

foreign degrees are inferior. In my view, however, Peacocke’s example is misdescribed: the academic does not really believe that foreign degrees are inferior, although she acts as if she believes this. What motivates this redescription is that her irrationality is practical, not epistemic: it stems from the subject’s actions, rather than her beliefs. After all, she knows full well that foreign degrees are equal in standard to domestic degrees: we may assume that she has studied the evidence carefully and formed her beliefs in a way that is appropriately responsive to the evidence. If so, then her rational failing is not that she has unjustified beliefs, but that her justified beliefs fail to exert an appropriate influence on her actions, which are consequently unjustified. This is not to deny that Peacocke’s academic has an attitude toward the proposition that foreign degrees are inferior to domestic degrees, which is functionally individuated by its role in disposing her to treat them as such. Rather, it is to deny that such an attitude is correctly described as ‘belief ’. Of course, there is some risk of descending into purely terminological dispute here, but the substantial point is that if one’s dispositions to judge a proposition come apart from one’s dispositions to act as if it is true, then our epistemic evaluations are sensitive to the former, but not the latter.30 In fact, the key point can be made without relying on any terminological claims about the correct use of the term ‘belief ’, since the case can be described in more neutral terms by employing David Chalmers’s (2011) subscript strategy in which the disputed term is eliminated and replaced by subscripted variants. Let us say that Peacocke’s academic believes1 that foreign degrees are equal, although she believes2 that foreign degrees are inferior. We can add that her belief1 is phenomenally individuated by its disposition to cause phenomenally conscious states of judgment, whereas her belief2 is individuated by its disposition to cause actions, which conflict with her best judgments. Now we can ask whether Peacocke’s academic has introspective access to these mental states. Intuitively, she has introspective justification to believe that she believes1 that foreign degrees are equal, since this is what she is disposed to judge.31 However, she does not have introspective justification to

30 There are at least four possible descriptions of Peacocke’s academic: (1) she believes that foreign degrees are inferior and does not believe that they are equal (Peacocke 1998); (2) she believes that foreign degrees are equal and does not believe that they are inferior (Zimmerman 2007, Gendler 2008); (3) she believes both (Shoemaker, this volume); or (4) she does not determinately believe either (Schwitzgebel 2010). My claim is that only the second option does justice to the agent’s epistemic rationality. See Zimmerman (2007) and Gendler (2008) for related discussion. 31 What if the disposition is blocked by drunkenness or temporary insanity? I maintain that the belief is introspectively accessible in the sense that the subject has introspective justification to believe that she has this belief, although she may be unable to take advantage of her justification unless the blockage is removed. In that case, the blockage constitutes a departure from full rationality.

276

Constitutivism

believe that she believes2 that foreign degrees are inferior, since she has no disposition to judge this. If she knows what she believes2, then she knows in the same way that others do—namely, by inference to the best explanation.32 Therefore, whether we are restrictive or permissive in our use of the term ‘believes’, we may conclude that a mental state is introspectively accessible if and only if it is phenomenally individuated. My proposal relies on the assumption that the phenomenology of judgment is proprietary, distinctive, and individuative. But why should we accept this controversial assumption? David Pitt (2004) argues that we need to make this assumption in order to explain our introspective access to our own phenomenally conscious thoughts and judgments.33 We can restate his argument as follows: 1. It is possible to know by introspection what one is thinking and judging. 2. If it is possible to know by introspection what one is thinking and judging, then one’s thoughts and judgments have proprietary, distinctive, and individuative phenomenology. 3. Therefore, one’s thoughts and judgments have proprietary, distinctive, and individuative phenomenology. The first premise is relatively uncontroversial, while the second premise is motivated by generalization from other cases. For instance, it is plausible that the phenomenology of perceptual experience explains how introspective knowledge of perceptual experience is possible. By analogy, it is plausible that the phenomenology of judgment explains how introspective knowledge of judgment is possible. The problem with Pitt’s argument is that the analogy breaks down in the case of belief, since we have introspective knowledge of belief, but there is no phenomenology of belief that explains how introspective knowledge is possible. Therefore, the second premise cannot be motivated by a general principle to the effect that introspective knowledge of attitudes is possible only if those attitudes have proprietary, distinctive, and individuative phenomenology. Moreover, this raises an objection to the second premise, since it remains to be seen whether introspective knowledge of judgment should be explained on the model of perceptual experience, rather than the model of belief. However, if the arguments of this section are successful, then we should understand introspective knowledge of belief on the

32 As Brie Gertler pointed out to me, the inference may be based on introspective knowledge of the results of an imaginative simulation exercise, rather than observation of one’s behavior. In that case, according to the simple theory, the premises of the inference are known by introspection alone, but not the conclusion. 33 Goldman (1993) gives a related style of epistemological argument for cognitive phenomenology.

A Simple Theory of Introspection

277

model of introspective knowledge of judgment, rather than vice versa. What I have argued, in effect, is that introspective knowledge of belief is possible if and only if belief is phenomenally individuated by its disposition to cause the proprietary, distinctive, and individuative phenomenology of judgment. This raises a challenge for the opponents of cognitive phenomenology. If judgments do not have phenomenology that is proprietary, distinctive, and individuative, then what explains why one’s beliefs are introspectively accessible, whereas one’s subdoxastic states are not?

4. Some Problem Cases I have argued that a mental state is introspectively accessible if and only if it is phenomenally individuated. We can formulate this thesis more precisely as follows: The phenomenal accessibility thesis: for all mental states M, necessarily, M is phenomenally individuated if and only if the following condition holds: one is in M if and only if one has introspective justification to believe that one is in M and one thereby has an introspective way of knowing that one is in M. In this section, I will provide further support for the phenomenal accessibility thesis by defending it against a range of objections. First, one might deny that the subject matter of introspection is restricted to one’s mental states, as opposed to one’s physical states. On this view, one has introspective justification to believe various propositions about one’s own body. For instance, the experience of proprioception might provide introspective justification to believe that one’s legs are crossed and the experience of agency might provide introspective justification to believe that one is raising one’s arm. If so, then in cases of illusion, one has introspective justification to believe that one’s legs are crossed, although it merely seems as if one’s legs are crossed; or one has introspective justification to believe that one is raising one’s arm, although it merely seems as if one is raising one’s arm. Therefore, we face counterexamples to the infallibility thesis in which one has introspective justification to believe false propositions. On the simple theory, the experience of proprioception or agency does not provide introspective justification to believe propositions about one’s own body. Certainly, one’s justification is peculiar in the sense that it is not available to others, but it is not privileged in the way that is characteristic of introspective justification, since the source of one’s justification is not identical with its subject matter. After all, it is not the case that one has justification to believe that one’s body is in a certain state just by virtue of its being in that state. Rather, one has justification to believe

278

Constitutivism

that one’s body is in a certain state by virtue of having an experience, which represents that the world is in that state. In cases of illusion, one’s experience misrepresents the world and so one has justification to believe falsely that the world is in that state. As such, these are cases of perceptual justification, rather than introspective justification. On the simple theory, one has introspective justification to believe propositions about one’s experiences of proprioception and agency. For instance, one has justification to believe that it seems as if one’s legs are crossed or that it seems as if one is raising one’s arms just by virtue of the fact that it seems that way. Arguably, one also has nonintrospective justification—perhaps a priori justification—to believe that one’s experience represents veridically and hence that things are the way they seem. On this view, one has fallible justification to believe that one is in bodily state B, which derives from one’s infallible introspective justification to believe that one is in some phenomenally individuated mental state M together with one’s fallible nonintrospective justification to believe that if one is in M, then one is in B. Therefore, one’s justification depends partly, but not wholly, upon introspection.34 A similar response can be given to those who deny that the subject matter of introspection is restricted to phenomenally individuated mental states, as opposed to externally individuated mental states. The phenomenology of visual experience does not provide fallible introspective justification to believe that one sees a cup or that one sees that the cup is white. Rather, one has introspective justification to believe that one is in a certain kind of phenomenally individuated mental state—namely, it seems as if one sees a cup or it seems as if one sees that the cup is white. Moreover, one has nonintrospective justification to believe that one’s experience represents veridically and hence that things are as they seem. In other words, one’s fallible justification to believe that one is in some externally individuated mental state M derives from one’s infallible introspective justification to believe that one is in some phenomenally individuated mental state M together with one’s fallible nonintrospective justification to believe that if one is in M, then one is in M. Again, one’s justification depends partly, but not wholly, upon introspection. On this view, the phenomenology of judgment cannot provide introspective justification to believe that one is judging the externally individuated content that water is wet. Rather, one has introspective justification to believe that one is judging

34 In the terminology of mediate versus immediate justification: one does not have immediate introspective justification to believe propositions about one’s own body, since one’s justification is mediated by nonintrospective justification to believe other propositions. I remain neutral on the further question whether one’s perceptual justification to believe propositions about one’s own body is mediate or immediate.

A Simple Theory of Introspection

279

some phenomenally individuated content—for instance, that watery stuff is wet— but one also has nonintrospective justification to believe that if one is thinking that watery stuff is wet, then one is thinking that water is wet. According to phenomenal mentalism, which propositions one has justification to believe is determined by one’s phenomenally individuated mental states. If so, then my phenomenal duplicate on twin earth has justification to believe that he is thinking that water is wet, which is false, since he is not thinking about water, but twin water. As before, however, this fallible justification derives from his infallible introspective justification to believe that he is thinking that watery stuff is wet and his fallible nonintrospective justification to believe that if he is thinking that watery stuff is wet, then he is thinking that water is wet. So, this case does not provide a counterexample to the infallibility of introspective justification. Nicholas Silins (this volume) argues that there are counterexamples to the infallibility of introspective justification in which one judges a proposition and so one has introspective justification to believe that one believes it, although one does not believe it because one’s judgment fails to manifest a stable disposition of the right kind. Here too, I claim that one’s justification depends partly, but not wholly, upon introspection. If one judges a proposition, then one has introspective justification to believe that one judges it. Moreover, one has nonintrospective justification— perhaps a priori justification—to believe that if one judges a proposition, then one believes it. So, if one judges a proposition, then one has fallible justification to believe that one believes it, which derives from one’s infallible introspective justification to believe that one judges it together with one’s fallible nonintrospective justification to believe that if one judges it, then one believes it. However, one has introspective justification to believe that one believes a proposition if and only if one believes it.35

5. The Explanatory Question In sections 3 and 4, I defended an answer to the generalization question: a mental state is introspectively accessible if and only if it is phenomenally individuated. This raises a version of the explanatory question: Why are phenomenally individuated mental states introspectively accessible, rather than others, or none at all? In order to explain which mental states are introspectively accessible, we need to invoke distinctive features of those mental states together with more general claims

35

See Silins (this volume) for further discussion.

280

Constitutivism

in the theory of justification, which explain why all and only mental states with those distinctive features are introspectively accessible. In this section, I consider and reject two reductionist strategies—namely, reliabilism and phenomenal conservatism— which attempt to answer the explanatory question by assimilating introspective justification to a more general model that applies equally to other nonintrospective kinds of justification. The simple theory, by contrast, is primitivist in the sense that it regards introspective justification as a primitive or sui generis form of justification, which cannot be assimilated to a more general model that applies elsewhere. In the next section, however, I will develop an alternative strategy for answering the explanatory question that is consistent with the primitivism of the simple theory. According to reliabilism, which propositions one has justification to believe is explained by the reliability of one’s doxastic mechanisms.36 What explains why one has justification to believe a proposition is that one has a doxastic mechanism, which is reliably disposed to yield belief in that proposition if and only if that proposition is true. In particular, what explains why one has introspective justification to believe a proposition is the fact that one has an introspective mechanism, which is reliably disposed to yield belief that one is in a mental state if and only if one is in that mental state.37 On this view, a mental state provides introspective justification to believe that one is in that mental state if and only if it is reliably tracked by an introspective mechanism. Therefore, an answer to the explanatory question takes the following form: 1. A mental state provides introspective justification to believe that one is in that mental state if and only if it is reliably tracked by an introspective mechanism. 2. A mental state is reliably tracked by an introspective mechanism if and only if it has feature F. 3. Therefore, a mental state provides introspective justification to believe that one is in that mental state if and only if it has feature F. However, this reliabilist answer to the explanatory question faces several problems. First, the existence of a reliable introspective mechanism is not sufficient to explain one’s introspective justification to form beliefs about one’s mental states, since there are well-known counterexamples to the claim that reliability is sufficient for justification. For instance, Norman has a reliable clairvoyant faculty, which reliably

36 I have formulated reliabilism in terms of doxastic mechanisms, but Goldman (1979) appeals to processes, Nozick (1981) appeals to methods, and Sosa (this volume) appeals to competences. These differences in formulation are not crucial for current purposes. 37 Examples of introspective mechanisms include Armstrong’s (1968) self-scanning mechanism and Nichols and Stich’s (2003: 160–164) monitoring mechanism.

A Simple Theory of Introspection

281

causes him to form true beliefs about the location of the president; Mr.Truetemp has a tempucomp implanted in his brain, which reliably causes him to form true beliefs about the temperature; the super-blindsighter has unconscious visual information, which reliably causes him to form true beliefs about objects in his blind hemi-field.38 In each of these cases, the subject’s beliefs are unjustified, but they are formed on the basis of a doxastic mechanism that is in fact reliable, although the subject in question does not know or have justification to believe that it is reliable. However, if mere reliability is not sufficient for justification in these cases, then why suppose that mere reliability about one’s own mental states is sufficient for introspective justification? Second, reliabilism is not sufficient to explain why introspective justification is infallible and self-intimated. If one’s introspective mechanisms are reliable, but not infallible, then there will be cases in which one has introspective justification to believe that one is in a certain mental state, although one is not in that mental state. Similarly, if one’s introspective mechanisms are reliable, but one’s mental states are not self-intimating, then there will be cases in which one is in a certain mental state, although one does not have introspective justification to believe that one is in that mental state. In order to rule out these possibilities, introspective justification must be taken to require not just the usual degree of reliability, but the highest degree of reliability, which is sufficient to ensure infallibility and self-intimation. However, reliabilism provides no principled motivation for imposing such a demanding requirement on introspective justification. Third, reliabilism is not sufficient to explain the modal status of the introspective accessibility thesis. Reliabilism offers no modal guarantee that one has introspective justification to believe that one is in a certain mental state if and only if one is in that mental state. On the contrary, whether or not one has introspective justification to believe that one is in a certain mental state depends entirely on contingent empirical facts about one’s psychology—namely, whether or not one has a reliable introspective mechanism of the relevant kind. According to reliabilism, then, it is a purely contingent matter which mental states, if any, provide one with introspective justification to believe that one is in those mental states. Therefore, we must look elsewhere in order to find a theory of justification that is capable of explaining the modal status of the introspective accessibility thesis.39

38 These examples are from BonJour (1985, 41), Lehrer (1990, 163–164), and Block (1997, 385–386). 39 For related criticisms of reliabilism, see Shoemaker’s (1996) arguments against the broad perceptual model and Burge’s (1996) arguments against the simple observational model, which hinge on the claim that they cannot explain the limits on the possibility of ignorance and error about one’s own mental states.

282

Constitutivism

According to phenomenal conservatism, which propositions one has justification to believe is explained by one’s nondoxastic, phenomenal seemings.40 What explains why one has justification to believe a proposition is the fact that it seems to be true and it is undefeated by any contrary seemings. On this view, seemings are distinct from beliefs: in some cases, a proposition seems true although one does not believe it. Moreover, seemings are distinct from conscious inclinations to believe: in many cases, it is because a proposition seems true that one has a conscious inclination to believe it. Thus, seemings may be said to comprise a sui generis category of conscious states, although they come in many different subcategories, including perceptual, memorial, and intellectual seemings. Phenomenal conservatism motivates a version of the inner sense theory on which introspective justification is explained by reference to a special category of introspective seemings.41 On this view, one has introspective justification to believe that one is in a certain mental state if and only if it introspectively seems that one is in that mental state. Therefore, an answer to the explanatory question takes the following form: 1. A mental state provides introspective justification to believe that one is in that mental state if and only if it introspectively seems that one is in that mental state. 2. A mental state is such that it introspectively seems that one is in that mental state if and only if it has feature F. 3. Therefore, a mental state provides introspective justification to believe that one is in that mental state if and only if it has feature F. However, this answer to the explanatory question also faces several problems. First, introspective seemings are epistemologically redundant. Consider my belief that I am in pain. If my belief is to be justified, then there must be some conscious state, which is distinct from my belief that I am in pain, but which explains and justifies my belief that I am in pain. Plausibly, however, this conscious state is just my pain. Why should we suppose that there is any further state of introspective seeming, which is distinct from my pain, but which explains and justifies my belief that I am in pain? Why isn’t my pain sufficient by itself to explain and justify my belief that I am in pain? Second, introspective seemings are phenomenologically redundant. According to some theories of phenomenal consciousness, a mental state is phenomenally conscious

40 Huemer’s (2001, 99) rule of phenomenal conservatism states that: “If it seems to S as if p, then S thereby has at least prima facie justification for believing that p.” Huemer (2007) also endorses the right-to-left conditional, which plays a role in his self-defeat argument for phenomenal conservatism. 41 Huemer (2007, 30) includes “apparent introspective awareness” in his inventory of seemings.

A Simple Theory of Introspection

283

if and only if it seems that one is in it.42 Arguably, however, this is either trivial or false. There is a trivial sense in which I am in pain if and only if seems to me that I am in pain. After all, pain is an experience; moreover, I experience my experiences in the same way that I smile my smiles and dance my dances. However, this does not imply that I represent my own experiences in the way that my experiences represent the world. There are no obviously compelling grounds to suppose that I represent my own experiences; indeed, there are some compelling phenomenological grounds to suppose otherwise.43 Third, if we assume that all and only phenomenally conscious states are accompanied by introspective seemings, then phenomenal conservatism generates an overly restrictive answer to the generalization question. After all, beliefs are not phenomenally conscious states and so they are not accompanied by introspective seemings. Nevertheless, as I argued in section 3, the fact that one believes a proposition is sufficient to provide introspective justification to believe that one believes it. Therefore, we must look elsewhere for an answer to the explanatory question that generates a more plausible answer to the generalization question. What these explanatory strategies have in common is the reductionist ambition to explain which mental states provide introspective justification by appealing to a more general theory of justification, which applies equally to introspective and nonintrospective forms of justification. According to the simple theory, by contrast, introspective justification is primitive or sui generis in the sense that it cannot be assimilated to a more general theory of justification that applies elsewhere. Introspective justification is by nature the distinctive kind of justification that one has to believe that one is in a certain mental state, which one has just by virtue of being in that mental state. But if introspective justification is primitive or sui generis in this sense, then what resources do we have to answer the explanatory question? How can we explain why some mental states provide introspective justification, rather than others, or none at all? My aim in the following section is to develop an answer to the explanatory question that is consistent with the primitivism of the simple theory. My strategy is to argue for the conditional claim that if access internalism is true, then the accessibility of facts about justification is best explained by the introspective accessibility constraint, which states that the determinants of justification are introspectively accessible facts about one’s mental states. If the introspective accessibility constraint is true, then we can answer the explanatory question as follows: 42

This assumption is often used as a premise in arguing for higher-order theories of consciousness. See Armstrong (1968), Lycan (1997), Rosenthal (1997), Carruthers (2000), and Kriegel (2009). 43 See Shoemaker (1996, ch. 10) and Siewert (this volume) for arguments against inner sense.

284

Constitutivism

1. A mental state provides introspective justification to believe that one is in that mental state if and only if it is among the determinants of justification. 2. A mental state is among the determinants of justification if and only if it has feature F. 3. Therefore, a mental state provides introspective justification to believe that one is in that mental state if and only if it has feature F. Thus, we can explain our answer to the generalization question by appealing to a more general account of the determinants of justification. Alternatively, we can derive a more general account of the determinants of justification by appealing to our answer to the generalization question.

6. Access Internalism One dimension of the debate between internalism and externalism in epistemology concerns the nature and extent of one’s epistemic access to facts about which propositions one has justification to believe. Roughly speaking, access internalism is the thesis that one has privileged epistemic access to facts about which propositions one has justification to believe, whereas access externalism is the denial of access internalism. For current purposes, access internalism can be defined more precisely as the thesis that justification is accessible in the following sense: Access internalism: necessarily, one has justification to believe that p if and only if one has justification to believe that one has justification to believe that p. Access internalism, like the accessibility thesis discussed in section 2, is a biconditional that conjoins a self-intimation thesis, which is the conditional that goes from left to right, and an infallibility thesis, which is the conditional that goes from right to left. The self-intimation thesis rules out the possibility of justified ignorance about which propositions one has justification to believe, whereas the infallibility thesis rules out the possibility of justified error. In other words, ideal rationality, according to access internalism, involves omniscience and infallibility about which propositions one has justification to believe. Access internalism is a controversial thesis. For instance, reliabilism offers no modal guarantee that if one’s first-order doxastic mechanisms are reliable, then one’s second-order doxastic mechanisms are also reliable. So, if reliabilism is true, then access internalism is false, since it is possible that one has first-order justification to believe a proposition, although one lacks second-order justification to believe that one does.

A Simple Theory of Introspection

285

Why should anyone suppose that access internalism is true? After all, it is not true in general that a condition C is accessible in the sense that C obtains if and only if one has justification to believe that C obtains. So, why suppose that justificatory conditions are any different? Here, I will briefly sketch three lines of argument, which I develop in much more detail elsewhere. First, access internalism explains intuitions about cases. Second, access internalism solves an epistemic version of Moore’s paradox. And third, access internalism explains why justification is worth caring about, since it plays an important role in epistemic evaluation as an ideal of critical reflection.44 First, access internalism explains intuitions about cases, including clairvoyance and envatment.45 Intuitively, my envatted duplicate has justification to believe propositions on the basis of perceptual experience, just as I do, although his beliefs are unreliable, whereas mine are reliable. But what explains this intuition? My envatted duplicate has second-order justification to believe that he has justification to believe propositions on the basis of perceptual experience. So, by the infallibility thesis, he has first-order justification to believe propositions in this way. Conversely, my clairvoyant duplicate lacks justification to believe propositions on the basis of blind hunches or wishful thinking, just as I do, although his beliefs are reliable, whereas mine are unreliable. But what explains this intuition? My clairvoyant twin lacks second-order justification to believe that he has justification to believe propositions on the basis of blind hunches or wishful thinking. So, by the self-intimation thesis in its contrapositive form, he lacks first-order justification to believe propositions in this way. Second, access internalism solves an epistemic version of Moore’s paradox, since it explains what is wrong with believing Moorean conjunctions of the following forms: 1. 2. 3. 4.

p and it is not the case that I have justification to believe that p. p and it is an open question whether or not I have justification to believe that p. I have justification to believe that p and it is not the case that p. I have justification to believe that p and it is an open question whether or not p.

If access internalism is true, then one cannot have justification to believe these Moorean conjunctions because one cannot have justification to believe each conjunct

44 See Smithies (forthcoming, b and c). For simplicity, I focus here on arguments that the presence of justification is accessible, but similar arguments show that the absence of justification is also accessible. 45 See BonJour (1985, ch. 3) for the clairvoyance case and Cohen (1984) for the new evil demon case, which is a variation on the envatment case.

286

Constitutivism

simultaneously. If access internalism is false, on the other hand, then it is left open that one can have justification to believe Moorean conjunctions. If self-intimation is false, then one can have justification to believe a proposition, although one lacks higher-order justification to believe that one has justification to believe it. In that case, one has justification to disbelieve or to withhold belief that one has justification to believe the proposition in question, so one has justification to believe either (1) or (2). Similarly, if infallibility is false, then one can have higher-order justification to believe that one has justification to believe a proposition, although one lacks justification to believe it. In that case, one has justification either to disbelieve or to withhold belief in the proposition in question, so one has justification to believe either (3) or (4). Therefore, access internalism is needed in order to explain what is wrong with believing these Moorean conjunctions. Finally, access internalism explains why justification is an important epistemic property that is worth caring about. Justification plays an important role in epistemic evaluation as an ideal of critical reflection: it is the epistemic property in virtue of which a belief has what it takes to survive ideal critical reflection. Roughly, one has justification to believe a proposition if and only if one would believe it after ideal critical reflection.46 And yet one would not believe a proposition after ideal critical reflection unless one has second-order justification to believe that one has first-order justification to believe it. After all, the whole point of critical reflection is to bring one’s beliefs into line with one’s reflections about which propositions one has justification to believe. Therefore, ideal critical reflection involves believing a proposition if and only if one has second-order justification to believe that one has first-order justification to believe it. Hence, the role of justification in epistemic evaluation as an ideal of critical reflection provides the basis of an argument that justification is accessible: 1. Necessarily, one has justification to believe that p if and only if one would believe that p after ideal critical reflection. 2. Necessarily, one would believe that p after ideal critical reflection if and only if one has justification to believe that has justification to believe that p. 3. Necessarily, one has justification to believe that p if and only if one has justification to believe that one has justification to believe that p. In sum, there are compelling arguments for access internalism. Still, it is one thing to argue that access internalism is true, but it is another thing to explain what makes

46 This needs further refinement to deal with conditional fallacy objections. See Smithies (forthcoming, c) for details.

A Simple Theory of Introspection

287

it the case that it is true. If access internalism is true, then part of the job description for a theory of justification is to give an account of the determinants of justification, which explains why access internalism is true. There are no brute facts about which propositions one has justification to believe. On the contrary, these epistemic facts are determined by non-epistemic facts in the sense that no two situations can differ with respect to which propositions one has justification to believe unless those epistemic differences are explained by, or grounded in, corresponding non-epistemic differences. The determinants of justification are the non-epistemic facts (whatever they are) that determine the epistemic facts about which propositions one has justification to believe. Arguably, the epistemic facts about which propositions one has justification to believe are determined a priori by non-epistemic facts. In other words, there are a priori conditionals of the form: necessarily, if such-and-such non-epistemic facts obtain, then such-and-such epistemic facts obtain. This is supported by the method of cases, in which a possible case is specified in non-epistemic terms and we can read off an epistemic specification without relying on any further empirical information. Given sufficient information about a case, it is a priori which propositions one has justification to believe in that case. If any further empirical information about the case is needed, then it can be included in a specification of the determinants of justification in which case the determination relation itself is a priori.47 One of the main tasks for a theory of justification is to specify the determinants of justification. According to reliabilism, for instance, the determinants of justification are non-epistemic facts about the reliability of one’s doxastic methods. However, access internalism constrains an account of the determinants of justification, since justification is accessible only if the determinants of justification are themselves accessible. After all, access to epistemic facts depends on access to nonepistemic facts that determine those epistemic facts.48 For instance, access internalism rules out a reliabilist account of the determinants of justification, since facts about the reliability of one’s doxastic mechanisms are not accessible in the sense that one has justification to believe that they are reliable if and only if they are in fact reliable. But then what are the determinants of justification? Which facts, if any, are accessible in the sense that one has justification to believe that they obtain if and only if they obtain?

47 See Chalmers and Jackson (2001, section 2) for related discussion of a priori entailment relations. 48 More precisely: for every accessible epistemic fact E, there is some non-epistemic fact N that determines E; however, this does not entail the implausible claim that every non-epistemic fact N that determines E is accessible.

288

Constitutivism

Access internalism provides the basis of an argument for a version of mentalism, according to which the determinants of justification are introspectively accessible facts about one’s mental states.49 The argument is as follows: 1. Justification is accessible. 2. Justification is accessible only if the determinants of justification are introspectively accessible facts about one’s mental states. 3. So, the determinants of justification are introspectively accessible facts about one’s mental states. Premise (2) is motivated by inference to the best explanation. If the determinants of justification are introspectively accessible facts about one’s mental states and facts about the determination relation are accessible by a priori reflection, then facts about which propositions one has justification to believe are accessible by means of a combination of introspection and a priori reflection. Moreover, there is no other plausible explanation available. Therefore, access internalism stands or falls with the thesis that the determinants of justification are introspectively accessible facts about one’s mental states. Here, in more detail, is how the accessibility of justification is to be explained. Suppose that one has justification to believe that p in virtue of the fact that one is in some mental state M. If M is introspectively accessible, then one has introspective justification to believe that one is in M. And if the determination relation is a priori accessible, then one has a priori justification to believe that if one is in M, then one has justification to believe that p. So, if one has justification to believe that p in virtue of being in M, then one has justification through a combination of introspection and a priori reflection to believe that one has justification to believe that p in virtue of being in M.50 If, on the other hand, M is not introspectively accessible, then it is not the case that if one is in M, then one has introspective justification to believe that one is in M. So, it is not the case that, if one has justification to believe that p in virtue of being in M, then one has justification on the basis of introspection and a priori reflection to believe that one does. In conclusion, if access internalism is true, then so is the introspective accessibility constraint, which states that the determinants of justification are introspectively accessible facts about one’s mental states. Of course, it is a further question which

49 Note that many epistemologists accept mentalism, but reject accessibilism, including Williamson (2000), Conee and Feldman (2001), and Wedgwood (2002). 50 Here I assume that one’s justification transmits across the entailment from (1) one is in M; and (2) if one is in M, then one has justification to believe that p; to (3) one has justification to believe that p.

A Simple Theory of Introspection

289

facts about one’s mental states are introspectively accessible. If the introspective accessibility constraint is true, however, then the answer to this generalization question constrains and is constrained by a more general account of the determinants of justification. This provides a strategy for answering the explanatory question. We can explain our answer to the generalization question by appealing to a more general theory of the determinants of justification. For instance, we can explain the phenomenal accessibility thesis by appealing to phenomenal mentalism: the thesis that the determinants of justification are phenomenally individuated mental states.51 Alternatively, we can motivate a more general theory of the determinants of justification by appealing to an independently motivated answer to the generalization question. For instance, we can argue for phenomenal mentalism by appealing to the phenomenal accessibility thesis. Moreover, we need not choose between these options, since coherentism is more plausible than foundationalism as an epistemology for philosophy. In conclusion, the following package of claims is best regarded as part of a coherent and mutually reinforcing theory of justification: 1. The introspective accessibility constraint: A mental state is introspectively accessible if and only if it is among the determinants of justification. 2. Phenomenal mentalism: A mental state is among the determinants of justification if and only if it is phenomenally individuated. 3. The phenomenal accessibility thesis: A mental state is introspectively accessible if and only if it is phenomenally individuated.

7. Rationality and Self-Knowledge What is the connection between rationality and introspective self-knowledge? According to the simple theory, there is a necessary connection between rationality and self-knowledge. If one is ideally rational, then one is omniscient and infallible about one’s phenomenally individuated mental states. In other words, introspective self-knowledge is a constitutive ideal of rationality. A recurring theme in the literature on self-knowledge is that there is a necessary connection between rationality and self-knowledge.52 And yet the existence of such

51

For more detailed discussion of phenomenal mentalism, see Smithies (forthcoming, d). Classic discussions of the relationship between rationality and self-knowledge include Burge (1996) and Shoemaker (1996), but see also the chapters by Moran, Shoemaker, and Stoljar (this volume) for further discussion. 52

290

Constitutivism

a connection stands in need of further explanation. Why does ideal rationality require one to be omniscient and infallible about one’s own mental states? Why can’t one be ideally rational and yet introspectively blind to one’s own mental states, just as one can be ideally rational and yet visually blind to the external world? My aim in this chapter has been to argue that the rational ideal of introspective self-knowledge is best explained as a consequence of access internalism in the theory of justification. Access internalism is the thesis that justification is accessible in the sense that one has justification to believe a proposition if and only if one has higher-order justification to believe that one does. If justification is accessible in this sense, then ideal rationality involves omniscience and infallibility about which propositions one has justification to believe. Moreover, I argued that justification is accessible in this sense only if the determinants of justification are phenomenally individuated mental states, which are introspectively accessible in the sense that one has introspective justification to believe that one is in a phenomenally individuated mental state if and only if one is in that mental state. Therefore, if justification is accessible, then ideal rationality involves omniscience and infallibility about one’s phenomenally individuated mental states, which determine which propositions one has justification to believe. Thus, the connection between rationality and self-knowledge is explained as a consequence of access internalism in the theory of justification combined with the claim that the determinants of justification are introspectively accessible facts about one’s mental states.53

References Armstrong, David. 1968. A Materialist Theory of the Mind. London: Routledge. Block, Ned. 1997. “On a Confusion about a Function of Consciousness.” In The Nature of Consciousness, edited by N. Block, O. Flanagan, and G. Guzeldere. Cambridge, Mass.: MIT Press. Bonjour, Laurence. 1985. The Structure of Empirical Knowledge. Cambridge, Mass.: Harvard University Press.

53

An early version of this chapter was presented for the Introspection and Consciousness Workshop at the Australian National University in October 2008. A more recent version was presented for my seminar on Introspection and Self-Knowledge at the Ohio State University in May 2010. Many thanks to the audiences on those occasions and to David Chalmers, Brie Gertler, Ole Koksvic, Ram Neta, Eric Schwitzgebel, Susanna Siegel, and especially Nico Silins, for helpful comments and discussion. Finally, special thanks to Daniel Stoljar for many hours of conversation about all the chapters of this volume. It has been a pleasure and an education to collaborate with him on this project.

A Simple Theory of Introspection

291

Burge, Tyler. 1996. “Our Entitlement to Self-Knowledge.” Proceedings of the Aristotelian Society 96: 91–116. Byrne, Alex. 2005. “Introspection.” Philosophical Topics 33: 79–104. Carruthers, Peter. 2000. Phenomenal Consciousness: A Naturalistic Theory. Cambridge: Cambridge University Press. Chalmers, David. 2003. “The Content and Epistemology of Phenomenal Belief.” In Consciousness: New Philosophical Perspectives, edited by Q. Smith and A. Jokic. Oxford: Oxford University Press. ———. 2004. “The Representational Character of Experience.” In The Future for Philosophy, edited by B. Leiter. Oxford: Oxford University Press. ———. 2011. “Verbal Disputes and Philosophical Progress.” Philosophical Review 120(4): 515–566. Chalmers, David, and Frank Jackson. 2001. “Conceptual Analysis and Reductive Explanation.” Philosophical Review 110: 315–361. Chomsky, Noam. 1957. Syntactic Structures. Hague: Mouton. Cohen, Stewart. 1984. “Justification and Truth.” Philosophical Studies 46: 279–295. Conee, Earl, and Richard Feldman. 2001. “Internalism Defended.” American Philosophical Quarterly 38: 1–18. Davies, Martin. 1987. “Tacit Knowledge and Semantic Theory: Does a Five Percent Difference Matter?” Mind 96: 441–462. ———. 1989. “Tacit Knowledge and Subdoxastic States.” In Reflections on Chomsky, edited by A. George. Oxford: Blackwell. Evans, Gareth. 1981.“Semantic Structure and Tacit Knowledge.” In Wittgenstein:To Follow a Rule, edited by S. Holtzman and C. Leich. London: Routledge and Kegan Paul. ———. 1982. The Varieties of Reference. Oxford: Oxford University Press. Gendler, Tamar. 2008. “Alief and Belief ” Journal of Philosophy 105: 634–663. Gertler, Brie. 2001. “Introspecting Phenomenal States.” Philosophy and Phenomenological Research 63(2): 305–328. Goldman, Alvin. 1979. “What is Justified Belief?” In Justification and Knowledge, edited by G. Pappas. Dordrecht: Reidel. ———. 1993.“The Psychology of Folk Psychology.” Behavioral and Brain Sciences 16: 15–28. Horgan, Terry, and Uriah Kriegel. 2007. “Phenomenal Epistemology: What Is Consciousness That We May Know It So Well?” Philosophical Issues 17: 123–144. Horgan, Terry, and John Tienson. 2002. “The Intentionality of Phenomenology and the Phenomenology of Intentionality.” In Philosophy of Mind: Classical and Contemporary Readings, edited by D. Chalmers. Oxford: Oxford University Press. Huemer, Michael. 2001. Skepticism and the Veil of Perception. Lanham, Md.: Rowman and Littlefield. ———. 2007. “Compassionate Phenomenal Conservatism.” Philosophy and Phenomenological Research 74(1): 30–55.

292

Constitutivism

James,William. [1890] 1981. The Principles of Psychology,Vol. 1. Cambridge, Mass.: Harvard University Press. Kriegel, Uriah. 2009. “Self-Representationalism and Phenomenology.” Philosophical Studies 143: 357–381. Lehrer, Keith. 1990. Theory of Knowledge. London: Routledge. Locke, Don. 1967. Perception and Our Knowledge of the External World. London: Allen and Unwin. Lycan, William. 1997. “Consciousness as Internal Monitoring.” In The Nature of Consciousness, edited by N. Block, O. Flanagan, and G. Guzeldere. Cambridge, Mass.: MIT Press. Marr, David. 1982. Vision: A Computational Investigation into the Human Representation and Processing of Visual Information. New York: Freeman. Neta, Ram. 2011. “The Nature and Reach of Privileged Access.” In Self-Knowledge, edited by A. Hatzimoysis. New York: Oxford University Press. Nichols, Shaun, and Stephen Stich. 2003. Mindreading: An Integrated Account of Pretense, Self-Awareness, and Understanding of Other Minds. Oxford: Oxford University Press. Nozick, Robert. 1981. Philosophical Explanations. Cambridge, Mass.: Harvard University Press. Peacocke, Christopher. 1992. A Study of Concepts. Cambridge, Mass.: MIT Press. ———. 1998. “Conscious Attitudes, Attention and Self-Knowledge.” In Knowing Our Own Minds, edited by C. Wright, B. C. Smith, and C. Macdonald. Oxford: Oxford University Press. Pitt, David. 2004. “The Phenomenology of Cognition or What Is It Like to Think That P?” Philosophy and Phenomenological Research 69(1): 1–36. Quine, Willard van Orman. 1970. “Methodological Reflections on Current Linguistic Theory.” Synthese 21: 386–398. Rorty, Richard. 1979. Philosophy and the Mirror of Nature. Princeton: Princeton University Press. Rosenthal, David. 1997. “A Theory of Consciousness.” In The Nature of Consciousness, edited by N. Block, O. Flanagan, and G. Guzeldere. Cambridge, Mass.: MIT Press. Schwitzgebel, Eric. 2008. “The Unreliability of Naïve Introspection.” Philosophical Review 117: 245–273. ———. 2010. “Acting Contrary to Our Professed Beliefs, or the Gulf between Occurrent Judgment and Dispositional Belief.” Pacific Philosophical Quarterly 91: 531–553. Searle, John. 1990. “Consciousness, Explanatory Inversion and Cognitive Science.” Behavioural and Brain Sciences 13: 585–596. Shoemaker, Sydney. 1996. The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press. Siewert, Charles. 1998. The Significance of Consciousness. Princeton: Princeton University Press.

A Simple Theory of Introspection

293

Smithies, Declan. 2011.“What Is the Role of Consciousness in Demonstrative Thought?” Journal of Philosophy 108(1): 5–34. ———. Forthcoming, a. “Mentalism and Epistemic Transparency.” Australasian Journal of Philosophy. ———. Forthcoming, b. “Moore’s Paradox and the Accessibility of Justification.” Philosophy and Phenomenological Research. ———. Forthcoming, c. “Why Justification Matters” In Epistemic Evaluation: Point and Purpose in Epistemology, edited by J. Greco and D. Henderson. Oxford: Oxford University Press. ———. Forthcoming, d. “The Phenomenal Basis of Epistemic Justification.” In New Waves in Philosophy of Mind, edited by J. Kallestrup and M. Sprevak. London: Palgrave Macmillan. Sosa, Ernest. 2003. “Privileged Access.” In Consciousness: New Philosophical Perspectives, edited by Q. Smith and A. Jokic. Oxford: Oxford University Press. Steup, Matthias. 2009. “Are Mental States Luminous?” In Williamson on Knowledge, edited by P. Greenough and D. Pritchard. Oxford: Oxford University Press. Stich, Stephen. 1978. “Beliefs and Subdoxastic States.” Philosophy of Science 45: 499–518. Strawson, Galen. 1994. Mental Reality. Cambridge, Mass.: MIT Press. Wedgwood, Ralph. 2002. “Internalism Explained.” Philosophy and Phenomenological Research 65: 349–369. Williamson, Timothy. 2000. Knowledge and Its Limits. Oxford: Oxford University Press. Zagzebski, Linda. 1994. “The Inescapability of Gettier Problems.” Philosophical Quarterly 44: 65–73. Zimmerman, Aaron. 2006. “Basic Self-Knowledge: Answering Peacocke’s Criticisms of Constitutivism.” Philosophical Studies 128: 337–379. ———. 2007. “The Nature of Belief.” Journal of Consciousness Studies 14: 61–82.

This page intentionally left blank

11 Judgment as a Guide to Belief Nicholas Silins

1. Introduction What is the role of consciousness in our introspective lives? In this chapter, I will focus on the role of conscious judgment in giving us access to our standing beliefs. The view I will defend is a special case of a more general position in the epistemology of introspection. I will start by sketching the general view, and then will give the details of my particular position. One plank of the general position concerns specialness: our self-ascriptions of mental states can be justified in a way that our ascriptions of mental states to others are not. Another plank of the position concerns fallibility: we can nevertheless make justified mistakes about what mental states we are in. For example, if you hallucinate that something yellow is present, with no indication that anything is going wrong, you can be justified in believing that you are in the relational mental state of seeing something yellow, even though you are not in that state. The moderate position says that our introspective beliefs are sometimes justified in a both fallible and noninferential way.1 Accepting this position is, I think, an 1

In maintaining the moderate view, one need not say that we never have inferential access to our mental states, or that we never have infallible access to our mental states. For example, perhaps your pain can give you infallible justification to believe that you are in pain. Or perhaps you can have

296

Constitutivism

attractive way of avoiding two bad views. According to one, we cannot be mistaken about our own minds; according to the other, there is nothing distinctive about our introspective beliefs. The moderate position is attractive, but the debate in the area is often set up so as to leave it nearly out of sight. It is standard to ask about how we have self-knowledge or about how we have access to our mental states. Both these questions foreground the cases in which we are in a given mental state. But we need to consider what our positive epistemic position can be for a mistake about our own mind. One might expect Descartes to be the archenemy of the moderate position, but he arguably is not. He commits himself to something close to it in the following passage, as it is translated by John Cottingham (the passage was drawn to my attention by Moran 2001, 12 n. 9): [M]any are themselves ignorant of their beliefs. For since the action of thought by means of which we believe something is different from that by means of which we know that we believe it, the one is often found without the other. (A Discourse on Method, part 3) I should start by clearing away a deflationary reading of the passage. On this reading, Descartes is merely saying that people sometimes have a belief without knowing they have it, say because they have not considered the question of whether they have the belief. This reading interprets the final phrase in the Cottingham translation as saying that “the former is often found without the latter.” A look at the French resolves the problem. Descartes’s own phrase is “elles sont souvent l’une sans l’autre,” which should be rendered as “each one is often found without the other.” So the passage in fact does make a quite striking claim: the action of thought by means of which we know that we have a belief (when all goes well) is often found without the action of thought by means of which we believe. There are open questions about what Descartes means here by “action of thought.” I will set them aside, and use Descartes’s remark as a springboard to express the position I will defend in this chapter (the position is developed in a related but different way by Peacocke 1999). The view is an instance of the moderate position. I would first adapt what Descartes says as follows: the “action of thought” by which we know we believe something is sometimes our conscious judgment, where we can consciously judge that p without believing that p, and vice versa. The action of thought by which we sometimes know we believe that p thus does not

inferential justification to believe that you are in pain by inference from observation of your behavior (in addition to having justification to believe that you are in pain stemming from your pain itself).

Judgment as a Guide to Belief

297

guarantee the presence of a belief that p. Moreover, one can believe that p without judging that p —each state can be found without the other. Rather than focusing on knowledge, however, I prefer to focus on justification. That is because I am interested in what your positive epistemic position can be for a mistake about your own mind, and almost everyone agrees you cannot know a false proposition about anything.2 Finally, I would add in particular that conscious judgments give us noninferential justification for second-order beliefs. Here I go well beyond what Descartes says. My central thesis about conscious judgments is thus the following instance of the moderate position: Conscious Judgment: Conscious judgments give us noninferential yet fallible justification for second-order beliefs. I will defend the thesis by looking closely at the famous “transparency method” discussed by Edgley (1969), Evans (1982), Moran (2001), and others. Very roughly, the key idea is that, when you answer the question whether p, you put yourself in a position to answer the question whether you believe that p. As I will develop the idea, judgment is a guide to belief. In particular, one’s conscious judgments are a basic yet fallible guide to one’s nonconscious standing beliefs. There are several reasons why it is important to take a much closer look at the transparency method. Of course, we need to understand the method to understand the epistemology of introspection more generally. But the topic is important for metaphysical reasons, as well as epistemological reasons. By considering the role of conscious states as a guide to our nonconscious states, we should gain a valuable constraint on views of consciousness. Our understanding of what consciousness is must allow it to play the epistemic role it in fact plays. In the conclusion I will highlight two specific upshots of our discussion for the nature of consciousness. I will discuss the transparency method in the first half of this chapter. Here I will argue that, on the best understanding of the transparency method, judgments give us noninferential yet fallible justification for second-order beliefs. The second half will respond to important objections to the view. The main challenge comes from so-called “constitutivist” views in the epistemology of introspection, which instead emphasize the role of beliefs themselves in giving one justification for second-order beliefs. Here I will give a systematic survey of quite different ways of developing this approach. Only some constitutivist claims will turn out to be incompatible with my view. I will argue that each of those claims is false or unmotivated. Before I take on the main projects of the chapter, let me introduce some key terms and clarifications. 2

For dissent, see Hazlett (2010).

298

Constitutivism

First of all, when I speak of “beliefs” in what follows, I will only have what one might call “standing beliefs” or “dispositional beliefs” in mind. You most likely had a standing belief a moment ago that the author of this chapter is Silins, even though you were not judging that the author of this chapter is Silins. When I speak of “judgments” in what follows, I will only have conscious judgments in mind, those which modify what it is like for you at the time you make them. I will not be concerned with nonconscious judgments (if there are any at all).3 To make a judgment of the kind I am interested in, you might sincerely assert to someone that p. But you can consciously judge that p without performing the linguistic act of assertion. In many cases conscious judgment will be the “inner analogue” of assertion, although it may well be that one can judge that p without in any way vocalizing or imagining a sentence with the content that p. On one view, judgments and beliefs are very closely related: a conscious judgment is simply a standing belief that has become conscious. One reason to doubt this view is that judgments are often caused by beliefs. For there to be causal relations between beliefs and judgments, we need two nonidentical states that are causally related. In what follows we will also see a stronger form of distinctness between judgment and belief— important cases in which one judges that p yet does not believe that p. To anticipate, one can judge that p as a result of a slip that fails to reflect one’s standing beliefs. Second, let me clarify what I have in mind by “direct” or “noninferential” access to one’s mental states. Let us say that you have immediate justification to believe that p just in case you have justification to believe that p, and you do so in a way that does not rely on your justification to hold any other belief (Pryor 2005). Notice that immediate justification is characterized in terms of how one gets to have it, rather than in terms of how strong it is, or in terms of when one gets to have it. The key requirement is that one is not made to have immediate justification by one’s having justification for any further belief. Third, let me clarify what I have in mind by “introspective justification”: you have introspective justification to believe that you are in a mental state M just in case you have justification to believe that you are in M, and you do so in a way such that no one else can have justification to believe that you are in M in that way.4 3

One might count the following sort of case as one involving nonconscious judgment: you have an occurrent rather than dispositional belief, given the active role of the belief in guiding your action, yet the belief is not conscious, given that the belief does not by itself modify what it is like for you at the time. I set aside the case of occurrent yet nonconscious belief in what follows. 4 At a minimum, no one else actually has the ability to access your mental states in the relevant way. I leave open whether there is any stronger sense in which it is impossible for others to access your mental states in the relevant way.

Judgment as a Guide to Belief

299

Notice that introspective justification is characterized in terms of who it is available to, rather than in terms of how strong it is, or in terms of when exactly one gets to have it. I characterize introspective justification by its “peculiarity” (Byrne 2005), not by its superiority. I also do not characterize introspective justification with any positive account of how it is acquired. The characterization leaves open whether introspective justification is immediate or not, whether it is acquired by “inner sense” or not, and so on. The term “introspective” is for this reason somewhat misleading. It might well suggest something like inner sense, and a corresponding view on which only phenomenally conscious mental states are available to introspection—no beliefs, only judgments. However, no such reading of the key term is intended, and it should be entirely open to say that we have introspective justification for beliefs concerning our nonconscious states of mind, given that we plausibly have a peculiar form of access to our nonconscious beliefs and desires. Finally, let me clarify what I have in mind by “fallible” access. I will say that a state j gives you fallible justification to believe that p just in case j gives you justification to believe that p and it is possible for you to be in j while it is not the case that p.5

2. Transparency and Belief A useful place to start in characterizing the transparency method is its famous discussion by Gareth Evans. We can learn a lot by reflecting on this discussion. However, much of what we will learn concerns how Evans is misleading or mistaken. He writes that in making a self-ascription of a belief, one’s eyes are, so to speak, or occasionally literally, directed outward—upon the world. If someone asks me ‘Do you think there is going to be a third world war?’, I must attend, in answering him, to precisely the same outward phenomena as I would attend to if I were answering the question ‘Will there be a third world war?’ I get myself in a position to answer the question whether I believe that p by putting into operation whatever procedure I have for answering the question whether p. . . . If a judging subject applies 5 According to Sutton (2007), it is not possible to have a justified false belief. This extreme position is actually compatible with the claim that one sometimes has fallible justification. My definition of “fallible” justification is actually silent about whether, when you have a fallible justification to believe that p, you could have justification to believe that p if it is not the case that p. The crucial question for my definition is whether a justifying state is such that one can be in it when it is not the case that p, leaving open whether it still gives justification to believe that p when it is not the case that p. The question is however not about whether one could have the relevant belief when it is false (see Pereboom 2011 for a recent book-length discussion of what sorts of mistakes one can make about one’s mind, and of the bearing of the issue on standard arguments against physicalist views about the metaphysics of mind).

300

Constitutivism

this procedure, then necessarily he will gain knowledge of one of his own mental states: even the most determined sceptic cannot find here a gap in which to insert his knife. (Evans 1982, 225) We can hint at the key lesson from this passage in the following way: Slogan: You can answer the question whether you believe that p by answering the question whether p. In order to make further progress, we need to consider several further questions. What is the method? When does it work? What does it do when it works? How does it work? To gain a grip on what the method is, let us start by looking at Evans’s claim that “I must attend, in answering him, to precisely the same outward phenomena as I would attend to if I were answering the question ‘Will there be a third world war?’” This remark suggests that, in order to find out whether one believes that p, one has to launch a new enquiry into whether p. The suggestion has at least two problems.6 First of all, if you judge that p, without having launched a new investigation into the matter, you are still in a good position to answer the question whether you believe that p. You need not do what Evans seems to demand in order to answer the question about your mind. Second, if you want to find out whether you already believed that p, you arguably should not do what Evans seems to demand, since a new enquiry might easily result in a new belief that p. For example, if you wondered whether you had a prior belief that God exists, you should not answer that question by considering considerations for and against the existence of God. In what follows I will largely bracket questions about your access to your antecedent beliefs. My main focus will instead be on the situation once you do answer the question whether p, and how you stand with respect to your current beliefs once you do. Since I am bracketing questions about one’s prior beliefs, I am also interested in the introspective upshot of answering the question whether p, whether or not you started out wondering about your beliefs before you asked the question whether p. We will therefore be concerned with a much wider range of cases than Evans. Let me now turn to a different question—when you answer the question whether p, which epistemic position are you in with respect to whether you believe that p? Evans boldly asserts that his procedure is “necessarily” a source of knowledge. We have some reason to doubt the bold claim. One complication arises if someone believes each of two contradictory propositions (I assume it is possible for a person to do so). 6 For further discussion of related complications, see Peacocke (1999, 215–216), Shah and Velleman (2005, 506–508), and Byrne (2005, 84–85).

Judgment as a Guide to Belief

301

Here the person’s belief that not-p might prompt her to answer the question whether p with a “no,” and to answer the question whether she believes that p with a “no.” Since she still believes the contradictory proposition that p, she does not achieve knowledge that she does not believe that p (we will see further problem cases in what follows). A more cautious view is answering the question whether p necessarily gives one justification for a second-order belief. The proposal is more plausible than the knowledge proposal given that the current proposal is less demanding. Also, in the case of contradictory belief, the person arguably does gain justification to believe she does not believe that p. There are two more wrinkles about what position you are in when you answer the question whether p. Notice that you can have justification to believe that p whether or not you in fact believe that p on the basis of the justification, and whether or not you in fact believe that p at all. On the view I will develop, answering the question whether p gives you justification whether or not you have a belief on the basis of that justification. In current jargon, I take answering the question whether p to provide what is known as “propositional justification” rather than “doxastic justification.” Let us now turn to the question of when answering the question whether p gives one justification. According to Edgley 1969 or Moran 2001, the question whether one believes that p is “transparent” to the question whether p. On this line of thought, presumably shared by Evans himself, whatever the answer one gives to the question whether p, one will be in a good epistemic position to give the very same answer to the question whether one believes that p. This view exaggerates the success of the transparency method (whose name might indeed be a misnomer). Sometimes one’s answer to the question whether p is just “maybe,” for example, if I ask myself whether it will rain one month from now. In many of those cases, however, one is still in a good position to answer the question of whether one believes that p with a “no,” rather than with a “maybe.” Strictly speaking, the question whether one believes that p is not transparent to the question whether p. I will work with a more cautious claim about when answering the question whether p gives one justification—it gives one justification when one judges that p. I thus take Evans’s procedure to be a guide to the presence of beliefs rather than to the absence of beliefs.7 I would therefore describe the method as follows: 7 I do not take the method to be our only guide to our beliefs. Evans seems to think otherwise, given his remarks about what one “must” do to answer the question whether one believes that p. To see why there are arguably other sources of introspective justification, consider that you could have introspective justification to believe that p when you may never have judged that p. Perhaps your standing belief that p could give you introspective justification for a second-order belief, without doing so via an intermediary judgment (Zimmerman 2006).

302

Constitutivism

Transparency Method: If you judge that p, believe you believe that p! Given that my main interest is in your epistemic position when you judge that p, whether or not you have taken advantage of that position, I will focus on the following kind of formulation: Transparency Thesis, First Pass: If you judge that p, then you have justification to believe that you believe that p.8 Although I have not formulated the thesis explicitly in terms of prima facie justification, please do read it and its successors with that in mind. When one judges that p, one’s justification to believe that one believes that p might well sometimes be defeated, say by opposing testimony from one’s therapist. But I will usually omit explicit reference to prima facie justification in order to streamline the discussion. We should now consider how to defend the transparency thesis. Notice that Evans says little to defend the Transparency thesis or anything like it. We can improve on his account here. In particular, we can support the transparency thesis by using it to explain an instance of “Moore’s Paradox,” that much-used beast of philosophical burden.9 To get the phenomenon in view, take a horse—“p”—and a donkey—“I do not believe that p,” and conjoin them to get the following mule: (MP): p and I do not believe that p. In particular, consider judgments of contents of the form MP. Other things being equal, it is irrational make such judgments. As Goldstein (2000) would nicely put it, such judgments tend to be “Mooronic.”10 Transparency provides a good explanation of why judgments of MP tend to be irrational. To see how, first consider that, when one judges the conjunction that 8 Although I will not discuss the further claims below in what follows, they may well be correct (as Daniel Stoljar pointed out to me):

(T2) If you judge that it is not the case that p, then you have justification to believe you do not believe that p. (T3) If you judge that it might not be the case that p, then you have justification to believe you do not believe that p. 9 Thanks to Alan Hájek for bringing me to consider potential connections between the transparency method and “Moore’s paradox.” 10 For a useful overview of forms of “Moore’s paradox,” and descriptions and explanations of their absurdity, see the introduction of Green and Williams (2007). For further discussion of Moore’s paradox and issues about self-knowledge, see Shoemaker (1995) and (this volume). A major divergence is that Shoemaker argues for a much stronger claim: if you believe that p (and are suitably rational and conceptually sophisticated), then you will believe that you believe that p. I suspect this claim goes too far. Consider the possibility of a thoroughgoing eliminativist with respect to belief. I take it such a character can still be rational and have the concept of belief, without ever believing she believes that p.

Judgment as a Guide to Belief

303

p and I do not believe that p, one judges that p, and judges that one does not believe that p. Now, if Transparency is true, one will have justification to believe that one believes that p whenever one judges that p, and thus whenever one makes a judgment of the form MP. The overall upshot will be that, when one judges the MP conjunction, one judges the second conjunct, while having justification to believe the negation of the second conjunct. As we will see in a moment, there might be cases in which your prima facie justification from your judgment is defeated. Nevertheless, those cases are nonstandard. Your standard epistemic position is such that, if you were to judge in that position that p and I do not believe that p, your prima facie justification from your judgment would not be defeated. Therefore, if the Transparency thesis is true, it is standardly irrational to judge contents of the form MP. Transparency also helps to capture the elusive way in which “Moore paradoxical” judgments are distinctively defective. It is somehow defective to judge that one does not exist, but there is nothing worse about judging that one does not exist and snow is white. There is no tension between judging that one does not exist and judging that snow is white. To capture what is distinctively defective about Moore paradoxical judgments, we can emphasize that judging one conjunct is in tension with judging the other, since judging the one is sufficient for having prima facie justification to reject the other.11 Since the Transparency thesis explains why it is typically irrational to judge contents of the form MP, we have some reason to believe the Transparency thesis. Here we have taken an important step beyond Evans’s remarks.

11

My explanation has been framed using the notion of prima facie justification, which may well be defeated. In protest, one might demand an explanation of why it is invariably irrational to make a judgment of the form MP. In response, I do not accept that there is such a general phenomenon to explain. For an interesting case in which it might not be irrational to judge a content of the form MP, consider the following remarks by Velleman and Shah: Arriving at the judgment that p doesn’t necessarily settle the question whether one now believes it, since one may find oneself as yet unconvinced by one’s own judgment. One may reason one’s way to the conclusion that one’s plane is not going to crash, for example, and yet find oneself still believing that it will. (2005, 16–17) We need to tweak what they say to get the crucial point. They say that one judges that not-p, yet finds oneself believing that p, but what we need is a case in which one judges that not-p while finding oneself not believing that not-p.Velleman and Shah’s example arguably does supply such a case: one might judge that one’s plane will land safely, and yet still find oneself failing to believe that one’s plane will land safely. Here one certainly is not rational overall, since one appreciates the right reasons, and yet is unable to muster belief on their basis. However, one might still be justified in making the specific judgment that my plane will land safely but I do not believe that my plane will land safely. So it arguably sometimes is rational to make a judgment of the form MP. For further discussion of such cases, see Gertler (2010).

304

Constitutivism

Let me now turn to our final main question, about how answering the question whether p puts you in a good epistemic position to believe that you believe that p. I will clear up two potential misunderstandings before developing my positive view. First, it might be tempting to think that, if you form a belief that you believe that p through Evans’s procedure, you do so somehow in virtue of an inference corresponding to the following argument: p So, I believe that p. The tempting thought should be resisted (but see Byrne 2005, 2010 for an important development of it). First, it jars with the phenomenology—we never find ourselves reasoning along the lines “p, so I believe that p.” Second, consider the wide ranges of cases in which one can achieve knowledge through Evans’s procedure (to say that there are many such cases is not to say that the procedure is invariably a source of knowledge). To put the point roughly, brains in vats or victims of Gettier cases have poorer access to the external world, but they need not have poorer access to their beliefs. More specifically, one can know through the transparency method that one believes that p even if (a) one has a false belief that p or (b) one has a justified true belief that p while failing to know that p. Other things being equal, however, one does not achieve knowledge through inference from false lemmas or unknown lemmas. When one achieves knowledge through the transparency method, then, it looks like one does not do so through inference from the premise that p.12

12

Byrne (this volume) responds to the problem concerning knowledge by challenging the motivation of the “no false lemmas” requirement for knowledge (he does not address the corresponding problem about justification). In particular, he takes the no false lemmas requirement to be motivated by reflection on classic Gettier cases and maintains that a better diagnosis of what has gone wrong in the cases is that the Gettierized subject fails to meet a safety requirement for knowledge. The rough idea is that the subject is in too much danger of falsely believing that p to know that p. My challenge does not rely on a full-strength necessary condition for knowledge, but instead on the heuristic that, other things being equal, one does not gain knowledge through inference from a false lemma. In any case, I doubt the safety requirement provides a better diagnosis of what has gone wrong in classic Gettier cases. Consider the original case in which Smith has a justified false belief that Jones owns a Ford, and infers the justified true belief that either Jones owns a Ford or Brown is in Barcelona. If we build into the case that it is an extremely robust fact that Brown is in Barcelona, so that Smith is safe from error with respect to the disjunction, Smith still plausibly fails to know the disjunction.

Judgment as a Guide to Belief

305

We can make a similar point about the case of justification. When one has an unjustified belief that p, one can use the transparency method to achieve a justified belief that one believes that p. Other things being equal, however, one does not obtain a justified belief by reasoning from an unjustified belief.13 In sum, we will not understand the transparency method if we think of it as a case of somehow reasoning from the premise that p to the conclusion that one believes that p. Inference tends to obey the principle “Garbage In, Garbage Out,” whereas the transparency method does not.14 A second tempting thought is that, when the transparency method is a source of justification to believe that one believes that p, “inner sense” or “inner observation” is not. It is hard to make precise the notion of inner observation (although valuable clarification is available in Shoemaker 1996). Whatever inner observation may exactly be, however, one might think that the transparency method has nothing to do with it. This tempting thought is not clearly correct. A proponent of inner observation could say that, when one judges that p, one has inner observation of one’s judgment (or of the fact that one has made the judgment), and thereby has inner observation of one’s belief that p (or of the fact that one believes that p). Compare: when one observes movement in a nest, one might thereby observe an object in the nest, or observe that there is an object in the nest.15 Reflection on the transparency method might indeed remove some of the motivation for thinking that we obtain self-knowledge through inner observation. Nevertheless, it remains perfectly possible that we obtain self-knowledge through inner observation and the transparency method at the same time.16 I will set inner observation aside and take up a quite different positive view about the source of one’s justification when one answers the question whether p. On this view, when one gains justification for a second-order belief through the transparency method, one’s judgment is itself a source of justification for the second-order belief:

13

According to the Evans-inspired approach of Fernández (2003, 2005) or Williams (2004), what justifies one in believing that p also justifies one in believing that one believes that p. This approach is at best incomplete, since the transparency method can give one a justified secondorder belief when one lacks a justified first-order belief. For extended critical discussion of the approach of Fernández and Williams, see Zimmermann (2004) or Vahid (2005). 14 For further discussion of the transparency method and inference, see Gallois (1996), Brueckner (1998), and Shoemaker (this volume). 15 Relevant here is the “displaced perception” account discussed in Tye (2002). 16 Contrast Richard Moran, who builds in that “a statement of one’s belief about X is said to obey the Transparency Condition when the statement is made by consideration of the facts about X itself, and not by either an ‘inward glance’ or by observation of one’s own behavior” (2001, 101).

306

Constitutivism

Transparency, Second Pass: If you judge that p, then your judgment that p gives you justification to believe that you believe that p. Our previous formulation of transparency was silent about the source of one’s justification for a second-order belief. The current formulation is not. The key idea here is that judgment is itself a guide to belief. This claim is highly plausible, since it explains the correlation between judging that p and having justification to believe you believe that p. There is a significant advantage of focusing on the epistemic role of judgment in our formulation of the Transparency thesis. If one were instead to focus on the content that is judged, the transparency method could easily seem puzzling or paradoxical. Typically, when one judges that p, the content that p does not serve as evidence regarding one’s beliefs. For example, that it is exactly 9:02 a.m. is hardly evidence I believe that it is exactly 9:02 a.m. If one were to try to understand the transparency method by looking at the epistemic role of the content one judges, it would be puzzling how the transparency method is a source of introspective justification. If we instead focus on the role of judgment itself, however, we elegantly avoid such puzzlement about how the transparency method is a source of introspective justification. There is not yet any puzzle or paradox of transparency.17 We still need to address how judgment is a source of distinctively introspective justification. Here it is essential to move beyond the slogan with which we began,

17 Contrast Byrne (2005), on whose approach a “puzzle of transparency” does arise. On his account:

this situation will be commonplace: trying to follow BEL [the rule,“if p, then believe you believe that p!”], one investigates whether p, mistakenly concludes that p, and thereby comes to know that one believes that p. (In these cases, one will know that one believes that p on the basis of no evidence at all.) (2005, 98) It is somewhat puzzling how one could know that one believes that p on the basis of no evidence at all, but I do not think we should accept that such a situation is commonplace. Other things being equal, when one person has evidence in favor of the proposition that p, and another does not, the person with evidence is entitled to be more confident that p than the person who lacks evidence. In the case of introspection however, I take it that your confidence that you believe that p is insensitive to whether it is the case that p—it is not as if you can normally be more confident that you believe that p when you have a true belief that p as opposed to a false belief that p. So I do not think we should accept the following asymmetry required by Byrne’s account: the person who uses the transparency method when it is the case that p bases their second-order belief on evidence, the person uses the transparency method when it is not the case that p does not base their second-order belief on evidence. I should say that writers on the topic sometimes take it we do not have any “evidence” for our introspectively justified beliefs—I take them to be using some highly demanding notion of “evidence.”

Judgment as a Guide to Belief

307

according to which one can answer the question whether one believes that p by answering the question whether p. Given the right background information, I sometimes can answer the question whether you believe that p by answering the question whether p. Now I cannot answer the question whether you believe you have food in your teeth by answering the question whether you have food in your teeth. However, given my background information, I can answer the question whether you believe you have gold teeth by answering the question of whether you have gold teeth. Given my background knowledge that you are unlikely to have gold teeth unawares, if I answer “yes” to the question whether you have gold teeth, I am in a good epistemic position to answer “yes” to the question whether you believe you have gold teeth. Be that as it may, I do not have introspective access to any of your beliefs. As far as I know, there is little discussion of how the transparency method is specifically a source of introspective or “peculiar” justification. A valuable exception is Byrne (2005). To explain the peculiarity of one’s access to one’s beliefs through the transparency method, Byrne (2005, 96) emphasizes a difference between the following two rules: BEL: if p, then believe you believe that p! BEL3: if p, then believe that Fred believes that p! To follow a rule in Byrne’s sense, one must comply with the consequent because one recognizes (and so knows, and so believes) that the antecedent is true. Therefore, if one follows BEL, one ends up with a true second-order belief—the rule is “self-verifying” in the sense that following it invariably produces a true belief. BEL3 is not self-verifying: when one follows BEL3, one may well end up with a false belief about what Fred believes. I suspect the contrast does not explain peculiarity. Suppose (please bear with me!) that there is an omniscient God, and consider the rule: BEL4: if p, then believe God believes that p! If there is an omniscient God, following BEL4 will be sufficient for forming a true belief that God believes that p. Whenever one follows BEL4 in Byrne’s sense of “follow,” one will recognize that p, and so know that p, so it will be the case that p. Therefore, whenever one follows BEL4 in Byrne’s sense of “follow,” the omniscient God will know that p, and so the omniscient God will believe that p. However, despite all this, one still does not have properly introspective access to the beliefs of the omniscient God. To ensure that judgment is a source of distinctively introspective justification, I will say that one does not rely on background information when one gains justification

308

Constitutivism

from one’s judgment to believe one believes that p. This secures a contrast between the first-person case and the third-person case, while being phenomenologically plausible. I will thus refine our characterization as follows: Transparency, Third Pass: If you judge that p, then your judgment that p gives you immediate justification to believe that you believe that p.18 I will make one more point about the way in which judgment is a guide to belief. Judgment is a fallible guide to belief. To adapt what Descartes said in the Discourse on Method, the action of thought by which we gain justification for a second-order belief can occur in the absence of the relevant first-order belief. For an important type of case, consider the following: you judge that your flight leaves at noon, and then realize that you do not and did not believe that your flight leaves at noon. In such cases, your judgment that p is a kind of performance error which fails to reflect an underlying belief—“what was I thinking?” you might go on to say. You “blurted out” that p, either in speech or in merely in thought, consciously endorsing the proposition that p, yet failing to have a standing belief that p. Judging that p is insufficient for believing that p, I take it, because believing that p requires having various dispositions, where judging that p is insufficient for having those dispositions. In protest, one might say that judgment that p is a species of conscious belief that p, so that it is impossible to judge that p without believing that p. This objection misses the point. Our focus is on the way in which judgment is a guide to what we might think of as standing beliefs, or as dispositional beliefs.19 For the purposes of

18 Here I do not assume that all introspective justification is immediate (we will see cases of inferential introspective justification in what follows). I should say that there are alternative explanations of the contrast. For example, one might say that you rely in your own case on the background belief that, if you judge that p, then you believe that p, whereas in the case of others you rely on the background belief that if p, then the other person believes that p. I think this explanation is worse than my own, on the grounds that it over-intellectualizes the transparency method. In particular, the transparency method is available in a wider range of cases than those handled by the alternative explanation. The first type of cases involves those who have the concept of belief but not yet the concept of judgment: such people may use the transparency method but do not have the background belief that if you judge that p then you believe that p. The second type of case involves those who have the concept of belief and the concept of judgment, but who are (for whatever bizarre reason) thorough eliminativists about judgment although not eliminativists about belief. Such people may also use the transparency method to acquire a justified belief they believe that p, but presumably will not be relying on a background belief that if they judge that p then they believe that p. 19 A separate question concerns how judgment might itself be a guide to judgment. It might turn out that judgments are an infallible guide to judgments, but my focus is on how judgments are a fallible guide to beliefs.

Judgment as a Guide to Belief

309

the chapter I have reserved the term “belief ” for such states. As far as the focus of this chapter is concerned, it does not matter whether there is a wider use of the term “belief ” which encompasses judgments themselves.20 Also, one might protest that, if a judgment gives one fallible justification for a second-order belief, then it cannot give one immediate justification for a second-order belief. On this line of thought, if a state gives one fallible justification to believe that p, then there is a “gap” between the obtaining of the state and the truth of the proposition that p, so that some intermediate belief will be needed so as to “bridge the gap” between them. However, if the intermediate belief is itself only fallibly justified, the line of thought can be repeated concerning its own justification. And since there are few infallibly justified beliefs, the line of thought threatens to show that there are few fallibly justified beliefs. Since many of our beliefs are fallibly justified if justified at all, the reasoning threatens to have skeptical consequences. We should not endorse it. We may now formulate the Transparency thesis as follows: Transparency: If you judge that p, then your judgment that p gives you immediate fallible justification to believe that you believe that p. This is the lesson to take away from reflection on Evans’s classic passage.21 20 Still, there is controversy about whether one can judge that p without having a standing belief that p. Although the affirmative view is defended for example by Peacocke (1999), it is denied for example by Zimmerman (2006). I should emphasize that, even if judging that p did suffice for believing that p, it would be enough for my purposes if there is some state phenomenologically just like judging that p, which can occur in the absence of belief. That is because, in the cases in which one is in a judgment-like state without believing that p, I will hold that the judgment-like state still gives one immediate fallible justification to believe that one believes that p. Although the fallback position is available, it is simpler to work with talk about judgment. Zimmerman (2006) objects to the fallback position as follows:

if experience with the phenomenal character of genuine judgment is not sufficient for belief, we can have experiences with this phenomenal character that are not real judgments (for they don’t initiate, sustain or accompany beliefs). If our second-order introspective beliefs are grounded in such judgment-like experiences, knowledge of our beliefs is not direct, but instead mediated by inconclusive inferential grounds or states of inner perception. (367) In response, I see no reason to accept the dilemma he proposes. Judgment-like experiences arguably can provide fallible immediate justification, we need an argument that they cannot. (I will say much more about fallible immediate justification in what follows). 21 For a different take on the transparency method, in terms of considerations about “rulefollowing,” see Byrne (2005) and Setiya (forthcoming). For criticisms of Byrne (2005), see Shoemaker (this volume). A question I leave open is whether the transparency method can somehow be generalized beyond the case of belief. For discussion of the case of visual experience, see Evans (1982), Peacocke (2008), and Byrne (2005). For discussion of further cases, see Gordon (1995, 2007), Byrne (2005, this volume), and Way (2007).

310

Constitutivism

3. The Case Against Transparency I will now consider key objections to the Transparency thesis. I will raise two challenges briefly before moving on to my main discussion.

3.1. Preliminary Challenges The first objection uses a constraint on immediate justification: Face Value Constraint: Necessarily, if a state M gives one immediate justification to believe that p, then M has the content that p. I grant that the Face Value Constraint is attractive. In particular, it meshes well with the plausible claim that, when one forms an immediately justified belief on the basis of a state, one does so by taking some content of the state at face value. Given that a judgment that p does not have the content that one believes that p, the Face Value Constraint predicts that judgments that p are not sources of immediate justification for self-ascriptions of beliefs that p. As attractive as it is, the Face Value Constraint is false. One way to see this is by considering the case of consciousness. Your state of being conscious can give you immediate justification to believe that you are conscious, whether or not you are in any state with the content that you are conscious.22 A somewhat more controversial case to consider is that of belief. Arguably one’s belief that p can itself give one immediate justification for a second-order belief (Zimmermann 2006). If that is right, then the Face Value Constraint is false, since one’s belief that snow is white does not have the content that one believes that snow is white, the first-order belief instead simply has the content that snow is white. The challenge from the Face Value Constraint fails.23 The next objection is inspired by a difficulty that arises in the epistemology of perception. The worry in the perception case is that, if our experiences do provide immediate justification for external world beliefs, then it will be too easy for us to reject skeptical hypotheses about our experiences. In particular, we might be able to justifiedly reject skeptical hypotheses simply by performing inferences that correspond to the following argument:

22 According to philosophers such as Searle (1983), visual experiences have contents that involve references to the experiences themselves. This is not yet to say that any experience of mine will have the content that I am conscious. 23 For a nice discussion of related issues, see Pryor (2005). One might wonder whether there are counterexamples to the Face Value Constraint that are not cases of introspective justification. I believe such cases arise in the epistemology of perception and discuss them further in my (2011).

Judgment as a Guide to Belief

311

I have hands. If I have hands, then I am not a handless brain in a vat. So, I am not a handless brain in vat. According to the challenge, since we do not gain justification to reject skeptical hypotheses by performing such inferences, we do not gain immediate justification from our experiences for external world beliefs either (Cohen 2002, 2005; Wright 2002). There is a parallel objection to my account of how judgment is a guide to belief. The accusation would be that, if judgment is a fallible source of immediate justification, then it will be too easy for one to gain justification to reject skeptical hypotheses that say I judged that p without believing that p. Since it is in fact not so easy for one to gain justification to reject those skeptical hypotheses, judgment is not a source of immediate justification after all. In both cases, the idea is that we lack basic justification, since we lack easy justification to reject skeptical hypotheses. In brief, I think the best reply separates issues about anti-skeptical justification from issues about introspective justification. We can enjoy basic justification for our introspective beliefs whether or not we enjoy easy justification to reject skeptical hypotheses. But it takes a paper or two to properly develop the line of objection and explain why it fails. I do the needed work in my (2008) and (forthcoming).

3.2. The Constitutivist Critique I will now turn to the main challenge I will address in this chapter. According to the moderate position I have developed, what justifies the subject’s second-order belief in some cases is compatible with the falsehood of the second-order belief. However, one might think I have underestimated infallible sources of justification. The line of objection is based on so-called “constitutivist” views about introspection, so-called because they propose a connection between the nature of belief and second-order beliefs. The approach has been taken up in very different ways by philosophers such as Sydney Shoemaker (1996, this volume), Richard Moran (2001), Jane Heal (2002), and Akeel Bilgrami (2006).24 24 A related approach appeals to the nature of introspective justification itself, rather than to the nature of mental states (Smithies, this volume; see also Neta 2010). On this line of thought, it is of the essence of introspective justification that, for certain privileged mental states M, one has introspective justification to believe that one is in M if and only if one is in M. I believe this approach fails to capture the essence of introspective justification, insofar as the target notion of “introspective justification” is grounded in ordinary cases of a “peculiar” route to justification, a way of

312

Constitutivism

Since the line of objection raises more general issues about what it is to believe that p, it is of much wider interest than just as a challenge to my view. The approach’s most prominent proponent, Sydney Shoemaker, develops the approach by focusing on the relation between first-order beliefs and second-order beliefs. In order to articulate a view specifically about the epistemology of introspection, however, we need to look at the relation between first-order beliefs and epistemic states such as justified second-order beliefs or knowledge. Given that there are many epistemic states, there are also many ways to develop the constitutivist position. My aim is to build the view from the ground up as a view in epistemology. I will therefore depart from what promising possibilities there are in logical space, rather than from quotations from key figures (although their remarks will still play a role in guiding our development of the view). I will also bring the view to engage with discussions in epistemology in general, rather than only in the philosophy of mind in particular. We will survey a number of constitutivist claims in what follows. I will argue that the stronger claims invoked are false or unmotivated, and the weaker claims invoked are compatible with my own position. The constitutivist must address several choice points. The first I will discuss is whether the view will concern an entailment from beliefs to epistemic states, or instead from epistemic states to beliefs. Let us start by considering entailments from beliefs to epistemic states, and in particular the epistemic state of knowledge. Here we are concerned with the manner in which beliefs might be “self-intimating.”

having justification to believe that one is in a mental state such that no one else can have justification to believe one is in the mental state in that way. In what follows in the main text, I argue that we can have such “peculiar” justification for false beliefs about our own mental states, as well as for inferentially justified beliefs about our own mental states. More demanding notions of “introspective justification” can be defined, but I take it to be highly controversial whether they have a useful theoretical role to play. For instance, on the notion of “introspective justification” developed by Smithies (this volume), we do not have introspective justification for beliefs concerning our bodily states (through proprioception), or for beliefs concerning our beliefs with wide contents such as our belief that water is H2O, or for any belief that one believes that p from a judgment that p. Smithies (this volume) does assign a role for his notion of “introspective justification” with respect to the formulation and explanation of a version of access internalism in epistemology. The matter deserves much more discussion, but it is not clear to me whether access internalism in epistemology is best understood and motivated by using so strong a form of “introspective” access as the one articulated by Smithies. Given the narrow scope of his notion of “introspective justification,” his version of access internalism allows no mental state with wide content to play a justificatory role. So, for example, your (justified) belief that water is H2O will not be a justifier of your belief that water is chemically complex. On the grounds that mental states with wide contents do justify us in holding some beliefs, I do not yet see a useful role for the demanding Smithies notion of “introspective justification.”

Judgment as a Guide to Belief

313

3.2.1. Self-Intimation A very demanding starting point would be the claim that Knowledge: Necessarily, if S believes that p, then S has introspective knowledge that she believes that p. This proposal is far too strong. Assuming that knowledge entails belief, the proposal straightaway requires an infinite regress of higher-order beliefs. Since the regress does not look benign, we should use a less demanding proposal.25 A weaker view in terms of knowledge is that Potential Knowledge: Necessarily, if S believes that p, then S is in a position to have introspective knowledge that she believes that p. Roughly speaking, one is in a position to know that p when one’s epistemic position is good enough for one to know that p, although one may not yet have met the psychological requirements for knowing that p, such as that of believing that p. Given that the current proposal abstracts from psychological requirements, it avoids the regress generated by the first.26 The proposal is still extremely demanding. I think the constitutivist should avoid commitment to it. A specific reason to do so is supplied by Williamson’s (2000) powerful antiluminosity argument. It is worth reviewing the argument to appreciate the challenge. On closer scrutiny,Williamson’s argument actually indicates that my fallibilist view about introspective justification is correct. So in the course of considering how to formulate a constitutivist view about self-intimation, we will actually see further evidence in favor of my fallibilist approach to introspection. The setup of the argument is as follows. Consider a series of times during which Belle’s confidence that p very gradually decreases, so that she believes that p at the beginning but does not believe that p at the end. Throughout the series Belle carefully considers whether she believes that p, so that whenever she is in a position to know she believes that p, she does know she believes that p. Finally, her confidence that she believes that p likewise gradually decreases throughout the series. 25

Although, see Shoemaker (this volume) for an argument that a similar regress is in fact benign. 26 Another way to proceed is by packing psychological requirements into the antecedent of the formulation of the view. These sorts of formulations would proceed as follows: If one believes that p and considers the question whether p . . . As far as I can tell, there will be no substantive difference between this family of formulations and the one I consider in the main text. One might wonder which family of formulations is more fundamental, assuming that one is, but I will not take on this question (although, see Smithies, this volume, for more).

314

Constitutivism

The crucial premise of the argument is that, if Belle knows at a given moment in the series that she believes that p, then she believes that p at the next moment in the series. To see why the premise is plausible, suppose it is false. If it is false, then there is some moment in the series at which Belle knows that she believes that p, and an adjacent moment at which she does not believe that p. This would require that she knows a proposition at the earlier moment despite having misplaced confidence at the next, where that misplaced confidence is at a very close level on a very similar basis. But knowledge plausibly requires the avoidance of such forms of error—there is plausibly a “safety” requirement on knowledge. So the crucial premise is plausibly correct.27 Let us now consider how to argue from the crucial premise to the falsehood of Potential Knowledge. At the first moment in the series, Belle knows that she believes that p. By the crucial premise, it follows that, at the second moment in the series, she does believe that p. By Potential Knowledge, at the second moment in the series, she is in a position to know (introspectively) that she believes that p. By the setup concerning her attentiveness, at the second moment in the series she does actually know that she believes that p. This reasoning can be repeated until we reach the conclusion that Belle believes that p at the last time in the series. Something has gone wrong! One might object to the crucial premise, or even to some aspect of the setup of Williamson’s argument, but I think the constitutivist is best advised to avoid engagement with it altogether.28 That is because there is a much more general reason to doubt the knowledge proposals: our epistemic humility should extend to propositions about our own minds and not just to about the world. Just as we should not infer that a worldly proposition is false from the fact that we are in no position to know it, we should not infer that a belief-ascribing proposition is false from the fact that we are in no position to know it. This humble line of thought does not beg the question against constitutivism. Constitutivists might want to say that our access to our own minds is superior to our access to the world, but we can comfortably accept this thought without going to the extremes of the knowledge proposal. 27 For some discussion of how to clarify the safety idea, see Sosa (1999) or Manley (2007). For some criticism, see Neta and Rohrbaugh (2004). Notice that given the setup of the anti-luminosity argument, it is unable to challenge one family of extremely strong claims about self-knowledge. Consider “revelation” theses about mental states, according to which one is in a position to know the essence of a mental state by being in it. Since one’s being in a given mental state does not belong to its essence, and more generally since the essence of a mental state does not change over time, Williamson’s argument will not threaten any revelation thesis. For further discussion of revelation theses about mental states, see Lewis (1995) and Stoljar (2009). 28 Useful critical discussions of the argument include those by Conee (2005) and Berker (2008).

Judgment as a Guide to Belief

315

The constitutivist view should be formulated in terms of some less demanding epistemic position than knowledge. One way to try to do this is as follows: Reason: If you believe that p, then one of the reasons you have to believe you believe that p is that you believe that p.29 It is tempting to think that this view is a cautious alternative to the earlier claims. This tempting thought is not clearly correct (it is also not clearly incorrect). To see how the complication arises, we should ask what it takes for one to have the reason that p to believe that q. According to the knowledge-theoretic tradition defended by Unger (1975), Williamson (2000), and Hyman (1999), Reason-Knowledge:You have the reason that p to believe that q only if you know that p. If this tradition is correct, the reason thesis entails the extremely strong knowledge thesis and is hardly any alternative to it. We should avoid taking a stand here on what reasons are and what it takes to have them. To avoid the complication, the constitutivist should not present her view specifically in terms of reasons, but instead more generally in terms of sources of justification. Consider that an experience can give a child justification for a belief, even if the child lacks the concepts required to know that she has the experience. Even if a knowledge requirement holds for reasons, there is still room for a more relaxed view about other sources of justification than reasons.30 The constitutivist is therefore better advised to work with the following proposal: Justification: Necessarily, if you believe that p, then your belief that p gives you introspective justification to believe you believe that p.31 This proposal more clearly avoids Williamson’s anti-luminosity argument. Here is a first approximation of the reason why. Although it is plausible that knowledge requires the avoidance of error in nearby cases, it is not plausible that justification requires the avoidance of error in nearby cases. If there were such a requirement on 29

Shoemaker and Zimmermann might be inclined to go along those lines. For instance, Shoemaker writes that “belief in a proposition provides a reason to believe a proposition—the normative proposition that one ought to be guided by that proposition in one’s thought and action—which is arguably coextensive with the proposition that one believes that proposition” (this volume). 30 One might of course not be so relaxed. For discussion of how to push the knowledge-theoretic approach further, see Williamson (2000). 31 I should say that I have in mind justification for outright belief as opposed to for a mere increase in confidence. There are less demanding formulations of constitutivism, but I will not pursue them further here. For further discussion of various fallback positions one might adopt in response to the anti-luminosity argument, see Greenough (this volume) and Smithies (forthcoming).

316

Constitutivism

justification, it would not be possible to have justification for a false belief, given that a false belief trivially involves error in a nearby case. I assume however that it is possible to have justification for a false belief.32 For this response to the anti-luminosity argument to be effective, it actually needs to be put in terms of introspective justification in particular rather than in terms of justification in general. Otherwise the threat would remain that there is a safety requirement for introspective justification, even though there is not a safety requirement for justification. However, if one adapts the reasoning in the previous paragraph all the way through, one will directly commit oneself to my view that one can have introspective justification for a false second-order belief. But then the constitutivist will be committed to one of my key claims. The more general strategy of the response to Williamson is to distinguish between what it takes to have introspective knowledge and what it takes to have introspective justification. So long as the constitutivist pursues this general strategy, it is not clear how she will be entitled to reject my view that one can have introspective justification for a false second-order belief. Given that introspective justification is unlike knowledge in that it lacks a safety requirement, it is not clear why introspective justification should be similar to knowledge in being factive. If introspective justification is compatible with the nearby possibility of error, it is unclear why, in the setup of Williamson’s argument, one should not retain introspective justification beyond the last moment at which one believes that p.33 In sum, to avoid the anti-luminosity argument, one should contrast what it takes to have introspective justification from what it takes to have knowledge. To make this move is to play into my hands. The less introspective justification looks like knowledge, the more it looks like it should be possible to have introspective justification for a false belief. Let me now return to the more general question of how to understand the constitutivist position. The formulations just given are all in terms of entailment. A further choice point for the constitutivist is whether to put their proposal in stronger terms, in particular 32

This point can also be found in Conee (2005) and Berker (2008). One might respond by making the further claim that a belief that p is the only potential source of introspective justification to believe that one believes that p. Such a position, of course, needs further defense, but it would explain why one might have introspective justification to believe that one believes that p only if one believes that p. Making this move would commit the constitutivist to the following position: 33

One has introspective justification to believe that one believes that p if and only if one believes that p. I give critical discussion of the position below.

Judgment as a Guide to Belief

317

in terms of the essence of belief rather than merely in terms of an entailment from belief. The two ideas are different. There is an entailment from my belief that snow is white to the fact that 2 + 2 = 4. However, the essence or nature of my belief is more fine-grained, and has nothing to do with the fact that 2 + 2 = 4. Mathematical facts are neither here nor there when it comes to acquiring a full understanding of what it is to believe that snow is white.34 The constitutivist is well advised to explain the necessity involved in Justification, rather than to leave it unexplained. A good way to do so, although not the only way to do so, is to endorse the following stronger claim: Essential Justification: If you believe that p, then it is of the essence of your belief that p to give you introspective justification to believe you believe that p. By focusing on the nature of beliefs,35 rather than merely on what having a belief entails, I think we have moved closer to the heart of the constitutivist approach.36 I set aside the further evaluation of Justification or Essential Justification. I do so because both of these claims are actually compatible with our account of the transparency method. Their main upshot is to give beliefs a role in the epistemology of introspection. They do not say anything, or clearly imply anything, about the role of conscious judgments in the epistemology of introspection. To include beliefs is

34

For valuable discussion of essence vs. necessity, see Fine (1994). Sometimes the constitutivist view is put in terms of the essence of rationality rather than in terms of the essence of belief. Consider the following comment by Shoemaker 2009: 35

this seems a step towards the view that beliefs are constitutively self-intimating—that it is part of being a rational subject that belief that p, together with the possession of the concept of belief and the concept of oneself, brings with it the belief that one believes that p. (2009: 36) If beliefs are to be constitutively self-intimating, I think we should focus on what it is to believe that p rather than on what it is to be rational. In particular I think we should instead say that it is part of believing that p that, if one believes that p, and is rational and has the concepts of belief and oneself, then one believes that one believes that p. 36

A separate question concerns our access to the absence of belief. Suppose that absences of belief are self-intimating, in the sense that, if you do not believe that p, then you have introspective justification to believe you do not believe that p. To give a parallel explanation, the constitutivist would need to invoke a claim about the essence of nonbelief, to the effect that the essence of nonbelief involves access to the absence of belief. But such claims are mistaken. My chair does not have any belief and also does not have any justification to think that it lacks beliefs. It does not have any mental states or justifications at all.Whether or not belief has a (partly) epistemic essence, nonbelief does not have an (even partly) epistemic essence. In any case, there is no need to give a special account of the nature of the absence of belief. In general, the essence of the absence of x is just the absence of the essence of x. So a constitutivist story about our access to the absence of beliefs will need to be developed in some other way. For further discussion of cases of absences of belief, see Sosa (2003).

318

Constitutivism

by no means to exclude judgments. Moreover, the proposals do not say or clearly imply anything about whether there is fallible introspective justification. The formulations are restricted to cases in which one does believe that p. They are silent about cases in which one does not believe that p. I hope so far to have improved our understanding of the options open to the constitutivist. As the view has been developed so far, it has turned out to be no threat to my own. To isolate the threat to our treatment of transparency, we need to look at an entirely different family of constitutivist views.

3.2.2. Infallibility To isolate a challenge to my view, we need to look at claims that at least concern an entailment from epistemic states to first-order beliefs, rather than entailments from first-order beliefs to epistemic states. As before, the epistemic state in question might be that of knowledge or that of justification (setting aside even stronger or even weaker states). I will focus on the case of justification and will start with the following proposal:37 Justification 2: If one has introspective justification to believe one believes that p, then one believes that p.38 This thesis concerns introspective justification, but it does not say anything further about sources of introspective justification. Given this omission, Justification 2 is compatible with the following more specific proposal: Observational Justification: Necessarily, if one has introspective justification to believe one believes that p, then one has inner observation of one’s belief that p.

37

The knowledge case is less controversial. Since almost everyone agrees that one can know a proposition only if it is true, almost everyone will agree at least that Necessarily, if one knows that one believes that p, then one believes that p. There are further questions to address about the ground of one’s knowledge, which are so far left open, but I will set them aside. So long as one knows only true propositions, thinking about knowledge will not tell us whether one can be in a good epistemic position through introspection for a false belief. 38 This claim is endorsed in Zimmerman (2006). Consider his remark that “when our secondorder introspective beliefs are formed and maintained in a first-person way they are grounded in the very first-order mental states that make them true” (370). However, he sometimes presents his view in a more qualified way: “if we have any false, justified beliefs about what we believe, the grounds for these beliefs will be different in kind from the grounds with which we hold our typical second-order introspective beliefs” (371). One could accept this quote while allowing for an introspectively justified yet false second-order belief.

Judgment as a Guide to Belief

319

The current proposal takes the stand that inner observation is a source of introspective justification. But this idea is odious to the typical constitutivist. We need to look further to capture the spirit of their view. We need at least the following more specific claim: Constitutive Justification: If you have introspective justification to believe you believe that p, then what gives you justification for the second-order belief is that you believe that p (rather than inner observation of your belief that p). The claim has very striking consequences. If it is true, then one cannot have a secondorder belief that is both false and introspectively justified. As far as second-order beliefs are concerned, then, CJ embodies an infallibilist conception of introspective justification.Also, the thesis requires that only beliefs provide introspective justification for second-order beliefs. So CJ rules out judgment is a source of justification at all. Constitutive Justification is striking but mistaken. We can have fallible introspective justification for our second-order beliefs, just as we can have fallible justification for other beliefs. The most convincing way to see why is by considering cases of inferential introspective justification. I will set out such a case in some detail, since it provides a principled way to argue against CJ. Consider a quite idealized subject: when something she believes has a strictly logical consequence, she reliably tends to believe the consequence. She also has justification to believe that she is thorough in this way. However, she is not perfect, since she does not always follow through with the logical consequences of what she believes, and she also sometimes makes justified mistakes about what is a strictly logical consequence of what. Now suppose she reasons in a quite indirect way about what she believes, in a way she would vocalize as follows: (1) I believe that p. (2) As a matter of logic, if p, then q. (3) If [I believe that p and, as a matter of logic, if p, then q], then I believe that q. So, (4) I believe that q. Focus on cases where she has a justified false belief in (2)—I take it we can all make such mistakes when logic gets hard. In such cases she can still have an introspectively justified belief in (1), and a justified belief in (3). As long as these pieces are in place, she presumably will end up with a justified belief as well in (4), given that it is an obvious consequence of the contents of other justified beliefs she has. Further, her belief in (4) should be introspectively justified in particular. All of these pieces can be in place whether or not (4) is true. When she reasonably misidentifies a logical entailment, she can fail to believe the proposition she

320

Constitutivism

takes to be a strict logical consequence of another proposition she believes. Remember that in doing the reasoning she would vocalize with (1) through (4), she is forming a belief she believes that q, and need not be forming a first-order belief that q. In particular, as far as her explicit reasoning is concerned, she is not reasoning from her belief that p to form a belief that q. Now, her lack of a belief that q need not interfere with her justification to believe any of (1) through (3). So her belief in the conclusion will be justified—introspectively—whether or not it is true. Given that this sort of case is possible, the Constitutive Justification thesis is false. In response, someone might say that the subject is not justified at all in believing (4). Whether or not there are counterexamples to the principle that justified belief is closed under obvious consequence, I take it that the current example is not such a case.39 In a separate response, someone might say that the subject is not introspectively justified in believing the conclusion, on the grounds that she is inferentially justified in believing the conclusion.40 But something can both be an inferential and an introspective source of justification. For example, it might be through reasoning about counterfactual situations that you realize you hope that p, rather than expect that p (Williamson 2000). In such a case you have inferential justification to believe you hope that p, but the case is still a paradigm of introspective justification, since no one else can gain justification to believe you hope that p in the way you did. 39 Shoemaker (this volume) argues that it is not even possible to falsely believe one believes that p. In particular, he writes that

Will it [the second-order belief] bestow the disposition to assent to the content of that putative belief? If it does, we will then have a case for saying that the person does believe that content, or at least that it is not determinately true that he does not. If it does not bestow that disposition, then the person will be liable to fall into a version of Moore’s paradox—saying, or thinking, “I believe that p, but not-p,” or “I believe that p, but I have no idea whether p is true.” (44) In response, if a subject with a false second-order belief must fall into a version of Moore’s paradox, I would simply accept that this can happen. Being subject to Moore’s paradox is no barrier to having a false second-order belief, or even to having an introspectively justified false second-order belief. We can agree that it is irrational to believe the relevant conjunction, while still maintaining that it is possible to believe the relevant conjunction. In particular, it could even be that the best explanation of why it is irrational to believe the conjunction proceeds in terms of one’s having introspective justification to believe the first conjunct. 40 See, e.g., Fernández (2005: 541–542): Whatever adopting that [first-person] perspective on our own beliefs ultimately amounts to, it is a way of forming beliefs about them that provides one’s meta-beliefs with a special kind of justification, in that: (i) It does not depend on reasoning. (ii) It does not depend on behavioral evidence.

Judgment as a Guide to Belief

321

One might have thought that all introspective justification is immediate justification. This thought is wrong. The case I just presented only concerns inferentially justified beliefs. The proponent of CJ therefore might fall back to the following weaker claim: Constitutive Justification 2: Necessarily, if you have immediate introspective justification to believe that you believe that p, then what gives you justification for the second-order belief is that you believe that p. Since CJ2 is merely concerned with immediate introspective justification, it is silent about the type of case just discussed. The claim is still striking. If it is true, then one cannot have a false second-order belief that enjoys immediate introspective justification. So CJ2 embodies an infallibilist conception of immediate justification, at least as far as second-order beliefs are concerned. Next, if CJ2 is true, only beliefs provide immediate introspective justification for second-order beliefs. The thesis thereby rules out that judgments also play this role. So the thesis is still a threat to my overall position.41 My case against CJ2 is supplied by my account of the transparency method, according to which your judgment that p can give you immediate introspective justification to believe you believe that p, even though judgments are distinct from beliefs. The point I most want to emphasize here is that CJ2 needs to be defended, where it is far from clear how to defend the claim. One way to defend CJ2 would be to say that, in general, immediate justification is infallible. Call this the infallibilist conception of immediate justification. Even if this demanding view is true, it does not quite get us to CJ2. The infallibilist conception tells us that every immediately justified second-order belief is true, but it does not tell us what the source is of their justification. Claims about the source of a justification are quite different from claims about the infallibility of a justification. For example, an immediately justified belief that 0 = 0 is trivially an infallibly justified belief, but it is dubious that the fact that 0 = 0 is somehow itself the source of the justification of the belief. Since CJ2 does make a further claim about the source of the immediate justification for second-order beliefs, CJ2 does not obviously follow from the infallibilist conception of immediate justification. 41

There is a different way to introduce the alternative formulation (for a related point, see Smithies, this volume). One might say that Constitutivism concerns only “pure” or “wholly” introspective justification, where all pure introspective justification is immediate. Perhaps my earlier cases failed to target the view as it is properly understood, since the cases involved a mixture of introspective and nonintrospective justification. However, this response involves the constitutivist in unnecessary controversy. When I realize that I hope that p through counterfactual reasoning, I arguably have pure introspective justification, while still failing to have immediate introspective justification. Again, we should not take the case of pain as the only paradigm.

322

Constitutivism

The challenge for the constitutivist here is serious. She needs to distinguish her position from the odious inner observation theories. The inner observation theory could take on board infallibilism about immediate justification, by modeling itself on externalist views in the epistemology of perception, according to which only seeing that p can provide one with perceptual justification to believe that p, where one sees that p only if p.42 So the constitutivist needs a more discriminating defense of her position. A major further problem is that there is no reason to believe the infallibilist conception. As I emphasized in the introduction, immediate justification is not characterized by its strength, but instead by its lack of dependence on one’s justification for background beliefs. Given that immediate justification is not characterized by its strength, there is no reason to expect it to be infallible. CJ2 needs special pleading in its defense. The most promising argument for CJ2 I am aware of returns to a “self-intimation” thesis.43 The argument relies in particular on the idea that absences of belief are strongly self-intimating, as well as the idea that you cannot have justification to believe each of two contradictory propositions at the same time: (H) If one does not believe that p, then one has immediate introspective justification to believe one does not believe that p. (I) If one has justification of any kind to believe that p, then one does not have justification of any kind to believe it is not the case that p. So, (J) If one does not believe that p, then one does not have immediate introspective justification to believe one believes that p. The conclusion of this argument is equivalent to the claim that, if one does have immediate introspective justification to believe that one believes that p, then one does believe that p. The conclusion is thus quite close to CJ2, although CJ2 does make the further claim that one’s belief that p is the source of one’s introspective justification. Setting aside the problem that an inner observation view has yet to be excluded, the main problem with the argument is that it equivocates. For the claims about self-intimation to be correct, they must be read in terms of prima facie justification. When one seems to not have a belief, it is still possible to gain evidence that one has the belief, say from one’s therapist. For the crucial claim about incompatibility 42

For views in this vicinity, see McDowell (1995) or Johnston (2006). Conversations with Declan Smithies and Daniel Stoljar were extremely helpful here. Smithies (this volume) provides a further case for CJ2 via his account of the nature of “introspective justification.” See n. 24 of this chapter for my line of response. 43

Judgment as a Guide to Belief

323

to be correct, however, it must be understood in terms of all things considered justification—there is no difficulty in having prima facie justification for each of two contradictory claims. The argument is therefore invalid when the premises are construed in the way in which they might be true. I am aware of no better way to defend the view that introspective justification is infallible. The constitutivist challenge to the moderate view does not succeed.

4. Conclusion Although there is much disagreement about self-knowledge, philosophers in the literature currently tend to agree that it should not be understood on the model of perception. They focus on the metaphysics of perceptual states and on the absence of appropriately similar states in the case of introspection. The idea is that there is no good sense in which we perceive our mental states. The current focus on metaphysics has obscured parallels between the epistemology of perception and the epistemology of introspection—in each case a state can give a kind of justification that is immediate yet fallible. Just as an experience can give one immediate justification to believe that p, even though one can have the experience when it is not the case that p, a judgment can give one immediate justification to believe that one believes that p, even though one can make the judgment without believing that p. That is the proper understanding of the poorly understood “transparency of belief.” Judgments play such a role in our introspective lives, not all the work can be done by beliefs themselves. I would hold that there is a noninferential yet fallible structure in many other cases of introspection. Consider our access to factive mental states such that of seeing that p or remembering that p, where one can be in such mental states only if it is the case that p. Or consider our access to relational mental states such as that of seeing o, where one sees o only if one is appropriately interacting with the thing. In each of these cases, we need to be able to account for beliefs that are firstpersonally justified, yet false. To get a good view of the epistemology of introspection, then, we should not look away from the epistemology of perception. In developing my account of the transparency method, I have emphasized the role of conscious judgment in our introspective lives. Doing this work should improve our understanding of consciousness itself. Let me briefly sketch two potential upshots. First consider the epiphenomenalist view that consciousness plays no causal role. In order to form a second-order order belief on the basis of a conscious judgment, however, it looks like the belief must be caused or causally sustained by the judgment. We thus have a new reason to avoid epiphenomenalist

324

Constitutivism

views of consciousness—consciousness must play a causal role to play its epistemic role. Second, our work should also inform current debate about the phenomenology of cognition.44 What is it like, if anything, to think that p? Given that different conscious judgments can justify us in self-attributing different beliefs, we might expect judgments with different contents to have different conscious characters. If what it is like to judge that p were the same as what it is like to judge the different proposition that q, it would be unclear how the judgments could still differ with respect to which self-attributions they justify. To see the point, consider the following (imperfect) analogy: if what it is like to see redness were the same as what it is like to see greenness, it would be unclear how the color experiences could differ with respect to which color attributions they justify. Our work therefore suggests there is some support to views on which the phenomenology of cognition is fairly rich. To develop and assess the argument in detail is a further matter.45

References Berker, S. 2008. “Luminosity Regained.” Philosophers’ Imprint 8: 1–22. Bilgrami, A. 2006. Self-Knowledge and Resentment. Cambridge, Mass.: Harvard University Press. Boyle, M. 2009. “Two Kinds of Self-Knowledge.” Philosophy and Phenomenological Research 78: 133–164. Brueckner, A. 1988. “Moore Inferences.” Philosophical Quarterly 48: 366–369. Burge, T. 1996. “Our Entitlement to Self-Knowledge.” Proceedings of the Aristotelian Society 96: 91–116. Byrne, A. 2005. “Introspection.” Philosophical Topics 33: 79–104. ———. 2010. “Knowing That I am Thinking.” In Self Knowledge, edited by A. Hatzimoysis. Oxford: Oxford University Press. Cassam, C. 2007. The Possibility of Knowledge. Oxford: Oxford University Press.

44

For some discussion, see Siewert (1998) or Pitt (2004). Thanks to seminar participants at Cornell, and audiences at an Arché Basic Knowledge Workshop, Princeton, Columbia, the Hong Kong Towards a Science of Consciousness conference of 2009, and the Research School of Social Sciences at the Australian National University.Thanks in particular to Paul Benaceraff, Selim Berker,Alex Byrne, David Chalmers, John Collins, Shamik Dasgupta, Dylan Dodd, Alan Hájek, Gilbert Harman, Mark Johnston, Thomas Kelly, Brian Kim, Boris Kment, Maria Lasonen Aarnio, Sarah-Jane Leslie, Errol Lord, Angela Mendelovici, Ram Neta, Derk Pereboom, James Pryor, Carol Rovane, Daniele Sagravatti, Kranti Saran, Jonathan Schaffer, Kieran Setiya, Sydney Shoemaker, Ralph Wedgwood, Timothy Williamson, Elia Zardini, and an anonymous reviewer for Oxford University Press. Finally, thanks especially to Declan Smithies and Daniel Stoljar. 45

Judgment as a Guide to Belief

325

Chalmers, D. 2003. “The Content and Epistemology of Phenomenal Belief.” In Consciousness: New Philosophical Perspectives, edited by Q. Smith and A. Jokic. Oxford: Oxford University Press. Cohen, S. 2002. “Basic Knowledge and the Problem of Easy Knowledge.” Philosophy and Phenomenological Research 65: 309–329. ———. 2005. “Why Basic Knowledge Is Easy Knowledge.” Philosophy and Phenomenological Research 70: 417–430. Conee, E. 2005. “The Comforts of Home.” Philosophy and Phenomenological Research 70: 444–451. Davies, M. 2004. “Epistemic Entitlement, Warrant Transmission and Easy Knowledge.” Aristotelian Society, Supplementary Volume 78: 213–245. Edgley, R. 1969. Reason in Theory and Practice. London: Hutchinson. Evans, G. 1982. The Varieties of Reference. Oxford: Oxford University Press. Fernández, J. 2003. “Privileged Access Naturalized.” Philosophical Quarterly 53: 352–372. ———. 2005. “Self-Knowledge, Rationality, and Moore’s Paradox.” Philosophy and Phenomenological Research 71: 533–556. Fine, K. 1994. “Essence and Modality.” Philosophical Perspectives 8: 1–16. Gallois,A. 1996. The World Without the Mind Within. Cambridge: Cambridge University Press. Gertler, B. 2010. “Self-Knowledge and the Transparency of Belief.” In Self Knowledge, edited by A. Hatzimoysis. Oxford: Oxford University Press. Goldstein, L. 2000. “Moore’s Paradox.” In Believing and Accepting, edited by P. Engel. Dordrecht: Kluwer Academic Publishers. Gordon, R. 1995. “Simulation Without Introspection or Inference from Me to You.” In Mental Simulation: Evaluations and Applications, edited by T. Stone and M. Davies. Oxford: Blackwell. ———. 2007. “Ascent Routines for Propositional Attitudes.” Synthese 159: 151–165. Green, M., and J. Williams. 2007. Moore’s Paradox: New Essays on Belief, Rationality, and the First Person. Oxford: Oxford University Press. Hájek, A. 2007. “My Philosophical Position Says ‘p’ and I Don’t Believe ‘p’.” In Moore’s Paradox: New Essays on Belief, Rationality, and the First Person, edited by M. Green and J. Williams. Oxford: Oxford University Press. Hazlett, A. 2010. “The Myth of Factive Verbs.” Philosophy and Phenomenological Research 80: 497–522. Heal, J. 2002. Mind, Reason, and Imagination. Cambridge: Cambridge University Press. Horgan, T., and U. Kriegel. 2007. “Phenomenal Epistemology: What Is Consciousness That We Know It So Well?” Philosophical Topics 17: 123–144. Hyman, J. 1999. “How Knowledge Works.” Philosophical Quarterly 49: 433–451. Johnston, M. 2006. “Better Than Mere Knowledge? The Function of Sensory Awareness.” In Perceptual Experience, edited by T. S. Gendler and J. Hawthorne. New York: Oxford University Press.

326

Constitutivism

Lewis, D. 1995. “Should a Materialist Believe in Qualia?” Australasian Journal of Philosophy 73: 140–144. Reprinted in his Papers in Metaphysics and Epistemology (Cambridge: Cambridge University Press, 1999). Manley, D. 2007. “Safety, Content, Apriority, Self-Knowledge.” Journal of Philosophy 104: 403–423. McDowell, J. 1982. “Criteria, Defeasibility, and Knowledge.” Proceedings of the British Academy 68: 455–479. Reprinted in Perceptual Knowledge, edited by J. Dancy (Oxford: Oxford University Press, 1988) ———. 1995. “Knowledge and the Internal.” Philosophy and Phenomenological Research 55: 877–893. Moran, R. 2001. Authority and Estrangement. Cambridge, Mass.: Harvard University Press. Neta, R. 2010. “The Nature and Reach of Privileged Access.” In Self Knowledge, edited by A. Hatzimoysis. Oxford: Oxford University Press. Neta, R., and G. Rohrbaugh. 2004. “Luminosity and the Safety of Knowledge.” Pacific Philosophical Quarterly 85: 396–406. Peacocke, C. 1996. “Entitlement, Self-Knowledge and Conceptual Redeployment.” Proceedings of the Aristotelian Society 96: 117–158. ———. 1999. Being Known. Oxford: Oxford University Press. ———. 2008. Truly Understood. Oxford: Oxford University Press. Pereboom, D. 2011. Consciousness and the Prospects of Physicalism. Oxford: Oxford University Press. Pitt, D. 2004. “The Phenomenology of Cognition, or What Is It Like to Think that p?” Philosophy and Phenomenological Research 69: 1–36. Prinz, J. 2004. “The Fractionation of Introspection.” Journal of Consciousness Studies 11: 40–57. Pryor, J. 2005. “There Is Immediate Justification.” In Contemporary Debates in Epistemology, edited by M. Steup and E. Sosa. Oxford: Blackwell. ———. Forthcoming a. “Uncertainty and Undermining.” Available at . ———. Forthcoming b. “When Warrant Transmits.” Available at . Ryle, G. 1949. The Concept of Mind. London: Hutchinson. Schiffer, S. 2004. “Vagaries of Justified Belief.” Philosophical Studies 119: 161–184. Schwitzgebel, E. Forthcoming. “Acting Contrary to Our Professed Beliefs, or the Gulf between Dispositional Belief and Occurrent Judgment.” Searle, J. 1983. Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press. Setiya, K. Forthcoming. “Knowledge of Intention.” Shah, N., and Velleman, D. (2005). “Doxastic Deliberation.” Philosophical Review 114(4): 497–534.

Judgment as a Guide to Belief

327

Shoemaker, S. 1995. “Moore’s Paradox and Self-Knowledge.” Philosophical Studies 77: 211–228. ———. 1996. The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press. ———. 2009. “Self-Intimation and Second-Order Belief.” Erkenntnis 71: 35–51. Siewert, C. 1998. The Significance of Consciousness. Princeton: Princeton University Press. Silins, N. 2008. “Basic Justification and the Moorean Response to the Skeptic.” Oxford Studies in Epistemology 2: 108–140 ———. 2011. “Seeing through the ‘Veil of Perception’.” Mind 120: 329–367. ———. Forthcoming. “Introspection and Inference.” Philosophical Studies. Smithies, D. Forthcoming. “Mentalism and Epistemic Transparency.” Australasian Journal of Philosophy. Sosa, E. 1999. “How To Defeat Opposition to Moore.” Philosophical Perspectives 33: 141–153. ———. 2003. “Privileged Access.” In Consciousness: New Philosophical Perspectives, edited by Q. Smith and A. Jokic. Oxford: Oxford University Press. Stoljar, D. 2009. “The Argument from Revelation.” In Conceptual Analysis and Philosophical Naturalism, edited by D. Braddon-Mitchell and R. Nola. Cambridge, Mass.: MIT Press. Sutton, J. 2007: Without Justification. Cambridge, Mass.: MIT Press. Tye, M. 2002. “Representationalism and the Transparency of Experience.” Nous 36: 137–151. Unger, P. 1975. Ignorance. Oxford: Oxford University Press. Vahid, H. 2005. “Moore’s Paradox and Evans’s Principle: A Reply to Williams.” Analysis 65: 337–341. Way, J. 2007. “Self-Knowledge and the Limits of Transparency.” Analysis 295: 223–230. Weatherson, B. 2007. “The Bayesian and the Dogmatist.” Proceedings of the Aristotelian Society 107: 169–185. White, R. 2006. “Problems for Dogmatism.” Philosophical Studies 131: 525–557. Williams, J. 2004. “Moore’s Paradox, Evans’s Principle, and Self-Knowledge.” Analysis 64: 348–353. Williamson, T. 2000. Knowledge and Its Limits. Oxford: Oxford University Press. ———. 2004. “Skepticism.” In The Oxford Companion to Analytical Philosophy, edited by F. Jackson and M. Smith. Oxford: Oxford University Press. Wright, C. 2002. “(Anti)-Sceptics Simple and Subtle: Moore and McDowell.” Philosophy and Phenomenological Research 65: 330–348. Zimmermann, A. 2004. “Unnatural Access.” Philosophical Quarterly 54: 435–438. ———. 2006. “Basic Self-Knowledge: Answering Peacocke’s Criticisms of Constitutivism.” Philosophical Studies 128: 337–379.

This page intentionally left blank

12 Discrimination and Self-Knowledge Patrick Greenough

1. Preamble On a certain strong Cartesian conception of the mental, nothing in our mental life is hidden from us: if one is in a certain token mental state then one knows, or at least can easily come to know, via introspection, that one is in that state. On such a view, introspection is an absolutely reliable method of coming to know what mental states we are in. Nobody these days sponsors such a strong view of the mental, if anyone ever really did.1 The state of feeling angry, the state of feeling jealous, or the state of feeling schadenfreude, to name but three examples, are all mental states which one can be in and yet one can easily fail to know, via introspection, that one is in such a state—even if one is fully rational, functioning normally, and giving the matter one’s full attention. Still, the Cartesian conception survives in a variety of weaker forms. On one prominent weaker form, nothing in our core mental life is hidden from us: if one is in a core mental state, then one knows, or at least can easily come to know, via 1

It is not entirely clear that Descartes himself held such a strong conception of the mental— though Williams (1978, 84–85) tentatively attributes a version of such a view to Descartes, as does Dicker (1993, 45–48). Locke (1690) seems to have held such a view when he says: “There can be nothing within the mind that the mind itself is unaware of.”

330

Constitutivism

introspection, that one is in such a state—where prototypical core mental states are such states as being in pain, feeling cold, and feeling hot.2 On such a weaker view, introspection is still an absolutely reliable method—but only when introspecting such core mental states as these (cf. Shoemaker 1996, ch. 3). The aim in this chapter is to show not only that this weaker Cartesian conception is untenable but also that various even weaker, and more plausible, conceptions are unworkable too. The first of these entails: if one is in a core mental state, then one is in a position to form a justified belief, via introspection, that one is in such a state. The second retreats to the doxastic principle: if one is in a core mental state (and one has actively wondered whether one is in that state), then one believes, via introspection, that one is in that state. The third retreats to the principle: if one believes, via introspection, that one is in a certain mental state, then one knows that one is in that mental state. The fourth retreats to the weaker principle of infallibility: if one believes, via introspection, that one is in a certain mental state, then one is in that mental state. In other words, all the principles that have traditionally taken to be hallmarks of our (core) mental states must be given up. In order to show this, I offer a variety of (novel) conceptions of what it is for our introspective powers of discrimination to be less than perfect. The upshot is, I hope, a much more powerful case against the Cartesian conception of the mental than has been advanced hitherto.

2. Williamson on Anti-Luminosity One prominent set of doubts concerning the weak Cartesian conception that nothing in our core mental life is hidden from us concerns the absolute reliability of introspection.3 The worry is that since our introspective powers of discrimination 2 Further candidate core mental states include phenomenal conditions of the form it appears to one that p (see Williamson 2000, 96). 3 The term ‘introspection’, here and throughout, is used in a fairly theory-neutral sense. It is taken to encompass a variety of potential processes via which a subject is canonically able to become aware of, or form beliefs about, the occurrence of token mental states, processes, or events. Consequently, it is taken to apply to both core and noncore mental states. On such a minimal conception, we can remain neutral on a variety of issues. Namely, the issue as to whether introspection is an interpretative or inferential process (see, e.g., Carruthers 2010, 76–79; Byrne 2005, also this volume); the issue as to whether introspection is akin to perception (see, e.g., Armstrong 1968, 323–338; Shoemaker 1986; Gertler, this volume; Horgan, this volume); the issue as to whether there is a plurality of introspective methods (see, e.g., Schwitzgebel, this volume), and the issue as to the relationship between introspected and conscious states. (Cf. the less minimal conception adopted by Carruthers 2010, 76–79.)

Discrimination and Self-Knowledge

331

are not absolutely perfect, then this conception is still too strong. But what does it mean to say that our introspective powers of discrimination are not absolutely perfect? Very roughly, it means that introspection is not sufficiently discriminatory to tell apart close cases. In other words, close cases are indiscriminable via introspection. As we shall see, there are various ways in which one can unpack such a thesis. Williamson (1994, 1996, 2000) suggests that it amounts to the following conception of limited discrimination: (DIS 1): If one is in a case α, and a case α' is close enough to α, then for all one knows, via introspection, one is in α' For example, if one is in the state of feeling hot at time t, then for all one knows, via introspection, at some later time t plus one second, say, one feels hot. On that basis,Williamson provides an argument against the thesis that our core mental states are, what he terms, luminous—where a condition C is luminous for subject s just in case if C obtains then s is in a position to know, via introspection, that C obtains. If this argument is sound, then there are no (nontrivial) luminous conditions. For that reason, Williamson alleges, we are ‘cognitively homeless’—there is no theater of thought and experience within which our core mental states obtain and to which we have some special kind of epistemic access. Let us review Williamson’s argument in a bit more detail. The thumbnail version of Williamson’s anti-luminosity argument is as follows:4 The main idea behind the argument against luminosity is that our powers of discrimination are limited. If we are in a case α, and a case α' is close enough to α, then for all we know we are in α'. Thus what we are in a position to know in α is still true in α'. Consequently a luminous condition obtains in α only if it also obtains in α', for it obtains in α only if we are in a position to know that it obtains in α. In other words, a luminous condition obtains in any case close enough to cases in which it obtains. (2000, 13) We can unpack this argument as follows: first, consider a gradual phenomenal transition whereby a subject s knows, at time t, that she feels hot, but does not feel hot, and so does not know she feels hot, at time t plus one hour, where at every state of this transition the subject is in a position to wonder whether or not they feel hot. At t, s forms the belief that they feel hot; at the end of the process, s forms the belief that they do not feel hot. Second, given our limited powers of discrimination with respect to introspection, then knowledge of the obtaining of a condition C (e.g., feeling hot) requires what Williamson calls ‘a margin for error’, as follows: 4 The subtler version of Williamson’s argument (2000, 96–98) applies reliability considerations to degrees of confidence. For simplicity, I focus on the less subtle version.

332

Constitutivism

(ME): For all cases α, β, if in α s is in a position to know, via method M, that C obtains, then C obtains in β (where β is close to α).5 Very roughly, ME says that one is in a position to know that C obtains only if it is not an easy possibility that C fails to obtain.6 This is just to say that if C obtains in α, and case β is close enough to α, then for all one is in a position to know, via M in β, C obtains, which is just a version of DIS1.7 Third, suppose that C is luminous: (L): For all cases α, if C obtains in α, then in α s is in a position to know, via M, that C obtains (see Williamson 2000, 95). Fourth, ME plus L straightforwardly entail (via the transitivity of the conditional) the following soritical principle: (SR): For all cases α, β, if in α s is in a position to know, via M, that C obtains, then in β s is in a position to know, via M, that C obtains (where β is close to α). Given SR, and given that s is in a position to know via introspection that they feel hot at time t, then s is in a position to know via introspection that they feel hot in all cases—in particular at time t plus one hour. But it is given that s does not feel hot at time t plus one hour and so s is not in a position to know that they feel hot at time t plus one hour—since being in a position to know is a factive state. Contradiction. Upshot: the condition of feeling hot is not luminous. Likewise for all other core mental states. Thus, the weak Cartesian conception, under which these core mental states are taken to be luminous, must be abandoned. There are three immediate points of note. First, Williamson’s anti-luminosity argument can also be employed to show that there are no (nontrivial) ‘negative’ (core) luminous mental states. This is because being in a position to know that these states obtain also requires a margin for error. So, the conditions of not being in pain, not feeling cold, and so on are also nonluminous. Second, Williamson takes his margin for error principle ME to be derived from the following safety principle on knowledge: (S): For all cases α, β, if in α s is in a position to know, via M, that C obtains, then in β it is not the case that: s believes, via M, that C obtains and it is not the case that C obtains (where β is close to α) plus the doxastic principle: 5 Cases can be thought of as comprising a subject, a time, and a possible world. The quantifiers should be taken to range over physically and psychologically feasible cases for normally functioning human subjects. For the purposes of this chapter, method M is introspection. 6 For more on what it is to be in a position to know, see Williamson 2000, 95. 7 I here assume that for all one is in a position to know, C obtains entails one is not in a position to know that C does not obtain.

Discrimination and Self-Knowledge

333

(B): If s forms a belief, via M, in a case α, then s could easily have formed this belief, via M, in a close (but distinct) case β (see Williamson 2000, 126–129). Very roughly, principle S says that one knows, via M, that C obtains only if one could not easily have formed the false belief, via M, that C obtains. Principle B, as Williamson (2000, 127) notes, is supposed to capture the idea that ‘belief is not perfectly discriminating’—very roughly, belief has the tendency to spill over, as it were, to close (but distinct) cases. More specifically, if one forms the belief that C obtains in a certain case, then there will always be some close (but distinct) case in which one also forms that belief via the same method. Third, while B is required to derive ME from S, S follows directly from ME without the use of B. Thus, if one rejects the margin for error principle ME, and holds on to the principle of luminosity L, one is committed to rejecting either the safety principle S or the doxastic principle B. However, if one rejects safety as a condition on knowledge, one must also reject the margin for error principle L, whether or not one takes B to be valid.

3. Worries There are various worries one might have with Williamson’s anti-luminosity argument. In the next few sections, I will merely focus on the following concerns: 1. The argument depends on a particular conception of reliability—a safety conception. It would be much better if a cogent argument against luminosity could be motivated on more neutral grounds.8 2. Even if a safety conception of reliability is correct, nonetheless, on certain conceptions of knowledge, the safety principle S is too strong. If that is so, then the margin for error principle ME is likewise too strong—since ME entails S. But Williamson’s anti-luminosity argument is only valid given such a strong version of ME (cf. Berker 2008). 3. Related to the above worry, the argument depends on a particular conception of what it is for our powers of discrimination to be limited (as given by DIS1). Again, it would be better if a cogent anti-luminosity argument could be given that employed a more neutral conception. 4. The use of principle B might be disputed (cf. Steup 2009). 8

S is questioned by Brueckner and Fiocco (2002) and Neta and Rohrbaugh (2004). It seems to me that the doubts raised in these papers against safety can be accommodated by recognizing that whether a belief is safe is a highly time-sensitive matter.

334

Constitutivism

5. The argument cannot be used to undermine various weaker (and more plausible) principles. So, even if one takes Williamson’s anti-luminosity argument to be sound one might nonetheless seek to defend a weaker form of the Cartesian conception via one, or more, of these weaker principles.

4. Strong and Weak Safety The safety principle S is a pretty demanding principle: a subject s in a case α fails to know, via M, that C obtains if there is just one case β, which is close to α in which s forms the false belief, via M, that C obtains. Such a principle entails, for example, that I do not know that my ticket in the National Lottery will be a losing ticket when the draw is made, despite the overwhelming odds that it will be. That is because in one close world my ticket is the winning ticket and yet I believe it is a losing ticket. Likewise, the margin for error principle ME also rules out such knowledge. Most epistemologists think that safety and ME are getting the right results, even if they do not accept these very principles.9 But suppose one thinks that one can know that one’s ticket in the National Lottery will be a losing ticket, then both S and ME are too demanding.10 To allow for this, one natural weakening of S is: (WS): For all cases α, if in α s is in a position to know that C obtains, then in nearly all (if not all) cases β it is not the case that: s believes that C obtains and it is not the case that C obtains (where β is close to α).11 But now the most that can be derived from WS, together with B, is the following weaker margin for error principle: (WME): For all cases α, if in α s is in a position to know that C obtains, then in nearly all (if not all) cases β C obtains (where β is close to α). 9

One might, for example, accept a sensitivity conception of reliabilism under s knows, via M, that C obtains only if were C to fail to obtain, then s would not believe, via M, that C obtains. Sensitivity and safety are independent conditions on knowledge (see Williamson 2000, ch.7 for some useful discussion). 10 Various forms of contextualism, for example, allow that one can know that the sentence “my ticket is a losing ticket,” as used in a low-standards context (where the stakes are low and the possibility of winning is not particularly salient), is true. On such conceptions, the relevant version of S is not valid relative to such low-standards (if one uses a fixed metric for closeness) or is valid but cases where one forms a false belief do not count as close. 11 WS is still strong enough to ensure that in the Goldman-Ginet barn façade case, Henry does not know that he is looking at a barn. A version of WS is defended by Pritchard (2005, 163).

Discrimination and Self-Knowledge

335

But then there is no immediate route to establish SR as was done above. So, on certain weaker safety conceptions of knowledge, luminosity remains in the running—or so it seems.

5. Minimal Margin for Error Principles As it turns out, wme can after all be used to undermine the luminosity principle but via a different form of argument than the one given by Williamson. To see why note that WME entails the following even weaker margin for error principle: (MME): There is no case α, β, such that in α s is in a position to know that C obtains and in β s is in a position to know that C fails to obtain (where β is close to α).12 Call this a minimal margin for error principle. This principle provides an alternative and more general conception of what it is for close cases to be indiscriminable via introspection: (DIS2): If one is in a position to know that one is in a case α, and a case α' is close enough to α, then for all one is in a position to know, via introspection, one is in α'. Indeed, given the factivity of being in a position to know, MME is entailed by, but does not entail, Williamson’s own strong margin for error principle ME. Likewise, DIS1 entails DIS2, but not vice versa.

6. Minimal Margins for Error and Luminosity To use MME to undermine luminosity requires an assumption that is not required for Williamson’s own anti-luminosity argument to go through: namely, the assumption that if a positive mental state C is luminous, then the corresponding negative state not-C is luminous too. This assumption is surely plausible.13 Given this assumption, the weak Cartesian conception is committed to the following instances of L as applied to the state of feeling hot: 12 Suppose there are two close cases α, β such that in α s is in a position to know that C obtains and in β s is in a position to know that C fails to obtain. Given WME, and the first conjunct of this supposition, C obtains in most of the closeβ cases. But if that is so, then in those cases, given WME, s is not in a position to know that C fails to obtain, contrary to the second conjunct of the supposition. Upshot: WME entails MME. 13 It is accepted by Shoemaker (1996, 51) and also Smithies (forthcoming) for example.

336

Constitutivism

(L1): For all cases α, if s feels hot in α, then in α s is in a position to know that they feel hot. (L2): For all cases α, if s does not feel hot in α, then in α s is in a position to know that they do not feel hot. The relevant instance of MME then is: there are no cases α, β, such that in α s is in a position to know that they feel hot and in β s is in a position to have know that they do not feel hot (where β is close to α). Substituting, L1 and L2 in this instance of MME yields: there are no cases α, β, such that in α s feels hot and in β s does not feel hot (where β is close to α). But that is classically equivalent to the soritical conditional: for all α, β, if in α s feels hot, then in β s feels hot.14 Given the phenomenal transition from feeling hot to not feeling hot, we can thus derive a contradiction. The upshot is that even if one accepts a weaker conception of safety via WS, we can still motivate a weaker, and more generally acceptable, conception of what it is for our introspective powers of discrimination to be limited under which luminosity fails (at least given B, together with the assumption that if a positive mental state C is luminous, then not-C is luminous too). We have thus found an anti-luminosity argument that is motivated on significantly broader grounds than Williamson’s own version.

7. Motivating Anti-Luminosity without Safety or Principle B The question then arises: Do we need either B or any kind of safety principle to motivate an anti-luminosity argument? Recall that Williamson uses principle B, plus S to derive ME. Recall also that B amounts to the claim that ‘belief is not perfectly discriminating’. The use of B is somewhat odd. In the first place, even if B is true, it does not sit too well with Williamson’s own knowledge-first epistemology under which knowledge is the basic epistemological notion and belief is decidedly 14 Even if one denies the classical equivalence because of doubts about classical logic, then the principle ‘there are no cases α, β, such that in α s feels hot and in β s does not feel hot (where β is close to α)’ still gives rise to a paradox: assume, for simplicity, that we have a one hundred case phenomenal transition from not feeling hot to feeling hot, and assume that in case β100 s does not feel hot and assume for reductio that in a case β99 that s feels hot. So, there is some case α, and some case β, such that in α s feels hot and in β s does not feel hot (where β is close to α). But that contradicts our principle and so we can conclude that in case β99 s does not feel hot. But then now assume for reductio that in case β98 s feels hot . . . and so on until we reach the absurd conclusion that in case β0, s does not feel hot.

Discrimination and Self-Knowledge

337

secondary.15 Here the worry is: How come insights about belief, and not knowledge, are underpinning the insight, along with safety, that our introspective powers of discrimination are limited? Does that worry depend on an unduly pure reading of Williamson’s knowledge-first epistemology? Arguably not. Williamson (1990) (rightfully it seems to me) observes that the notions of discriminability and indiscriminability are, first and foremost, epistemic notions— notions that are canonically defined in terms of knowledge. So, surely the insight that the methods via which we form our introspective beliefs are less than perfectly discriminatory ought to be canonically given in terms of knowledge. But then Williamson merely needs to cash this claim out in terms of ME—he does not need to go via the doxastic principle B together with the safety principle S. Moreover, if the basic insight behind the argument against luminosity is just that close cases are indiscriminable via introspection, then arguably that insight is better captured via WME than via ME. How so? Williamson is concerned in his 1990 book with what may be termed with numerical indiscriminability. An updated version of his conception goes as follows: an object a and an object b are indiscriminable via method M just in case one is not in a position to know, via M, that a and b are numerically distinct.16 As it turns out, it is MME, and not ME, that is the qualitative counterpart of this numerical notion.17 So, the insight that close cases are qualitatively indiscriminable is best captured via MME and DIS2, and not the stronger principles ME and DIS1.

15

One may, of course, also question whether B is true (or indeed question whether one can really use B to get from the safety principle S to ME). I will not questions these assumptions here. 16 As Williamson notes, indiscriminability is better thought of holding not between objects per se, but between presentations of objects. This important insight does not matter for present purposes.The conception here is updated, since in this earlier work Williamson speaks of activating the knowledge that a and b are distinct, rather than being in a position to know that a and b are distinct. 17 Proof: It is a given that: (1) for all α, β, if α = β then in α, C obtains iff in β, C obtains. From (1), via contraposition, we can derive: (2) for all α, β, if in α, C obtains but, in β, C does not obtain, then α and β are distinct. Given that a subject s is in a position to know (2), then via distribution over ∀, and closure and collection for being in a position to know, we have: (3) for all α, β, if s is in a position to know, in α, that C obtains and s is in a position to know that, in β, C does not obtain, then s is in a position to know that α and β are distinct. Given contraposition, from (3) we get: (4) for all α, β, if s is not in a position to know that α and β are distinct, then it is not the case that: s knows, in α, that C obtains and s knows that, in β, C does not obtain. Now, suppose that α, β range over close cases: (5) for all α, β, α is close to β. And suppose that close cases are numerically indiscriminable: (6) for all α, β, if α is close to β, then s is not in a position to know that α and β are distinct. It thus follows from (4), (5), (6) that close cases are qualitatively indiscriminable: (7) for all close cases α, β, it is not the case that: s knows, in α, that C obtains and s knows, in β, that C does not obtain, which is equivalent to MME. Thus MME and not ME is the qualitative counterpart of Williamson’s notion of numerical indiscriminability.

338

Constitutivism

A further reason to dispense with principle B comes from reflecting on the fact that if B is taken to be at all plausible, then the following principle is surely just as plausible: (KB): For all cases α, β, if in α s is in a position to know, via M, that C obtains, then, in β, s believes, via M, that C obtains (where β is close to α). KB says, in effect, that if one knows that C obtains, via M, then it is not an easy possibility that one fails to believe, via M, that C obtains. KB enforces the following conception of limited discrimination: (DIS3): If one does not believe via introspection that one is in a case α, and a case α' is close enough to α, then for all one is in a position to know, via introspection, one is in α'. Crucially, the strong margin for error principle ME follows directly from KB and S, and so we can dispense with principle B in motivating ME. However, if KB is at all plausible the following weaker principle is surely very plausible: (KNB): For all cases α, β, if in α s is in a position to know, via M, that C obtains, then in β s does not believe that C does not obtain (where β is close to α). This principle enjoins the following conception of what it is for our powers of discrimination to be limited: (DIS4): If one believes, via introspection, that one is in a case α, and a case α' is close enough to α, then for all one is in a position to know, via introspection, one is in α'. But now note that, given that failure to believe that C obtains entails failure to know that C obtains, KNB entails a version of the minimal margin for error principle MME (stated in terms of knowledge rather than being in a position to know). Since it is given that the subject is at all times wondering whether C obtains, then the difference between knowing that C obtains and being in a position to know that C obtains can be effectively ignored. In other words, KNB effectively entails MME. Likewise, DIS4 effectively entails DIS2. The upshot is that we do not need any kind of safety principle, weak or strong, plus B, to motivate MME; we merely need the principle KNB (and the equivalent principle DIS4). Which is to say that we have found a conception of what it is for our introspective powers of discrimination to be less than perfect that both undermines luminosity and does not depend in any way on a safety conception of reliabilism or on a purely doxastic conception of what it is for our introspection to be less than perfectly discriminatory.18 That is, we

18

It might be thought that one reason to prefer ME over MME is that the former principle can be used to underline the KK principle while the latter principle cannot. However, in Greenough “Discrimination and Access Internalism” (ms a), I show that MME can undermine the KK principle while a correlate principle can undermine the JJ principle.

Discrimination and Self-Knowledge

339

can motivate an anti-Cartesianism on significantly stronger grounds than those advanced by Williamson.

8. J-Luminosity Even if one accepts the soundness of an anti-luminosity argument that proceeds via the margin for error principle ME or via the minimal margin for error principle MME, one might still hope to save some form of the Cartesian conception of the mental via some weaker form of luminosity. There are three basic ways in which this strategy might be carried out: weaken the consequent of L, strengthen the antecedent of L, or do both.19 Here the focus will be mostly be on the first type of strategy.20 One natural fallback luminosity principle exploits the notion of being in a position to have a justified belief rather than the notion of being in a position to know.21 The idea is that while our limited powers of discrimination undermine luminosity, they do not undermine the following weaker principle of J-luminosity: (JL): For all cases α, if C obtains in α, then in α s is in a position to have a justified belief, via introspection, that C obtains. Can any form of Williamson’s anti-luminosity argument be used to undermine JL (and so, in turn, L)? It depends what is meant by ‘being in a position to have a justified belief ’. If there is no gap between having a justified belief that C obtains and knowing that C obtains, then Williamson’s anti-luminosity argument can also be used to undermine JL. However, suppose that being in a position to have a justified belief is weaker than being in a position to know. In particular, let us suppose that being in a position to have a justified belief is both nonfactive and merely satisfies the following weaker safety principle: (JS): For all cases α, if in α s is in a position to have a justified belief, via M, that C obtains, then in nearly all (if not all) cases β it is not the case that: s believes, via M, that C obtains and it is not the case that C obtains (where β is close to α).22 19 One might also restrict the range of the quantifiers in L, e.g., by having a more fine-grained conception of what constitutes a case or what constitutes a core mental state. I shall not be discussing such options here. 20 I discuss the second and third types of strategy in my “How to be a Neo-Cartesian” (ms b). 21 This is effectively the notion of lustrousness defended in Berker (2008). 22 This notion of justification is such that while one cannot have a justified belief that one is standing before a barn in the standard Goldman-Ginet barn-façade case, one can nonetheless have a justified belief that one’s ticket is not the winning ticket in a fair lottery (and so the relevant version of multi-premise closure for justified belief fails). Note also that JS is neutral as to whether being in a position to have a justified belief that C obtains entails that C obtains.

340

Constitutivism

Together with B, one can then derive the corresponding margin for error principle: (JM): For all cases α, if in α s is in a position to have a justified belief, via M, that C obtains, then in nearly all (if not all) cases β, C obtains in β (where β is close to α). Again, just as with the weak margin for error principle WME, it looks like there is no immediate route to establish a version of the soritical principle SR stated in terms of being in a position to have a justified belief. So, despite the validity of JS, J-luminosity looks to be in good standing and a weakened form of the Cartesian conception remains in the running.

9. Against J-Luminosity However, a little reflection ought to show that, just as WME entails MME, JM entails the following justificationist minimal margin for error principle: (MJM): There is no case α, β, such that in α s is in a position to have a justified, via M, belief that C obtains and in β s is in a position to have a justified belief, via M, that C fails to obtain (where β is close to α). MJM is designed to capture the idea that our powers of discrimination are limited in the following sense: (DIS5): If one is in a position to have a justified belief that one is in a case α, and a case α' is close enough to α, then for all one is in a position to justifiably believe, one is in α'.23 Now, just as above, we have all the materials for an argument that shows that MJM plus JL entails that there are no cases α, β, such that in α s feels hot and in β s does not feel hot (where β is close to α), which, again, is classically equivalent to the soritical conditional: for all α, β, if in α s feels hot, then in β s feels hot. (Assuming JL is taken to hold for both positive and negative core mental states.) There are four immediate points of note. First, MJM depends upon JS and B. But, as with MME, one might simply motivate MJM directly, or derive MJM from a principle cognate to NJB, namely: for all cases α, β, if in α s is in a position to have a justified belief, via M, that C obtains, then in β s does not believe that C does not obtain (where β is close to α). Second, Williamson’s own conception of limited discrimination via DIS1 and ME is not equipped to undermine JL. Third, given the 23 MJM entails MME (given that being in a position to know entails being in a position to have a justified belief), but not vice versa.

Discrimination and Self-Knowledge

341

assumption that that being in a position to know that C obtains requires being in a position to have a justified belief that C obtains, then L entails JL. So, under that assumption, an argument against J-luminosity also shows that luminosity fails too. Fourth, a general moral emerges: any notion of warrant or justification that is sufficiently interesting to sustain some suitably weakened form of Cartesian conception of the mental will always be strong enough to satisfy some weak margin for error principle along the lines of MJM. So, any sufficiently interesting luminosity principle along the lines of JL is destined to fail.24

10. B-Luminosity Perhaps the Cartesian has been looking in the wrong place. Perhaps the Cartesian conception of the mental is to be expressed using principles that do not (overtly) involve knowledge or justification (and cognate notions) but rather is to be given in terms of principles involving belief. On that basis, one further fallback principle is the following doxastic principle: (BL): For all cases α, if in α one is in a core mental state (and one has considered the matter), then one believes, in α via introspection, that one is in such a state.25 But even this principle can be shown to be false. There are (at least) two arguments via which one can do this. The first proceeds via the following doxastic margin for error principle: (BM): There is no case α, β, such that in α s believes, via M, that C obtains and in β s believes, via M, that C fails to obtain (where β is close to α).26

24

I take it that this point applies to a version of JL formulated using a propositional notion of justification as follows: (JL): For all cases α, if C obtains in α, then in α s is justified in believing that C obtains. Smithies (forthcoming) defends the intriguing idea that JL when stated using propositional rather than doxastic justification is in fact valid. Accordingly, he also denies that there is a propositional justification version of MJM. There is not space to address such a proposal here but let me record the worry that a propositional justification reading of JL fails to capture much of what was wanted from the Cartesian conception of the mental in the first place, since it allows that one can be propositionally justified in believing that one is in a certain mental state and yet one cannot avail oneself of this justification to form a justified belief that one is in such a state.What then constitutes a propositional justification that C obtains above and beyond the fact that C obtains? 25 This is often called self-intimation. 26 BM entails a version of MME (stated in terms of knowledge rather than in terms of being in a position to know), but not vice versa. Likewise BM entails a version of MJM (stated in terms of justified belief rather than being in a position to have a justified belief), but not vice versa. In other words, BM is a pretty strong principle.

342

Constitutivism

BM records the following purely doxastic conception of what it is for close cases to be indiscriminable. (DIS6): If one believes, via M, that one is in a case α, and a case α' is close enough to α, then for all one believes, via M, one is in α'. If BL applies to both positive and negative core mental states, then, via substitution of the relevant positive and negative instances of BL into BM, we can derive there are no cases α, β, such that in α s feels hot and in β s does not feel hot (where β is close to α), which, again, is classically equivalent to the soritical conditional: for all α, β, if in α s feels hot, then in β s feels hot. So, even BL it seems must be given up. Indeed, KL and JL both entail BL (if the subject is wondering whether C obtains), and so all forms of the Cartesian conception we have considered so far will fail too. One immediate worry with this argument is that BM is simply false. Here the (plausible) thought is that surely introspection is such that a competent, normally functioning subject just can form the belief, via introspection, at time t, that they feel hot and a moment later form the belief, via introspection, that they do not feel hot. In other words, introspection just can and indeed often does yield contrary beliefs across close cases. But this reply will not help. If one doubts BM, then one is also committed to a denial of MJM, since BM can be shown to entail MJM if BL is valid. How so? Suppose BL applies to both the positive mental states of feeling hot and the negative mental state of not feeling hot as follows: (BL1): For all cases α, if in α one feels hot (and one has considered the matter), then one believes, in α via introspection, that one feels hot. (BL2): For all cases α, if in α one does not feel hot (and one has considered the matter), then one believes, in α via introspection, that one does not feel hot.27 If introspection is a method via which we can gain knowledge of our mental states, then the following principle ought to valid also: (D): There is no case α such that in α s believes, via introspection, that C obtains and in α s believes via introspection that C does not obtain. Principle D just enforces the plausible thought that introspection, if at all reliable, should not issue in contrary beliefs in a single case. But then from D and from BL1 and BL2 we can derive the following principles of infallibility: (IN1): For all cases α, if one believes, in α via introspection, that one feels hot, then in α one feels hot. 27 In Shoemaker’s terminology (1996, 51), the state of feeling hot is strongly self-intimating if and only if it satisfies both BL1 and BL2, and weakly self-intimating if and only if it satisfies BL1. Shoemaker takes core mental states to be strongly self-intimating.

Discrimination and Self-Knowledge

343

(IN2): For all cases α, if one believes, in α via introspection, that one does not feel hot, then in α one does not feel hot. Note that we cannot derive BL1 and BL2 from IN1 and IN2 plus D.28 More generally, we can derive the following infallibility principle (which applies to both positive and negative core mental states): (IN): For all cases α, if one believes, in α via introspection, that C obtains, then in α C obtains from D plus BL. But now note that IN and MJM entail BM (again ignoring the difference between knowledge and being in a position to know). But if BM is invalid, while MJM is in good standing, it follows that IN must be given up. Since, IN follows from BL together with D, then BL must be given up too. Thus, one cannot stabilize the Cartesian conception by retreating to BL. Indeed, there is a further, independent against BL that does not rely on the assumption that our introspective powers of discrimination are less than perfect.

11. The Quick and Dirty Argument against Both Luminosity and Omniscience First, note that the luminosity principle L entails the following principle of omniscience: (KO): For all cases α, if C obtains in α (and s has actively wondered whether or not C obtains), then in α s knows that C obtains.29 Second, given that knowledge entails belief, KO entails BL. Third, given that L, KO, and BL are valid for both positive and negative core mental states, then from L we can derive BL1 and BL2. Fourth, consider again the gradual phenomenal transition whereby s feels hot at time t and does not feel hot at time t plus one hour, such that at every stage of this transition s has wondered whether or not they feel hot. There are stages midway through the transition whereby no belief either way need be triggered—s neither believes that they feel hot nor believes that they do not feel hot despite having attentively and fully considered the matter. It seems 28

Shoemaker (1986) alleges that we can derive BL1 and BL2 from IN1 and IN2 plus B when only the converse entailment is valid. 29 Or perhaps better: s has done what s is in a position to do with respect to knowing whether or not the condition C obtains.

344

Constitutivism

perfectly acceptable for an alert, normally functioning, and competent subject to fail to reach a verdict either way and so withhold belief as to whether or not they feel hot. Given BL1 and BL2, a contradiction follows: if one fails to form the belief that one feels hot, then via BL1, one does not feel hot. If one also fails to form the belief that one does not feel hot, then, via BL2, it is not the case that one does not feel hot. So, BL must be given up, and since BL follows from L, then the luminosity principle L must be given up too. Equally, consider a gradual and slow onset of toothache. At each stage in the process one considers whether or not one feels pain. It seems perfectly acceptable for an alert, normally functioning, and competent subject, at certain stages midway in the transition, to refrain from forming a belief as to whether or not they are in pain—one might feel a faint throbbing in one’s tooth and yet feel unsure that this feeling amounts to a feeling of being in pain and unsure that this feeling does not amount to feeling in pain. Doubtless, many competent subjects, indeed perhaps most subjects, may not express such doxastic hesitancy. However, all that is required to get the anti-luminosity argument in hand to work is for there to be one subject, who is fully alert, fully competent with the concepts in question, who is functioning normally, who is given sufficient time to reflect as to whether or not they are in the core mental state, and yet refrains from forming a belief as to whether or not the core mental state obtains when that state does/does not obtain. This style of argument involves no overt appeal to reliability considerations or to a margin for error. Rather, the quick and dirty argument utilizes a form of what Shoemaker (1996, 51) calls ‘first-person agnosticism’.30 Should one seek to deny that our introspective powers of discrimination are less than perfect then this quick and dirty argument remains effective.31 Moreover, note that JL entails the following version of omniscience: (JO): For all cases α, if C obtains in α (and s has actively wondered whether or not C obtains), then in α s has a justified belief that C obtains. JO, like KO, entails BL. So, if BL (as applied to both positive and negative core mental states) fails because of first-person agnosticism, then JO fails too.

30

Reliability considerations may play a covert role in that epistemically successful subjects may have evolved to fail to form beliefs where there is a danger that their beliefs may be false. 31 Shoemaker finds such agnosticism plausible for certain mental states (such as whether the subject has an Oedipus Complex) but thinks that such agnosticism is totally implausible when it comes to “sensations and ordinary beliefs and desires.” It is clear, however, that Shoemaker is not considering cases where a sensation or core mental state obtains, but only very faintly. In such cases, first-person agnosticism is surely very plausible—hence the quick and dirty argument. Cf. Greenough (2005, 175) on “the problem of silence.”

Discrimination and Self-Knowledge

345

12. Infallibility and the Cartesian Conception of the Mental Suppose one accepts all of the arguments given so far, but one now proposes that the Cartesian conception of the mental should simply validate the infallibility principle IN. Here the thought is that the Cartesian has been looking in the wrong place: infallibility and not luminosity is the true hallmark of the mental. Recall that while BL (as applied to both positive and negative core mental states) entails IN (given D), the converse entailment does not go through. So, it certainly looks like one could deny BL and L and yet retain IN. But is such an option really in the running? There does not seem to be any immediate (and uncontroversial) quick and dirty argument against IN as there is against BL. However, as it turns out, there are two kinds of arguments one can give against IN. The first involves principle KB (and so DIS3). The second, and arguably better argument, uses MME and the failure of BM.

13. The Anti-Infallibility Argument Using KB Suppose that our introspective beliefs concerning positive and negative core mental states are infallible. Crucial question: Is infallibility sufficient for knowledge? Since such beliefs are never false, then arguably that fact is both necessary and sufficient for them to constitute knowledge. In other words, it is not a necessary condition that such beliefs are such as to also satisfy BL. So, if IN is valid (for both positive and negative mental states), it follows that the following principle is valid also: (INK): For all cases α, if one believes, in α via introspection, that C obtains, then in α one knows that C obtains. Now recall principle KB: (KB): For all cases α, β, if in α s is in a position to know, via M, that C obtains, then, in β, s believes, via M, that C obtains (where β is close to α). KB together with INK entail the soritical principle SR (just as ME plus L entails SR). But since SR leads to a contradiction, and KB is in good standing, then INK must be given up. But since INK follows from IN, then IN must be given up also. Indeed, since IN follows from BL (given D), and since BL follows from KO (or WO), which follows from L (or JL), this anti-infallibility argument is also an anti-luminosity argument. Thus, KB is a much more interesting principle than ME, since it can be used to undermine all of the Cartesian principles we have encountered so far.

346

Constitutivism

14. The Master Anti-Infallibility and Anti-Luminosity Argument Still, suppose one thinks that KB is too strong.32 Suppose, in particular, that one thinks that knowledge via introspection can tolerate a bit of unbelief, as it were, in close cases. Can we then still undermine all forms of the Cartesian conception without using KB? We can. There is a master infallibility argument that proceeds via MME rather than KB. And since luminosity entails infallibility, this argument also functions as a master anti-luminosity argument also. The argument goes as follows: Take the introspective belief that one feels hot. Suppose that this belief is luminous: 1. For all cases α, if s feels hot in α, then in α s is in a position to know that they feel hot. Moreover, plausibly, if positive introspective beliefs are luminous, then so are negative introspective beliefs. And so, from (1) we have: 2. For all cases α, if s does not feel hot in α, then in α s is in a position to know that they do not feel hot. If one actively wonders whether one feels hot and one is in a position to know that one feels hot, then one will know, and so believe that one feels hot. And so, from (1) and (2) we have: 3. For all cases α, if in α one feels hot (and one has actively considered the matter), then one believes, in α via introspection, that one feels hot 4. For all cases α, if in α one does not feel hot (and one has actively considered the matter), then one believes, in α via introspection, that one does not feel hot (i.e., BL1 and BL2). But if introspection is at all reliable, then it had better not issue in contrary beliefs in a single case. And so, the following principle, principle D, is valid: 5. There is no case α such that in α one believes, via introspection, that one feels hot and in α one believes via introspection that one does not feel hot. From (3), (4), and (5), it follows, given some pretty simple logic, that these introspective beliefs are infallible: 6. For all cases α, if one believes, in α via introspection, that one feels hot, then in α one feels hot

32

Just as one might think that S and ME should be replaced with WS and WME, respectively.

Discrimination and Self-Knowledge

347

7. For all cases α, if one believes, in α via introspection, that one does not feel hot, then in α one does not feel hot (i.e., IN1 and IN2). Moreover, it is also highly plausible that all infallible beliefs constitute knowledge, and so any introspective belief that satisfies (6) and (7) will also satisfy: 8. For all cases α, if one believes, in α via introspection, that one feels hot, then in α one knows that one feels hot 9. For all cases α, if one believes, in α via introspection, that one does not feel hot, then in α one knows that one does not feel hot (i.e., the relevant instances of INK). Suppose further that our introspective powers of discrimination are less than perfect such that close cases are indiscriminable via introspection. That is, introspection is governed by the minimal margin for error principle MME: 10. There is no case α, β, such that in α one is in a position to know, via introspection, that one feels hot and in β one is in a position to know, via introspection that one does not feel hot (where β is close to α). Given (8) and (9), this entails the doxastic limited discrimination principle BM (given that knowing entails being in a position to know): 11. There is no case α, β, such that in α one believes, via introspection, that one feels hot and in β one believes, via introspection that one does not feel hot (where β is close to α). But (11) is false. The method of introspection, broadly construed at least, is such that competent subjects can and indeed just do form contrary beliefs across close cases (under normal conditions). Across a gradual phenomenal transition from feeling hot to not feeling hot, at one instant I may hold the belief, via introspection, that I feel hot, and yet, say, half a second later, I may now form the belief, via introspection, that I do not feel hot.33 But since (11) is false and given that introspection is not perfectly discriminatory, it follows, together with the thesis that infallible beliefs constitute knowledge, that

33

Of course, this is not to say that across all phenomenal transitions a subject goes from the belief that C obtains to the belief that C does not obtain across some close cases—sometimes one will simply go from belief that C obtains to a state of first-person agnosticism (as the quick and dirty argument shows). Rather, it is simply that a fully alert, rational, and normally functioning subject can and indeed on many occasions just does go from the belief that C obtains to the belief that C does not obtain across some close cases. Indeed, the phenomenon is akin to what subjects typically do in the so-called forced march sorites paradox; see Raffman 1994; Shapiro 2006.

348

Constitutivism

the belief that one feels hot and the belief that one does not feel hot are not both infallible. Indeed, since they are not both infallible, given that introspection does not yield contrary beliefs in any case, they not both luminous. Moreover, given that both or neither are luminous, and that both or neither are infallible, it follows that the belief that one feels hot is neither infallible nor luminous. We have thus shown that even considerably weakened forms of the Cartesian conceptions of the mental are unworkable—and moreover, we have done so without appealing to anything like the strong margin for error principle ME.34 Many issues remain. What role does and should vagueness play in these (and related) anti-Cartesian arguments? Should one endeavor to save the Cartesian conception of the mental via some strengthening of the antecedent of the luminosity principle L or the self-intimation principle BL? Can one save the Cartesian conception by invoking some form of response-dependent conception of phenomenal states? These, and other issues, are taken up elsewhere.35

34

A lingering worry may remain. Even if our introspective powers of discrimination are indeed less than perfect, one might think that one cannot use that very thesis in a persuasive argument against either luminosity or infallibility. For example, one might take the minimal margin for error principle MME to be too close to a failure of luminosity, in argumentative space, as it were, for an anti-luminosity argument to have any suasive force. There are two replies. First, note that this objection does not touch the quick and dirty anti-luminosity argument as this argument does not employ the thesis that our introspective powers of discrimination are less than perfect. Second, the weak Cartesian grants that introspection is less than perfectly discriminatory in respect of noncore mental states. Moreover, since the luminosity and infallibility principles are exceedingly strong principles, we need an independent argument from the weak Cartesian to show why introspection suddenly (and magically) becomes perfectly discriminatory with respect to our core mental states. In other words, the burden of proof is on the Cartesian to show why core mental states differ so dramatically from noncore mental states so as to allow core mental states to be luminous or infallible. In other words, in the absence of such an argument, we are entitled to assume that introspection is less than perfectly discriminatory with respect to both core and noncore mental states. 35 See Greenough “How to Be a Neo-Cartesian” (ms b). The main ideas in this chapter were presented at: Bergamo (The Italian Society for Analytic Philosophy), ANU, Melbourne (AAP annual conference), Leeds, Nottingham, and St. Andrews. For very useful feedback, particular thanks go to: Ralf Bader, Selim Berker, Dave Chalmers, Josh Dever, Paul Dimmock, Zoe Drayson, Padraig Graenfell, Jonathan Schaffer, Susanna Schellenberg, Declan Smithies, Daniel Stoljar, Michael Tye, Crispin Wright, and the two OUP referees.This work was begun while I was a Postdoctoral Fellow in the ARC funded Epistemic Warrant Project at ANU (2007–2008) and completed while I was a Postdoctoral Fellow in the ARC funded Pragmatic Foundations of Language Project at the Centre for Time, University of Sydney.

Discrimination and Self-Knowledge

349

References Armstrong, D. 1968. A Materialist Theory of the Mind. London: Routledge and Kegan Paul. Berker, S. 2008. “Luminosity Regained.” Philosophers’ Imprint 8: 1–22. Brueckner, A., and M. O. Fiocco. 2002. “Williamson’s Anti-Luminosity Argument.” Philosophical Studies 110: 285–293. Byrne, A. 2005. “Introspection.” Philosophical Topics 33: 79–104. Carruthers, P. 2010. “Introspection: Divided and Partly Eliminated.” Philosophy and Phenomenological Research 80: 76–111. Dicker, G. 1993. Descartes: An Analytical and Historical Introduction. Oxford: Oxford University Press. Greenough, P. 2005. “Contextualism about Vagueness and Higher-Order Vagueness.” Proceedings of the Aristotelian Society, Suppl. 78: 167–190. ———. ms a. “Discrimination and Access Internalism.” Manuscript. ———. ms b. “How to be a Neo-Cartesian.” Manuscript. Neta, R., and Rohrbaugh, G. 2004. “Luminosity and the Safety of Knowledge.” Pacific Philosophical Quarterly 85: 396–406. Pritchard, D. 2005. Epistemic Luck. Oxford: Oxford University Press. Raffman, D. 1994. “Vagueness Without Paradox.” Philosophical Review 103: 41–74. Shapiro, S. 2006. Vagueness in Context. Oxford: Oxford University Press. Shoemaker, S. 1986. “Introspection and the Self.” Midwest Studies in Philosophy 10: 101–120. ———. 1996. The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press. Smithies, D. Forthcoming. “Mentalism and Epistemic Transparency.” Australasian Journal of Philosophy. Steup, M. 2009. “Are Mental States Luminous?” In Williamson on Knowledge, edited by P. Greenough and D. Pritchard. Oxford: Oxford University Press. Williams, B. 1978. Descartes:The Project of Pure Inquiry. London: Penguin. Williamson, T. 1990. Identity and Discrimination. Oxford: Blackwell. ———. 1994. Vagueness. London: Routledge. ———. 1996. “Cognitive Homelessness.” Journal of Philosophy 93: 554–573. ———. 2000. Knowledge and Its Limits. New York: Oxford University Press.

This page intentionally left blank

Part IV Introspection and the Nature of Experience

This page intentionally left blank

13 Introspection, Explanation, and Perceptual Experience: Resisting Metaphysical Disjunctivism Aaron Zimmerman

1. Seeing, Experiencing, and Reflecting on Experience According to disjunctivists, veridical visual experience and hallucination do not have a common nature. I will argue that disjunctivism so defined is incompatible with commonsense views of introspection and psychological explanation. The preponderance of evidence therefore points away from disjunctivism toward a view on which hallucinatory and veridical visual experiences are representations of a common kind. To focus the discussion, I direct my arguments at the metaphysical form of disjunctivism defended by Michael Martin (2006): the claim that veridical visual experience and hallucination are not species of a single fundamental or explanatorily important genus. That is, according to Martin, there is no nondisjunctive, fundamental kind of thing—visual experience—such that hallucination and veridical experience are both types of things of this kind. I think there are good reasons to reject such a view.

354

Introspection and the Nature of Experience

Let us begin our assessment of the disjunctivist view by supposing that a subject, Max, does indeed see a sphere in front of him. And let us suppose that I establish this fact by locating the sphere, locating Max, and discerning a certain relation between them. When I verify that the latter sees the former, what fact do I therein come to know? If common thinking is any guide, I have established the existence of a complex event—an event that involves, as constituents, both Max and a sphere wholly distinct from him. Now we might ask, as metaphysicians of mind, whether the existence of this seen object or any other is an “essential” aspect of the complex event that is Max’s seeing this sphere. If this sphere were not in front of Max, and there were no other sphere in view, might not Max still be seeing something sphere-like in nature? Suppose that Max is hallucinating extremely hard, and while it looks like there is a sphere in front of him, he is really just staring into space. Does Max see something in this case? To suppose not, is to do no more than reject a particularly aggressive version of the sense data theory, as most of us now do. If there is nothing distinct from your mind before you, you are not seeing anything at all. So while it looks to the hallucinating subject as though there is a sphere in front of him, and he really does think that he is seeing a sphere, he is simply mistaken in this belief. Note, though, that there is a natural way of thinking about disjunctivist metaphysics of the sort embraced by Martin on which these relatively commonsense reflections on what it is to see something already result in a mild form disjunctivism, albeit a form of disjunctivism limited to the seeing of objects.1 Often when one hallucinates one does not see anything, as there is nothing there to be seen. (And a subject seriously impaired by hallucination often cannot see those things that are in fact there to be seen.) Seeing something is therefore an entirely different kind of thing than is hallucinating something. Seeing involves both an observer and a seen object. In contrast, hallucination is nothing more than an event in the observer’s mind. Of course, the disjunctivist view of perception advocated by Martin is much more controversial than this, as Martin’s thesis concerns visual experiences rather than instances of seeing things. The naïve realist claims that some sensory experiences are relations to mindindependent objects. That is to say, .  .  . taking experiences to be episodes or events, the naïve realist supposes that some such episodes have as constituents mind-independent objects. In turn, the disjunctivist claims that in a case of veridical perception like this very kind of experience that you now have, the 1 Timothy Williamson’s (2000) view of “seeing that p” is a comparably mild form of disjunctivism. Cf. John Hyman (1999) and (2001).

Introspection, Explanation, and Perceptual Experience

355

experiential episode that you enjoy is of a kind which could not be occurring were you having a hallucination. (2006, 354)2 Now Martin claims that his “naïve realist” account of visual experience, “best articulates how sensory experience seems to us to be just through reflection” (ibid.). Indeed, his stated goal in “On Being Alienated” is to save this wholly “reflective” understanding of experience from a skeptical train of thought: the kind of thinking that drives some philosophers to countenance sense data as the objects of hallucinatory experience, and lands many more into thinking that even veridical visual experience is representational in nature.3 So our first question is whether Martin is right about this. If our accurate visual experiences really are “as [they] seem to us to be just through reflection” must they have mind-independent objects as essential constituents? Indeed, this first question might turn out to be our last. For if we do not ordinarily take our accurate visual experiences to essentially involve extra-mental objects of perception, we will not have to worry with Martin about whether our visual experiences really have this feature. Of course, as we have already noted, our commonsense conception of someone’s seeing something is indeed a disjunctivist one. So I take it that what seeing something seems to us to be through “reflection” alone is something that (essentially) involves a mind-independent object as one of its constituents. But surely, whatever they are, visual experiences are supposed to be located in our heads. And if the common folk, when they reflect on their experiences, picture these episodes as unfolding in their heads, they are unlikely to think that a certain range of them—namely, the veridical ones—essentially involve things that are located outside of their heads. To be fair, this does not mean that Martin’s disjunctivist account of visual experience cannot be right. But it would suggest that few of us are going to be convinced of its truth in the absence of a compelling argument.4 It would suggest that Martin must do more than merely defend the disjunctivist account from skeptical assault.5 2 See too, “The Naïve Realist . . . claims that our sense experience of the world is, at least in part, non-representational. Some of the objects of perception—the concrete individuals, their properties, the events these partake in—are constituents of the experience . . . The motivation for disjunctivism, I suggest, is a desire to hold on to Naïve Realism” (Martin 2004, 273–274). 3 The modern sense data theorists against whom Martin argues include Jackson (1977), Robinson (1994), and Foster (2000); the representationalists are too numerous to name, though see Byrne (2001) for extensive citations. 4 I have in mind here an argument like Hilary Putnam’s (1975) famous case for the conclusion that the meanings of our words are not in our heads. Chalmers and Clark (1998) argue for an externalist account of the memories, thoughts, and other semantic “vehicles” that we commonly suppose are in our heads, an externalism that resembles Martin’s in greatly exceeding Putnam’s conclusions about meaning or semantic content. 5 Might Martin just mean by “having a veridical visual experience of x” what the folk mean by “seeing x”? It is telling that Susanna Siegel, before critiquing Martin’s view, feels she must inform

356

Introspection and the Nature of Experience

Indeed, one might wonder whether the folk even have the concept that philosophers associate with “veridical visual experience.” So one might doubt whether there is a pre-theoretical conception of veridical visual experience to rescue from skeptical critique. Admittedly, if our concept of visual experience is theoretical in nature, the theory is an extraordinarily old one, as philosophers since at least Aristotle have discussed cases in which the appearances of objects vary without change in the objects themselves. Why is it that to those who are very drunk everything seems to revolve in a circle, and as soon as the wine takes hold of them they cannot count objects at a distance. . . ? Is it because the vision is continually disturbed by the heat of the wine? The same thing then happens to those who are drunk as when an object appears double if one puts it close to the eye. (Aristotle [384–322 b.c.] 874a; 1984, 1346)6 In his famous discussion of “Skepticism with Regard to the Senses,” the young Hume lumps this observation together with others made in the intervening millennia. When we press one eye with a finger, we immediately perceive all the objects become double, and one half of them to be remov’d from their common and natural position. But as we do not attribute a continu’d existence to both these perceptions, and as they are both of the same nature, we clearly perceive that all our perceptions are dependent on our organs, and the disposition of our nerves the reader of a departure from ordinary thought and talk. “Throughout this essay, I’ll use ‘veridical experience’ to pick out instances of perception [e.g., seeing] that are not hallucinations or illusions. So there will be no such thing as a veridical hallucination in this sense of ‘veridical’” (2008, 205). Once the concept associated with “veridical experience” is distorted in this way—once it becomes analytic that hallucinations are inaccurate—“veridical experience” may well be cognitively equivalent to “seeing” and the events to which both terms apply may well have objects as their essential constituents and so differ (in at least this way) from even perfect hallucinations. (The question I am asking is whether Siegel could have achieved the same effect by saying, “‘Having a veridical visual experience of x’ is herein to be read ‘seeing x’.”) At any rate, I choose to think that Martin is not pushing for linguistic or conceptual reform. On the reading of the texts I find most natural, Martin is merely arguing (albeit on wholly a priori grounds) for naïve realism, where this implies no more than that having a veridical visual experience of something is a great deal like seeing it in that both events essentially involve an experienced/seen object as an extra-mental constituent. (That is, on my reading, Martin treats naïve realism as a synthetic—albeit an a priori knowable— truth.) I suspect that Martin also embraces the associated property identity—that having a veridical visual experience of something just is seeing something—but I am unsure of this, as there are passages (e.g., 2004, 294) in which Martin seems to countenance veridical hallucinations. 6 While of unquestionable antiquity, the authorship of “Problems” is disputed.

Introspection, Explanation, and Perceptual Experience

357

and animal spirits. This opinion is confirm’d by the seeming encrease and diminution of objects, according to their distance; by the apparent alterations in their figure; by the changes in their colour and other qualities from our sickness and distempers; and by an infinite number of experiments of the same kind; from all which we learn, that our sensible perceptions are not possest of any distinct or independent existence. (Hume [1739–40] 2000, T 1.4.2.45)7 Let us focus on the first of these experiments, the double vision of which both Aristotle and Hume speak. When I press my eyeball while looking at a sphere, something will of course change. (Hume is right about that.) But there is no reason to infer that what I am seeing changes in the case. (This is where Hume goes off the rails.) I am still just seeing a sphere, and I know that the sphere has not changed. (There remains a single sphere before me, and I know this without the aid of philosophy.) Instead, Hume’s experiment puts me in a position to infer that the change I am aware of—the doubling of my vision—must be caused by the depression of my eyeball. And since the depression of my eyeball is unlikely to cause a change in the sphere before me, it must instead change something in my head: a visual image, impression, sensation, or experience. Moreover, if common sense is any guide, the visual experience I become aware of when I conduct Hume’s experiment can be evaluated in something like that manner in which we evaluate maps, portraits, and other iconic representations. My experience is accurate prior to the depression, as there is then a single sphere before me, and that is how things then look. But my visual experience is then distorted when I depress my eyeball; it then becomes inaccurate. For after the eyeball’s depression there is a sense in which it now looks to me as though there is a second sphere, a sphere removed from the “common and natural” position occupied by its original, even though I know that there is no sphere at this second location. After the depression, things are not entirely as they look. It seems, then, that the conception of visual experience we get from a simple experiment known to philosophers for thousands of years is of something in the head that represents things outside the head in either an accurate or inaccurate manner. It is not, therefore, the naïve realist’s conception of visual experience. Now, what does all this show? It shows that when Martin writes of how “sensory experience seems to us to be just through reflection” he must not have in mind the experiments of Aristotle and Hume and the lessons common sense draws from them. It would seem, that is, that Martin is not using “reflection” to denote 7

As to the sources of these experiments, David and Mary Norton cite in addition to Aristotle: Lucretius, Sextus Empiricus, Hobbes, Montaigne, Rohault, and Collier. See their edition of the Treatise referenced below at 477 n. 45.

358

Introspection and the Nature of Experience

the hybrid of observation, introspection, and inference described above. For that method of thought leads us away from naïve realism toward a distinctively representational view of experience. Let us call the form of reflection driving Aristotle and Hume “a-introspection” and characterize it as follows. A-Introspection: Forming beliefs about one’s feelings, thoughts, and experiences making full use of any experiments, observations, and inferences that seem relevant. Clearly, a-introspection involves introspection properly so called, as Aristotle and Hume rely on introspective judgments about the “doubling” that marks their visual experience when their eyeballs are depressed or the object before them is too close to clearly see. But a-introspection clearly involves much else besides introspection. For instance, both theorists must infer their representationalist conclusions from their introspective observations and the further (extra-introspective) assumptions articulated above. Now I have been arguing that this augmented form of “introspection” yields a conception of visual experience at odds with naïve realism. The question for Martin, then, is whether there might be another more reliable guide to the nature of visual experience—a method or form of introspection that might lead us to discount the results of a-introspection and embrace Martin’s naïve realism instead.8

2. A Purely Introspective Conception of Experience When Aristotle and Hume reflected on the nature of their visual experiences, they felt free to conduct experiments and reason from their observations. And, as we have seen, this process of reflection speaks against naïve realism. So if a-introspection does not support naïve realism, what does? To answer this question, Martin turns away from observation and inference to the wholly introspective methods employed by phenomenologists. While looking at the sphere (with wholly undepressed eyeballs) I am supposed to simultaneously

8

Note that it would be wrong to suggest, as Martin sometimes does, that theorizing about the nature of hallucinations is the only thing driving theorists away from naïve realism. As the quotes from Aristotle and Hume make clear, hallucinations are only one among a variety of disturbances, disruptions, and patent inaccuracies in our typical visual experience that would seem to have proximate (or internal) causes and are consequently cited as evidence for the view that experiences are head-bound representations. Scott Sturgeon (2006, 2008) makes a related observation when he describes the spectrum of cases that has entirely accurate experience and pure hallucination as its “good” and “bad” endpoints.

Introspection, Explanation, and Perceptual Experience

359

see the sphere and introspect the largely accurate experiential episode I am then enjoying. And naïve realism is supposed to be the conception of my visual experience I find most congenial when I am introspecting in this manner. We can call this form of reflection “b-introspection” to distinguish it from its more inclusive rival. B-Introspection: Forming beliefs about one’s feelings, thoughts, and experiences in an “immediate” way—i.e. without relying on experiments, extra-introspective observations, or any inferences from them.9 Of course, even if we were convinced that the results of b-introspection support naïve realism, we would not yet be warranted in accepting that view. For we also know that the results of a-introspection run contrary to naïve realism. Thus, even if Martin convinces us that b-introspection really does support a naïvely realistic view of visual experience, we must still weigh our competing sources of evidence to determine which source, if either, should be trusted in the case on hand. Which guide to experience is more reliable: the a-introspective hybrid of introspection, observation, and inference that leads us to infer a representational view of experience from double vision and the like, or the purely a priori b-introspective method that is supposed to lead us into naïve realism? Now I do not think we can answer this difficult methodological question in a fully general way. For there are, admittedly, b-introspective judgments that are properly assigned enormous epistemic authority: my current judgment that I am in pain and my current judgment that I am thinking about disjunctivism are two clear examples. And it is indeed hard to see how the judgments in this privileged class could be overturned by experimentation, observation, and inference. Nevertheless, one might reasonably doubt whether Martin’s b-introspective judgment that his accurate visual experience involves an extra-mental object as an essential constituent falls into this same highly authoritative category (cf. Spener, this volume). For surely, highly abstract, highly theoretical judgments about one’s experience are at least more amenable to third-person refutation than are those core cases of introspective report on which discussions of first-person authority rightfully focus.

9 Eric Schwitzgebel equates introspection with what I have been calling “a-introspection” (this volume), and Nicholas Silins claims that introspectively justified beliefs can be grounded in inference from what is immediately known (this volume). But I can see no objection to isolating the noninferential aspects of the a-introspective process and introducing a label for them alone. The process so identified roughly corresponds to the cautious form of introspection identified by Brie Gertler (this volume). It is of course a further claim that when we use “introspection” we more often intend to pick out the less inclusive belief-forming process (b-introspection) than the more inclusive one (a-introspection). It may be that philosophers typically use “introspection” to pick out b-introspection while the usage of lay people is less restrictive.

360

Introspection and the Nature of Experience

But the case against metaphysical disjunctivism is even stronger than these reflections suggest. We do not have to content ourselves with arguing that a-introspective arguments for representationalism overwhelm the support naïve realism gains from b-introspection. For Martin is just wrong in thinking that the results of b-introspection support naïve realism. Instead, as I will now explain, it seems to me that both a-introspection and b-introspection support the representationalist view against its rival. Note that there are many different thoughts I can have about my visual relation to the sphere before me. Crucially, I can think or reflect on the fact that I am seeing the sphere and I can also reflect on the fact that I am having (or “enjoying”) a visual experience of it. And while it is relatively uncontroversial that the event that is my seeing the sphere has an extra-mental object as one of its (essential) constituents, if my attention is instead turned to my experience of the sphere—if I am thinking of what is now happening as an accurate experience of something extra-mental rather than as the seeing of that thing—I am not thinking of what is now happening in a naïvely realistic manner. Or, at any rate, Martin fails to argue that naïve realism supplies us with the best description of what I take my experience to be when I reflect in the manner he recommends. In sum, experiments known to the ancients lead most of us to posit visual experiences that are representational in nature. If I eschew these experiments and try to get a more direct fix on the kind of visual experience I am enjoying while seeing the sphere before me, I initiate a process of b-introspection. But to form beliefs about the experience I am having I must decide how to best conceptualize or characterize that experience. If I think of what is going on as my seeing the sphere, I am led to a naïvely realistic view of what is going on. But if I think of what is going on as my having an accurate visual experience of a sphere, I am not led into thinking of the episode as the naïve realist does. Instead, since this episode is nothing more than an experience (albeit an accurate one), it would seem to be something in my mind, something that could exist in the absence of that wholly external object of which it is an experience. Disjunctivism does not emerge from either a-introspection or b-introspection on our experiences but must instead be motivated with an independent argument or source of evidence.

3. Nonconceptual Purely Introspective Awareness of Experience Although Martin does not explicitly address the line of objection advanced above, there are crucial moments in his defense of disjunctivism in which I find him positing a nonconceptual form of introspective awareness of visual experience.

Introspection, Explanation, and Perceptual Experience

361

(Whether Martin really believes in nonconceptual awareness of experience is something I will address below.) And if we are indeed “acquainted” with our visual experiences when we introspect, Martin can invoke this awareness in reply to the case I have presented against him. Perhaps, as I have suggested, it is up to Max to decide whether to think about what is going on in the scenario I have described as his seeing a sphere or his having a veridical experience of such, where in the first case he is apt to think of what is going on in disjunctivist terms, and in the second case he is not. But Martin might insist that Max is also introspectively acquainted with what is going on in the case. He has, that is, an introspective awareness of what is going on that does not await his decision to think of it as an episode of seeing or a veridical visual experience. We can call the method we deploy to secure this kind of acquaintance with states of our own minds “c-introspection.” C-Introspection: Focusing one’s attention on one’s feelings, thoughts, or experiences without necessarily categorizing them (as, say, visual experiences or events of seeing) or drawing conclusions about what they really are. Now if there is such a thing as c-introspection, Martin might insist on its providing evidence for naïve realism. When Max c-introspects his visual experience of the sphere before him, Martin might claim, that experience will seem to him to essentially involve a sphere. Martin’s only remaining task, then, would be to argue, as he acknowledges he must, that visual experience really is as it is presented to us when we c-introspect. Our question, then, is whether there is such a thing as c-introspection, and if there is, whether this primitive take on our visual experiences presents them to us as essentially external-object-involving. Now there are those who doubt the very coherence of c-introspection. For example, Sydney Shoemaker rejects a “narrow perceptual model” of introspection by arguing that introspective knowledge is “fact awareness” not grounded in “thing awareness” (1996, ch. 10; cf. Tye, 2009b, ch. 5; and Sosa, this volume). According to Shoemaker, Max can know via introspection that he is having an experience of a sphere, and he can know (in some way or other) that he is seeing a sphere. But Max cannot be introspectively aware of the event in question if this is supposed to be something distinct from his awareness that it is an event of a certain type. According to Shoemaker, that is, Max cannot simply focus his attention on his visual experience without therein thinking of it as a visual experience, an event of seeing something, or an event of some other kind. In consequence, neither Max nor we can c-introspect. On Shoemaker’s lights, introspection is nothing more than the formation of introspective beliefs or judgments. Introspective judgment has no extraconceptual (or preconceptual) preliminary.

362

Introspection and the Nature of Experience

I have argued elsewhere that Shoemaker’s critique of the narrow perceptual model of introspection is too ambitious (Zimmerman 2008). Although our introspective knowledge of our beliefs and thoughts is indeed wholly nonperceptual in nature, our knowledge of our sensations and experiences resembles paradigmatic perceptual knowledge in various respects. For instance, in stark contrast with our thoughts and beliefs, we can quite easily describe mistaken judgments about our own experiences that do not have their source in irrationality or conceptual confusion. And a careful examination of these mistaken judgments unearths at least some evidence of c-introspective awareness of visual experience. The phenomenon is perhaps most clearly exemplified by a scenario Gareth Evans discussed: Consider a case in which a subject sees ten points of light arranged in a circle, but reports that there are eleven points of light arranged in a circle, because he has made a mistake in counting, forgetting where he began. Such a mistake can clearly occur again when the subject reuses the procedure in order to gain knowledge of his internal state: his report ‘I seem to see eleven points of light arranged in a circle’ is just wrong. (1982, 228–229) Evans’s subject does not have an irregularly malleable visual experience that morphs to match his miscalculations. Instead, he has a somewhat inaccurate view on the character and content of a largely stable stream of experience. He seems to see ten points of light in front of him, but he mistakenly thinks that he seems to see eleven. There are at least three features of the case worth noting. First, the events it describes are fairly unexceptional. Relevantly similar cases can be described that do not involve numerical concepts, as when one is asked by a doctor to characterize bodily discomfort and mistakenly describes gas pain as nausea, or is asked by a cook to identify the flavors of a complex broth and mistakes the taste of lime for that of vinegar. Second, Evans’s case lacks even the faintest whiff of irrationality. The subject he describes is guilty of nothing more exotic than a failure to keep track when counting. Third, and most important, the example demonstrates that cognitive or conceptual acts like counting often intervene between the enjoyment of a visual experience and the formation of an introspective belief or judgment affirming the existence and character of that experience. Admittedly, it would be odd to posit a (second-order) seeming experience that intervenes between one’s (first-order) visual experience and the introspective belief or judgment that one is enjoying that experience. And since visual experiences do intervene between the extra-mental objects we see and our knowledge of the existence of those objects, introspective knowledge of experience and perceptual knowledge do

Introspection, Explanation, and Perceptual Experience

363

differ in this important respect. But we have seen that introspection of experience resembles perception in an equally important way. Both acts require categorization, counting, and other conceptual preliminaries. And when these preliminary acts of conceptualization are unsuccessful—as when Evans’s subject miscounts—mistaken introspective judgments can result. In consequence, there are introspective errors similar in kind to errors of perceptual judgment.10 At the very least, then, the relation between our introspective judgments and our experiences can be seen to be different in kind from that which obtains between our first-order beliefs and thoughts and our (second-order) introspective knowledge of them. Descartes’s belief that he exists is self-verifying—the judgment is made true by its execution. And the same is probably true of Tyler Burge’s (1988) class of cogito-like judgments: judgments of the form “I am thinking p.”11 But our introspective knowledge of our experiences is not like this. Indeed, the difference remains in view when we turn from our introspective knowledge of those thoughts we merely entertain to our knowledge of our own beliefs. Evans’s subject does not figure out whether he believes that there are eleven points of light before him by turning his attention to his belief and counting. Instead, his knowledge of his belief is drawn directly from it.12 In contrast, Evans’s subject does turn his attention to his experience when determining how many lights he seems to see. He first becomes c-introspectively aware of that experience. And only then, after (incorrectly) counting the lights he seems to see, does he form a (mistaken) judgment as to the number of lights he is experiencing. C-introspection is not only possible; it is a typical preliminary to b-introspective judgment. There are, admittedly, other possible descriptions of the case that do not invoke c-introspective awareness of experience. One might insist, for instance, that Evans’s subject can only exercise perceptual attention, and then either replace or augment its more natural product—the judgment that there are x number of lights in front 10 These reflections rebut Martin’s (2006, 407) claim that one must either: (a) accept (secondorder) introspective appearances to which our (first-order) experiences need not conform; (b) posit an infallible mechanism of introspection; or (c) deny, as Martin does, that we have “a distinct perspective on our inner lives from that we take when experiencing the [extra-mental] world” (2006, 407). I have argued that we have a “distinct perspective” on our “inner lives.” It is constituted by our (second-order) introspective beliefs about—and, perhaps, our c-introspective awareness of—various first-order states of mind (experiences of spheres and the like). On this view, introspection is fallible, but not because our (first-order) experiences are not always as they introspectively seem. (Again, there are no seeming experiences intervening between our experiences and our knowledge of them.) Introspection is fallible because we can incorrectly conceptualize or categorize our experiences when formulating our introspective judgments. 11 See Zimmerman (2006b) for caveats. 12 I defend this view at length in Zimmerman (2006a). For criticism, see Silins (this volume).

364 Introspection and the Nature of Experience of him—with the (wholly conceptual) judgment that he is having an experience of x number of lights. According to this line of thought—which may indeed be Evans’s take on the scenario—advocates of the perceptual model of introspection mistake cases in which we more carefully focus our perceptual attention on external objects for cases in which we c-introspect upon the visual experiences that represent (or contain) those objects (cf. Byrne, this volume). But this reply is not plausible in the case of hallucination. If Evans’s subject is hallucinating ten lights, focuses his attention on what he takes to be the circle of lights before him, and then counts as best he can, he will of course think that he has focused his attention on the lights he can see. (He will think he is more carefully examining those points of light, not his experience of them.) But we will know, looking in, that his attempts to engage the mechanisms of perceptual attention were for naught. Since there were no lights there for him to see, he only succeeded in making himself more fully aware of features of his experience. And surely, it is introspection that makes one aware of these features of one’s experience, not vision. Our subject’s visual experience represented ten points of light arrayed in a circle before him. And, unbeknownst to him, when he tried to focus on the lights before him, he wound up dwelling upon those features of his experience that help make it a representation of ten lights in a circle. He c-introspected. Now, though I think these considerations make the existence of c-introspection a reasonable conjecture, I will not pretend that they force its acceptance. And this is clearly not the place to evaluate the wealth of arguments that have been presented both for and against the perceptual model.13 So let us just suppose, if only for the moment, that reflection on Evans’s example (or its hallucinatory variant) has convinced us that Shoemaker is mistaken, and that we can indeed get an introspective fix on our visual experiences without conceptualizing or describing them as events of experiencing or events of seeing. With this assumption in place, we can allow our initial subject Max “thing awareness” of his veridical visual experience of the sphere before him—a kind of awareness that cannot be equated with either his knowledge

13 I find Charles Siewert’s case against baldly perceptual models of introspection largely congenial (this volume). I am unsure, however, whether the kind of attention to experience he allows is really, as he contends, best described as a form of pure thought (as distinguished from experience). Surely, when we focus our visual attention on extra-mental objects, we more acutely see these things. We do not simply succeed in more carefully thinking about what we see. And if introspective attention resembles visual attention in this regard, when we c-introspect our visual experiences we do more than just carefully think about these experiences. And yet, because there are no introspective experiences of our visual experiences, c-introspection cannot be said to involve an inward sensing of experience. So c-introspection is like sense perception is some respects but not others.

Introspection, Explanation, and Perceptual Experience

365

that he is having an experience of a sphere or his knowledge that he is seeing a sphere. So when Max c-introspects in this manner, will his experience seem to him to involve an extra-mental constituent? To answer this question we must each do our best to c-introspect and report what we find. So I encourage you to turn your attention to your visual experience and to verify whether the claims I am about to make jibe with those to which you are inclined. Now I find it difficult to focus my attention on my current veridical visual experience itself; it is much easier to focus perceptual attention on the sphere I can now see. But it is not nearly as hard for me to focus my attention on the veridical visual experience I enjoy when my vision is blurred, doubled, or distorted in some other manner.14 But when, on these occasions, I manage to focus on my experience itself, I do so by shifting my attention away from the extra-mental objects I am seeing. Indeed, if I do not do this, how can I be sure that I am not just more acutely seeing those extra-mental objects that I can see, or more carefully dwelling on the fact that I am seeing them? What distinguishes such acts from c-introspecting the veridical visual experience itself? Surely, if there is such a thing as c-introspecting an experience of a sphere, there must be something that distinguishes this act from attentively focusing on a seen sphere, as attentively watching something is not itself an introspective act. (Only primates can introspect, but many “lower” species can look, watch, and examine what they see.) But it seems that the only feature that could distinguish the two acts from one another is a difference in the focus of attention. When I attentively focus on the sphere, I more carefully examine its shape, color, and the like. In contrast, when I c-introspect my visual experience of the sphere, I focus my attention on its blurriness and other clearly perspectival properties of the event. But if c-introspection essentially involves such an “inward” shift of one’s attention, the object of c-introspection will not present itself as an extra-mental object, nor will it seem to contain such an object as an essential constituent. Instead, I will have used c-introspection to focus my attention on precisely those features of my visual experience that do not reflect the properties of the sphere it enables me to see. And though it is an open question whether my current visual experience really could exist in the absence of any seen object, it does not seem to me that c-introspection provides any evidence that this experience—or an experience of the same type—could not be wholly hallucinatory.

14

Note that these are still instances of veridical experience. (They are neither hallucinations nor illusions.) I am still seeing the blurry object in view; as we representationalists would say, my experience still correctly represents the existence of this object despite mischaracterizing its color, shape, and exact location.

366

Introspection and the Nature of Experience

Surely, my vision’s blurriness is not an indication of the sphere’s true shape. Isn’t it then possible that none of the features on which I am focused have any extra-mental reality? As far as I can tell, c-introspection can do nothing to answer this question in the negative.15 In the end, then, I suspect that Martin rejects c-introspection altogether. (I will say more about this below.) For if Martin did countenance c-introspection, he would stand guilty of conflating it with attentive seeing. He would stand accused, that is, of running together the act of focusing introspective attention on experience with the act of focusing perceptual attention on its objects. Why else would he say that when you turn your attention to your visual experience, it is presented to you as essentially external-object-involving? At any rate, whatever explains the divergence between Martin’s introspective reports and my own, I am certain that at least my case does not answer to Martin’s description of how attention to experience is supposed to pan out. When I do my best to c-introspect my current visual experience, it does not seem to me to essentially involve the external object of which it is an experience.

4. Explaining Introspective Indiscernibility I have argued that a-introspection, b-introspection, and c-introspection all speak in favor of representationalism and against a naïvely realistic view of visual experience. Still, I do not want to hang too much on what might prove to be peculiarities in my own case.16 So I want to admit, if only for the sake of argument, that when we c-introspect on our veridical experiences we take them to essentially involve external objects in the way in which events of seeing essentially involve the extra-mental thing seen. And yet, even if the disjunctivist is granted this, the view is still open to dispute. For, as Martin admits, there will be cases in which a subject cannot tell via introspection alone whether she is hallucinating or instead enjoying a veridical visual experience. And when we consider cases in which a subject cannot tell via introspection whether she is, say, hallucinating a sphere or actually seeing one, it is

15 I here join Maja Spener in denying the “Exclusive Apparent Mind-Independence” thesis, according to which ordinary visual experience presents properties, objects, or phenomena as exclusively mind-independent (this volume). 16 My worries may be well founded. On the basis of a (rather limited) survey of those philosophers who publish their introspective reports, Hellie (2007, 266–269) argues that “phenomenological study” of our experiences provides support for something like naïve realism.

Introspection, Explanation, and Perceptual Experience

367

natural to suppose that there is something common to both the hallucinatory and veridical experience beyond the mere fact that they cannot be told apart by the subject in question, where this further commonality explains why the subject cannot tell these experiences apart.17 There are actually several forms such explanation might take. First, we must ask whether b-introspection is preceded by c-introspection in the case. Does the subject merely judge that her visual experience has not changed despite the move from veridical experience to hallucination? Or does she ground this judgment in some more primitive (nonconceptual, quasi-perceptual) act of c-introspection? Next, we must decide which features of the subject’s experience cause and rationalize her judgment that nothing has changed. Does she judge that her experience remains the same because her pre- and post-hallucinatory experiences are qualitatively identical? Or might distinct representational features play this role? Does she judge that her experience has not changed because what things are like for her when she is hallucinating is exactly the same as what things are like for her when she accurately perceives what is before her? Or does she judge that nothing has changed in her experience because her hallucinatory experience represents precisely what her veridical experience represents, where what an experience represents cannot be equated with what things are like for the subject enjoying that experience? Different answers to these questions yield different substantive theories of introspection. But what is striking about disjunctivism, is its incompatibility with any such theory. For the disjunctivist’s commitments prevent him from accepting any substantive explanation of introspective indiscernibility. We can make this criticism more precise by concisely stating the explanatory thesis that metaphysical disjunctivists must reject. Substantive Explanation of Introspective Indiscernibility (SEII): If a mature, reflective subject judges on the basis of careful introspection that her (relatively simple) visual experience before t is the same as her experience after t, where one of these experiences is hallucinatory and the other veridical, her experience before t must be similar to her experience after t along some dimension, where the similarity in question explains (i.e. causes and rationalizes) her judgment that nothing has changed. The disjunctivist must reject seii. For if there were a positive (representational or qualitative) characteristic common to both hallucinatory and veridical visual experience, we could use this characteristic to pick out the genus of which both kinds

17 This is a common criticism of disjunctivism ably mounted by Siegel (2008, 218–223); Hawthorne and Kovakovich (2006, 179); Byrne and Logue (2008, 89–90); and Tye (2009a, 560).

368

Introspection and the Nature of Experience

of experience are species (Martin 2006, 367). And this would impugn the disjunctivist’s claim that hallucination and veridical experience of a sphere are not members of a natural or nondisjunctive kind. They would instead be experiential representations of a sphere, or experiences of the qualitative kind: sphere. Thus, it seems that even if we grant the disjunctivist the phenomenological support he claims for himself, we can rationally reject the position on offer as explanatorily impoverished. Now Martin tries to reply to this critique by arguing that no explanation of the introspective indiscernibility of hallucinatory and veridical experience is necessary (2006, 393–396). We can, he thinks, fully characterize the agent’s “perspective” when she is hallucinating by saying that she cannot discriminate her situation from one of veridical visual experience (ibid.). So we need not say that she is then aware of some positive (representational or qualitative) features of her experience: features that would equally qualify her experience were she actually seeing something external to her mind. Of course, it is not uncommon to say that on both occasions it looks to the observer “as though” there is a sphere before her; and it is not uncommon to say that “what it is like” to enjoy a perfect hallucination of a sphere is no way differs from what it is like to really see one. But, in Martin’s view, what we are trying to convey here is just the fact that we cannot tell the two episodes apart.“It looks to the observer in both cases as though there is a sphere in front of her” and “What its like to see a sphere in no way differs from what it is like to vividly hallucinate one” do not state facts of similarity from which sameness in experiential kind can be inferred. Might this be true? The first thing to note is that the view is unattractive as stated. Suppose we focus on a case in which a subject, Sam, sees a sphere before him right up until the moment at which the sphere is removed and he begins to enjoy a perfect hallucination of it.18 When we ask Sam whether anything has changed (in either the sphere or his visual experience of it), he will honestly answer in the negative, as it looks to Sam as though nothing has changed. And those of us who reject Martin’s disjunctivism will explain Sam’s inability to discriminate the hallucination from the veridical perception in terms of what we at least take to be distinct facts about Sam’s experience. Despite the disappearance of the sphere, Sam’s experience of it has remained qualitatively the same, and his experience still represents to be the case precisely what it did before. But Martin must insist that these quite natural explanations are defective. According to the disjunctivist, our descriptions are

18

William Fish might reject the very possibility of Sam’s case as I have described it, as he insists the possibility of an unnoticeable shift from veridical perception to perfect hallucination cannot be safely assumed (2009, 87). But I cannot see why Fish demands defense of a possibility so easy to imagine and describe.

Introspection, Explanation, and Perceptual Experience

369

either false, or they fail to assert anything beyond the fact that Sam cannot tell his two different experiences apart. But why should we abandon our explanatory practices as Martin requests? Because veridical experience (purportedly) seems to us to involve objects when we c-introspect upon it? But why should the latter consideration weigh more heavily in our thinking than the former? Why not save seii by rejecting the (purported) introspective appearances on which Martin relies?

5. An Internal Tension in Martin’s Reply We have seen that neither a-introspection nor b-introspection provides support for naïve realism, and that Martin must therefore accept the existence of c-introspection if he is to provide the least shred of evidence for naïve realism and the disjunctivist conception of hallucination and veridical experience he bases upon it. Moreover, as we have also seen, Martin maintains that no explanation of the introspective indiscernibility of veridical visual experience and perfect hallucination should be provided. Since all such explanations depict hallucination and veridical experience as species of an explanatorily potent common genus, the disjunctivist must reject them all. But there is a tension here. For in arguing against seii—and so against substantive explanations of introspective indiscernibility—Martin would seem to reject the very existence of substantive introspective knowledge. And in so doing, he robs himself of the only potential source of support for his belief in naïve realism. Martin’s argument for why we need not explain why subjects cannot tell their perfect hallucinations apart from their veridical experiences hangs on the idea that introspection is importantly different from perception. Although the sphere that Sam can see is, for example, distinct from Sam’s initial visual experience of it—as the sphere is an extra-mental object—Martin claims that Sam’s visual experience of the sphere is not distinct from his introspective awareness of that experience. On Martin’s lights, we do not need to explain why Sam judges that nothing has changed when his experience shifts from veridical perception to perfect hallucination because the fact that his experience remains qualitatively or phenomenologically the same is nothing beyond his introspective sense that nothing has changed in it. However things seem from the subject’s perspective with respect to her phenomenal consciousness is how phenomenal consciousness must be . . . . The subject’s perspective on her own sense experience constitutes sense experience being that way for her . . . . If she really is in a situation in which from her perspective it is as if she is having an experience as of a white picket fence, then that constitutes her being in the situation of having an experience as of a white picket fence. (2006, 392–393)

370

Introspection and the Nature of Experience

Now I think it is natural to read these statements as claiming a kind of infallibility for our introspective judgments. And if this reading is correct, Martin must deny the very cogency of Evans’s example of mistaken introspection judgment discussed above. (One cannot be led by introspection to the false conclusion that one seems to see eleven points of light if false introspective judgments are simply impossible.) Moreover, whatever one thinks about the plausibility of positing infallible powers to introspect, the dialectical power of this maneuver cannot be denied. For if our judgments about our visual experiences really were infallible, Martin would be right to reject the folk psychologist’s attempt to explain Sam’s judgment that nothing has changed in his experience by appealing to qualitative and representational constancies in that experience. Why? Well, Martin is right to reject as incoherent infallible mechanisms of introspective discrimination (2006, §7). And if we cannot be credited with infallible powers of introspective discernment, our (purportedly) infallible introspective judgments about what we are experiencing must be assigned some other source. (Perhaps those theorists driven to accept the invariable truth of our introspective judgments would feel pressed to invoke conceptual considerations. Like Descartes’s judgment that he exists, and the clever man’s belief that he believes at least one thing, our introspective judgments about what we are experiencing would have to be construed as making themselves true.) So Martin is right about at least this: if Sam’s introspective judgments about his experience were truly infallible, his judgment that nothing changes in his experience when his veridical experience gives way to hallucination really could not be grounded in an appreciation of some respect in which his veridical experience and his hallucination are independently similar to one another. There is a problem, however, with endorsing this line of reasoning and following Martin along a train of thought that would seem to lead him from the infallibility of our introspective judgments to a denial of seii. And this is just the patent implausibility of the claim that introspection is infallible in the sense we have described. Surely, Evans’s subject is just wrong that he seems to see eleven points of light. Surely, our introspective judgments are sometimes mistaken. Of course, as we have already noted, there are indeed important differences between perception and introspection. But fallibility is something the two sources of knowledge would seem to have in common. It is perhaps because of these worries that Martin swears off the natural reading of the text that we have been considering. The disavowal comes when Martin confronts the Wittgensteinian objection that infallible knowledge is not really knowledge at all. If my opinion on some matter cannot be wrong, the Wittgensteinian protests, its truth is not much of an achievement; and, the critic continues, knowledge must be won, not awarded by default. So if our introspective judgments really cannot be wrong, they cannot really constitute knowledge.

Introspection, Explanation, and Perceptual Experience

371

In reply to this sort of objection, Martin says that “a subject’s perspective on her own mind” is not meant to denote her beliefs and judgments about how things seem to her. The constitutive connection is between the subject’s perspective on his or her own mind, how it seems to be, and how his or her mind then is. This need not be identified with the judgments he or she actually makes. (2006, 392 n. 42) Presumably Martin is not just saying that the subject’s perspective on her experience need not be identified with her introspective judgments. Instead, the reader is being told that she should not so identify the introspective perspective when interpreting Martin’s claim that this perspective is infallible. For if the reader did so interpret the phrase, she would conclude that Martin thinks our introspective beliefs about what we are experiencing are invariably true. And Martin is assuring the reader that he does not endorse this patently false doctrine. But if the introspective perspective is not constituted by introspective judgments, and it is not constituted by introspective experiences of our experiences— because, as Martin rightly asserts, there are no introspective appearances—what exactly is the introspective perspective supposed to be? I do not know whether Martin has arrived at a firm answer to this question. The remarks he does make suggest that he endorses something like Shoemaker’s (1996) view according to which a first-order mental state will “double” as an introspective awareness of itself so long as it is embedded in a mind with the requisite concepts and powers of reasoning (cf. Shoemaker, this volume). And if this is indeed Martin’s view, he would have Sam’s visual experience of the sphere itself constituting Sam’s introspective perspective on that experience (given its embedding in Sam’s sufficiently sophisticated and well-functioning mind). But even if we suppose that Sam’s initial veridical visual experience of the sphere somehow “doubles” as his awareness of such, Martin must tell some story about how this minimal form of introspective awareness might somehow morph into the full-blown judgment that the sphere is an essential constituent of that experience. For recall that Martin does not support his assertion of naïve realism by citing experiments, observations, and inferences. Instead, the view is supposed to be motivated by introspection or “reflection” alone. But how exactly is a belief in naïve realism supposed to emerge from introspection when a subject’s introspective awareness is nothing more than the (suitably embedded) experience of which she is aware? We can bring the problem this question poses more clearly into focus by shifting our discussion from our imagined subject Sam to Martin, the theorist himself. To do this, let us suppose that Martin is enjoying a veridical visual experience of a

372

Introspection and the Nature of Experience

lavender bush in bloom before him. How does Martin transition from his minimal introspective awareness of this visual experience—an awareness that is supposed to consist in his suitably embedded visual experience of the blooming bush—to his confident judgment that this experience involves the lavender bush as one of its essential constituents? Surely a veridical visual experience of a bush is not itself a belief in naïve realism. But Martin says nothing about how a subject might transition from the one to the other. In sum, Martin argues against seii by claiming that introspective awareness of a visual experience of a mind-independent object is nothing more than that very experience embedded in a suitably sophisticated mind. And it is clear why he argues for this extraordinarily minimalistic conception of introspection. For if introspective awareness of experience were more substantial than this, it would be hard to see how it could be infallible. And if introspection were fallible, folk psychologists would be right to explain its successes in cases like Sam’s by appealing to those positive representational features his hallucination and veridical experience of the sphere have in common; features that justify and render true his belief that his experience has not changed; features that are not present when, as sometimes happens, a subject mistakenly judges that nothing has changed in her experience. But Martin cannot consistently characterize introspection in this minimalistic way. For naïve realism is supposed (by Martin) to derive its support from introspection alone. And it is clear that Martin must characterize introspection in a more substantial light if he is to convince us that it really does provide evidence for a naïvely realistic conception of veridical visual experience. Indeed, it is hard to see how Martin’s varying descriptions of introspection—invoked, as they are, in response to two very different challenges to his view— might be unified in such a manner as to leave his case for disjunctivism intact.

6. Disjunctivism and Psychological Explanation Visual experience does not just interact with our beliefs about it. Crucially, psychologists of both the folk and academic varieties invoke experience to explain beliefs about external objects—beliefs that are not themselves preceded by introspection upon experience. And it can be seen that many of these beliefs respond to hallucination and veridical perception in exactly the same way. Thus, reflection on cases reveals that we often treat a subject’s hallucinatory and veridical experience as equivalent in kind when explaining her efforts to render correct perceptual judgments. And Martin has not successfully argued for the impropriety of this practice. To illustrate the kinds of folk psychological explanation I have in mind, let us suppose that Sam initially sees not one large sphere before him, but ten small ones,

Introspection, Explanation, and Perceptual Experience

373

and that at some point he begins to hallucinate the spheres without being able to tell that a change has taken place. There are two variations to consider. In one case, Sam is asked how many spheres are before him prior to the onset of his hallucination, and after correctly counting them up, he concludes there are ten. Now a third party asked to explain what has occurred will no doubt point out that both Sam’s eyesight and his ability to count were involved. For in the absence of either he would not have correctly ascertained the number of spheres. But to arrive at something more determinate than this our folk psychologist must run afoul of one of the two differing philosophical theories on hand, as each offers a different account of just how looking and counting operated to affect Sam’s response. On Martin’s account Sam has a veridical visual experience of the spheres and correctly counts the spheres that partly constitute that experience. According to the representationalist, Sam has a veridical visual experience of the spheres, and he correctly counts the spheres his experience represents. But now suppose that Sam is not asked for the number of spheres until the hallucination has already begun, and that he again does his best to count and arrives at the conclusion that there are ten spheres before him. Now in this case, the two theories will diverge. Despite the change in distal stimulation, the representationalist’s explanation of how Sam arrives at his answer will not change. Again Sam correctly counts the spheres his experience represents. But the disjunctivist must now provide a radically different explanation. Sam cannot now count the ten spheres that partially constitute his experience, as there are now no spheres to be counted. So the disjunctivist must in this case appeal to a representation of the spheres, or, at the very least, invoke something other than Sam’s counting spheres. And yet, though the disjunctivist might appeal to an experiential representation to explain Sam’s numerical judgment in the hallucinatory case, he cannot give the same explanation of the conclusion Sam reaches when veridically perceiving. For to do so would be to treat hallucination and veridical visual experience as instances of a common kind. It would turn out that both of Sam’s experiences are representations of ten spheres: experiences that represent ten spheres in such a way as to inform and render intelligible Sam’s efforts to count what is before him.19 So we must ask: When our ordinary subject explains the two cases under discussion will she embrace the disjunctivist’s conclusions? Again, I have only my own case to go on, but I predict that you will join me in a negative answer. When explaining Sam’s answer in the veridical case our folk psychologist will feel pressed

19

Note that an operation of conceptualization can be shaped and informed by (or based on) an experience without its being about (or of) that experience. (Sam’s counting the spheres can be based on his experience of them without his therein counting his experience or elements of it.) See Evans (1982, 227) for discussion.

374

Introspection and the Nature of Experience

to cite the very same relation between looking and counting that obtains when Sam hallucinates. Again, Martin’s disjunctivism runs into implausibility. We are left looking for a good reason to revise the commonsense psychological practice that we have identified. To be clear, I am not denying that various facts about Sam’s thinking, reasoning, and acting are best explained by adverting to the (necessarily spheres-involving) fact that he sees ten spheres. His successfully sorting the spheres into two groups of five is surely a fact of this kind.20 But we have limited our view to cases in which “visual experience” is used to explain someone’s judgments or behavior, so explanations in terms of “seeing spheres” or “seeing that there are spheres” are beside the point. And it is instructive to note that the fact that Sam might correctly sort the spheres in the veridical case could not have a corollary in the hallucinatory case, for when he is hallucinating, Sam cannot find any spheres to sort. Indeed, it is because Sam counts those spheres he seems to see in precisely the same way in both of our initial scenarios that we feel compelled to provide a uniform explanation of what has happened. The commonality in effect calls out for a commonality in cause. We posit a single experiential origin to explain a remarkable similarity in cognitive result.

7. Explaining the Nature of Experience I have argued that Martin’s disjunctivist approach is incompatible with folk psychological practice. But critics have also argued against Martin’s attempt to explain the nature of experience in terms of veridical experience and events or states indiscriminable from it. My aim in this section is to explain this criticism of Martin’s account, to describe Martin’s response to it, and to explain why Martin’s response cannot be applied to the case we have already presented against the disjunctivist view. Consider, as an instance of the kind of criticism I have described, Susanna Siegel’s (2008) argument against Martin’s equating visually experiencing a sausage with having an experience introspectively indiscriminable from a veridical visual experience of a sausage, an equation that is supposed (by Martin) to capture the idea that both hallucinations of sausages and veridical visual experiences of sausages are indeed visual 20

See Williamson (2000, 60–64) for cases in which we quite naturally cite an agent’s factive (or external-object-entailing) states of mind to explain her actions. Cf. Pettit (1986) and Child (1994, 204–216). Martin claims, without argument, that explanations of judgment and behavior phrased in terms of “perceptual states” (e.g., states of seeing something) are in “competition” with explanations in terms of “sensory experience as something common across the [veridical and hallucinatory] cases” (2006, 371–372). But he never considers the view, defended in the text above, that different explananda (counting v. sorting) demand differing explanans (experiencing v. seeing).

Introspection, Explanation, and Perceptual Experience

375

experiences of sausages even though they fail to constitute a common, important (nondisjunctive) kind. Siegel argues that when we turn to a classification of the hallucinations of nonhuman animals, the equation we have described will come to grief. She argues with a dilemma, the first horn of which has Martin characterizing introspective indiscriminability in a positive manner, so that two experiences are introspectively indiscriminable if the subject enjoying them is (positively) disposed to judge them the same (2004, 96). Since dogs cannot introspect, they are not positively disposed to judge hallucinations of sausage to be the same as veridical experiences of such. And yet, despite this failing, the dog’s hallucination of a sausage supper is surely an experience of sausage. Thus, Martin must reject the positive characterization of introspective indiscriminability on offer as too restrictive. But this just brings Martin to rest on the dilemma’s second horn. For suppose that he instead adopts a negative characterization of introspective indiscriminability by allowing that two experiences are indiscriminable so long as the subject enjoying them is not disposed to judge them different. The dog is not disposed to judge that his hallucination of sausage is different from his veridical experience of such because he cannot make—and so is not disposed to make—introspective judgments of any kind. So the negative characterization of introspective indiscriminability correctly classifies both events as experiences of sausage. But then the dog also lacks a disposition to judge that his veridical experiences of his ball and his bowl of water differ from his veridical experiences of sausage. In consequence, the negative characterization of introspective indiscriminability incorrectly classifies as experiences of sausage the dog’s veridical visual experience of his ball and water bowl (2008, 97). In sum, the positive characterization that Siegel considers is too strong, and the negative characterization too weak. Martin (2006, 379–383) replies to this objection by invoking an “impersonal” notion of introspective indiscriminability—a strategy explicated and endorsed by Sturgeon (2006, 2008). To say that the dog’s perfect hallucinations of sausage are introspectively indiscriminable from veridical visual experiences of such is not to say that the dog would judge them the same. It is instead to say—speaking impersonally now—that these events just cannot be told apart after even the most careful efforts to introspect. Introspecting a difference between them is, to use Martin’s phrase, “impossible simpliciter,” and not just impossible for this or that subject or species (2006, 381). Correlatively, to say that the dog’s veridical experiences of sausage are introspectively discriminable from his veridical experiences of balls and bowls of water is not to say that he can tell these experiences apart. It is just to say—without relativizing to the dog or any other subject—that they can be told apart on the basis of introspection. When we are trying to give an account of what sausage experience is, we should feel free to idealize in this manner, Sturgeon suggests, just as we should

376

Introspection and the Nature of Experience

feel to idealize when giving an account of doxastic rationality in terms of logical omniscience and full probabilistic coherence (2006, 196–197). Let us suppose that Martin and Sturgeon are right about this. And let us suppose, in consequence, that being a visual experience of ten spheres in a circle can be equated with—or shown to be necessarily coextensive with—being introspectively indiscriminable (in the abstract) from a veridical visual experience of such. Still, what does this have to do with our efforts to explain why Sam, when hallucinating the spheres before him, came to judge that there were ten spheres before him? Perceptual concepts and perceptual judgments develop before introspective concepts and judgments. So Sam might hallucinate his spheres and judge there to be ten spheres before him without judging that he is having an experience of ten spheres and without actually trying and failing to discriminate his experience from a veridical experience of ten spheres. So what does the fact of the hallucination’s impersonal indiscriminability from a veridical visual experience of ten spheres have to do with that hallucination’s motivating and rendering intelligible the judgment Sam actually makes in the case on hand? When doxastic rationality is supposed to consist in logical omniscience and full coherence, it is not at the same time supposed to account for our real-world, realtime judgments and decisions. We invoke these idealizations to characterize the working of an ideal mind and the norms to which many ordinary thinkers aspire, but we must instead cite the nonidealized properties of the reasoning processes of (nonideal) people to provide adequate explanations of their real-world, real-time beliefs and actions. Indeed, modern economic theory is learning precisely this lesson from behavioral economists. You can use idealized conceptions of rationality when constructing a model of ideal economic activity. But if you can replace these idealizations with a more realistic description of the processes that actually generate our beliefs and choices you will invariably arrive at better explanations and predictions of our actions and the economic system they embody. So which event drove Sam to judge there to be ten spheres before him? His nonideal, all-too-real hallucination of ten spheres. And which features of this hallucination best explain why it had this effect on his judgment? Its nonideal, alltoo-real representational and qualitative features. Sam’s perfect hallucination is indeed impersonally introspectively indiscriminable from a veridical visual experience of ten spheres. But when he does not in fact introspect upon that experience, and he does not in fact mistake it for a veridical experience of ten spheres, the observation that Sam’s hallucination is (in some impersonal sense) indiscriminable from a veridical perception does not enter into (much less exhaust) the best folk psychological explanation of the judgments to which this experience actually gives rise.

Introspection, Explanation, and Perceptual Experience

377

8. Scientific Explanation What then of scientific psychology? I suspect that investigation of Sam’s nervous system would find exactly the same interactions between the neural correlates (or realizations) of counting and visual processing occurring both before and after Sam begins to hallucinate. That is, though it is an empirical conjecture—a conjecture that could only be conclusively verified by impracticably (and immorally) producing a series of perfect hallucinations through surreptitious cortical and (perhaps) retinal stimulation—I suspect that there is no neurological difference between veridical perception and perfect hallucination, nor any difference in the impacts these events have on the neural realization of counting at a distance and the judgments of number in which such counting issues. And since the neural realizations of perfect hallucination and veridical visual experience will turn out to be identical, and the two will turn out to have identical impacts on many of the forms of cognition to which vision is essential, neuroscientists will be tempted to infer that the realizers of hallucination and veridical visual experience are species of a single neurological genus. Won’t they then infer that this neurological genus is the correlate or realizer of visual experience: the folk psychological genus of which veridical visual experience and hallucinatory experience are both species? I suspect that they will, and that Martin’s claims about the intuitive nature of naïve realism will not move them from this stance. Only an explanatory or predictive advantage could tempt a neuroscientist to alter her interpretation of the results I think likely. And Martin’s theory of experience offers neither.21 Of course, Martin might reject my predictions as unlikely or overly speculative (as would Fish 2009, ch. 5). Or he might just ignore them altogether. But even if he focuses exclusively on folk explanation, the case against disjunctivism will not go away. For, as we have seen, we folk do introduce a common experiential kind to explain commonalities in the perceptual judgments of those who are seeing and those who are merely hallucinating. (And we use “seeing” and “not seeing,” not

21 Based on a careful examination of contemporary vision science, Tyler Burge (2005, 22) argues for a “Proximity Principle” on which, holding fixed background psychology, the same proximal stimulation of the visual system will yield tokens of the same perceptual state type regardless of differences in distal cause. (Burge’s phrase “proximal stimulation” is meant to include light striking the retina, though because this light need not have been reflected off the surface of an object, Burge countenances analogs of perfect hallucination: illusory perceptions—entertained in the absence of external objects—that have the same proximal origins as—and so are type identical to—the veridical experiences from which they are introspectively indiscernible.) Burge is surely right about the contemporary science, so the predictions I have made in the text are conservative.

378

Introspection and the Nature of Experience

“veridically experiencing” and “hallucinating” to explain any differences in the judgments and actions of such subjects.) Neither folk nor scientific psychology recommends Martin’s disjunctivist view.

9. Conclusion I have argued at length that Martin is mistaken when he says that our reflective conception of visual experience is only accurate if naïve realism is true. Introspection does not support but actually undermines naïve realism. This is especially true of conceptual forms of introspection. When a subject reflects on the fact that he is seeing something, he thinks of what is happening in a naïvely realistic light, but when he reflects on the distinct fact that he is having a veridical visual experience of something, he thinks of what is happening in representational terms. Martin thus needs to invoke nonconceptual introspective awareness of experience (c-introspection) to motivate the naïvely realistic picture of veridical visual experience on which his disjunctivism is based. But it is hard to see how Martin can countenance c-introspection while at the same time rejecting substantive explanations of the introspective indiscernibility of veridical visual experience and hallucination. Moreover, even if Martin were right that introspection supports naïve realism, our explanatory practices would provide us with a strong reason to reject the disjunctivist conception of visual experience to which naïve realism gives rise. If common thought is any guide, our hallucinatory and veridical experiences share a positive representational and qualitative character.

References Aristotle. [384–322 b.c.] 1984. “Problems.” Pp. 1319–1527 in The Complete Works of Aristotle,Vol. 2, edited by J. Barnes. Princeton: Princeton University Press. Burge, Tyler. 1988. “Individualism and Self-Knowledge.” Journal of Philosophy 85(11): 649–663. ———. 2005. “Disjunctivism and Perceptual Psychology.” Philosophical Topics 33(1): 1–78. Byrne, Alex. 2001. “Intentionalism Defended.” Philosophical Review 110(2): 199–234. Byrne, Alex, and Heather Logue. 2008. “Either/Or.” Pp. 57–94 in Disjunctivism: Perception, Action and Knowledge, edited by A. Haddock and F. Macpherson. Oxford: Oxford University Press. Chalmers, David, and Andy Clark. 1998. “The Extended Mind.” Analysis 58: 10–23. Child, William. 1994. Causality, Interpretation, and the Mind. Oxford: Clarendon.

Introspection, Explanation, and Perceptual Experience

379

Evans, Gareth. 1982. The Varieties of Reference, edited by J. McDowell. Oxford: Clarendon. Fish, William. 2009. Perception, Hallucination, and Illusion. New York: Oxford University Press. Foster, John. 2000. The Nature of Perception. Oxford: Clarendon. Hawthorne, John, and Karson Kovakovich. 2006. “Disjunctivism.” Proceedings of the Aristotelian Society, Supplementary Volume 80: 145–183. Hellie, Benj. 2007. “Factive Phenomenal Characters.” Philosophical Perspectives: Philosophy of Mind 21: 259–306. Hinton, J. M. 1967. “Visual Experiences.” Mind 76: 217–227. Hume, David. [1739–40] 2000. A Treatise of Human Nature, edited by D. F. Norton and M. J. Norton. Oxford: Oxford University Press. Hyman, John. 1999. “How Knowledge Works.” Philosophical Quarterly 49: 433–451. ———. 2001. “Knowledge and Self-Knowledge.” Pp. 171–193 in Wittgenstein and Contemporary Philosophy of Mind, edited by S. Schroeder. Basingstoke: Palgrave. Jackson, Frank. 1977. Perception: A Representative Theory. Cambridge: Cambridge University Press. Martin, M. G. F. 2004. “The Limits of Self-Awareness.” Philosophical Studies 120: 37–89. Reprinted in Disjunctivism: Contemporary Readings, edited by A. Byrne and H. Logue (Cambridge, Mass.: MIT Press, 2009), 271–317. ———. 2006. “On Being Alienated.” Pp. 354–410 in Perceptual Experience, edited by T. S. Gendler and J. Hawthorne. Oxford: Clarendon. Pettit, Philip. 1986. “Broad Minded Explanation and Psychology.” Pp. 17–58 in Subject, Thought, and Context, edited by P. Pettit and J. McDowell. Oxford: Clarendon. Putnam, Hilary. 1975. “The Meaning of ‘Meaning’.” Pp. 131–193 in Minnesota Studies in Philosophy of Science, Vol.VII: Language, Mind and Knowledge, edited by K. Gunderson. Minneapolis: University of Minnesota Press. Reprinted in Putnam’s Mind, Language, and Reality: Philosophical Papers,Vol. 2 (Cambridge: Cambridge University Press, 1975), 215–271. Robinson, Howard. 1994. Perception. London: Routledge. Shoemaker, Sydney. 1996. The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press. Siegel, Susanna. 2008. “The Epistemic Conception of Hallucination.” Pp. 205–224 in Disjunctivism: Perception, Action, and Knowledge, edited by A. Haddock and F. Macpherson. Oxford: Oxford University Press. Sosa, Ernest. 2007. A Virtue Epistemology:Apt Belief and Reflective Knowledge,Vol. 1. Oxford: Oxford University Press. Sturgeon, Scott. 2006. “Reflective Disjunctivism.” Proceedings of the Aristotelian Society, Supplementary Volume 80: 185–216. ———. 2008. “Disjunctivism about Visual Experience.” Pp. 112–143 in Disjunctivism: Perception, Action, and Knowledge, edited by A. Haddock and F. Macpherson. Oxford: Oxford University Press.

380

Introspection and the Nature of Experience

Tye, Michael. 2000. Consciousness, Color, and Content. Cambridge, Mass.: MIT Press. ———. 2009a. “The Admissible Contents of Visual Perception.” Philosophical Quarterly 59(236): 541–562. ———. 2009b. Consciousness Revisited. Cambridge, Mass.: MIT Press. Williamson, Timothy. 2000. Knowledge and Its Limits. Oxford: Oxford University Press. Zimmerman, Aaron. 2006a. “Basic Self-Knowledge: Answering Peacocke’s Criticisms of Constitutivism.” Philosophical Studies 128: 337–379. ———. 2006b. “Self-Verification and the Content of Thought.” Synthese 149: 59–75. ———. 2008. “Self-Knowledge: Rationalism v. Empiricism.” Philosophy Compass 3(2): 325–352.

14 Mind-Independence and Visual Phenomenology Maja Spener

1. Introduction Many philosophers maintain that visual experience presents things as part of the objective world. Their claim is about the phenomenal character of visual experience—about what it is visually like for one to undergo an experience. In their view, an apt description of such phenomenal character must register the fact that in having experience, one appears to be visually aware of ordinary public objects and properties in the physical environment rather than, say, of mental objects inside one’s head. In this vein, for instance, P. F. Strawson says that mature sensible experience (in general) presents itself as . . . an immediate consciousness of the existence of things outside us. . . . [T]he human commitment to a conceptual scheme of a realist character is not properly described, even in a stretched sense of the words, as a theoretical commitment. It is, rather, something given with the given. (1979, 47) The thought that visual experience presents things as part of the objective world plays an important role in debates about the nature of conscious visual experience.1 One dividing line in that debate separates subjectivist and world-individuating 1

From now on, I will use ‘experience’ to mean visual experience.

382

Introspection and the Nature of Experience

views about experience. The former embrace the idea that an account of the nature of experience must (although perhaps not exclusively) appeal to subjective phenomena of some sort. These subjective phenomena could be properties of experience, ranging from immaterial qualia to neurological properties of some kind, or they could be special sensory objects such as sense data. The key point is that the relevant objects and properties are not individuated with respect to the external perceptual environment of the experiencing subject. By contrast, worldindividuating views—for instance, standard versions of representationalism and disjunctivism—aim to account for the nature of experience only in terms of resources that are individuated by reference to the external perceptual environment of the experiencing subject. Proponents of world-individuating views typically hold the thesis that experience is transparent. The transparency thesis is a thesis about what is introspectible as part of the phenomenal character of one’s experience. There are different versions of the thesis, but I concentrate on what I take to be the most common one.2 It says that all one is aware of when introspecting the phenomenal character of one’s experience is what this experience visually presents or is as of. The way proponents sometimes put it is that one is introspectively aware of merely what experience visually presents and not of the experience itself. The transparency thesis also includes a view of what experience is as of: ordinary objects and their properties in one’s external environment. That is, when introspecting the phenomenal character of experience, all one is aware of are the ordinary worldly objects putatively seen in virtue of having the experience. It is important to keep in focus that the transparency thesis concerns the phenomenal character of experience, that it is about visual appearances or seemings. The claim is that introspection reveals that in experience it is as if there are ordinary worldly objects before one, not that there really are such objects before one. This claim about what experience is as of is just the thought introduced in the beginning of this chapter. In the context of the transparency thesis, though, it is considered an important bit of introspective evidence. As such, it is used by proponents of world-individuating views against their rivals in the following two ways. On the one hand, proponents of world-individuating views point out that one’s introspective evidence tells against postulating subjective properties or objects. So, for instance, Gilbert Harman says

2

The following are prominent examples of philosophers putting forward various versions of the transparency thesis: Harman 1990; Tye 1995, 2003; Dretske 1995; Lycan 1995; Byrne 2001; Strawson 1979; Martin 2002; and Thau 2002. Criticism and discussion of the transparency thesis: Kind 2003; Martin 2002; Siewert 2004; and Stoljar 2004.

Mind-Independence and Visual Phenomenology

383

Look at a tree and try to turn your attention to the intrinsic features of your visual experience. I predict you will find that the only features there to turn your attention to will be features of the presented tree . . . (1990, 39) If this is right, then not only is an account of experience that endorses qualia or sense data unmotivated, it is phenomenally inadequate as well. Such an account would distort what it is visually like to undergo experience. On the other hand, proponents of world-individuating views typically maintain that introspective evidence provides a positive reason to endorse a worldindividuating view, such as representationalism or disjunctivism. They say theories of this kind account nicely for the introspective evidence and are phenomenally adequate. For example, one notable defender of disjunctivism insists that the central naïve realist component of disjunctivism is put forward on the grounds that it gives an accurate description of how the subject’s situation strikes her when consciously perceiving. Focusing on the tower, I can note its distinctive shape and colouring; turning my attention inward, and reflecting on the character of my looking at the tower, I can note that the tower does not disappear from the centre of my attention. The tower is not replaced by some surrogate, whose existence is merely internal to my mind, nor are its various apparent properties, its shape and colours, replaced by some merely subjective qualities. So my perceiving is not only a way of providing me with information about an external world, when my attention and interest is directed towards action and the world; in its very conscious and so subjective character, the experience seems literally to include the world. (Martin 1997, 84) Such use of introspective evidence is an instance of a general trend in philosophical theorizing about consciousness. Philosophers typically use introspection-based phenomenal adequacy conditions—claims about the phenomenal character of experience arrived at via introspection—to constrain their theory of the nature of experience. In particular, introspection-based claims about what it is visually like to undergo experience often function as fairly neutral starting points and independent arbiters with which to adjudicate debate about the metaphysics of experience. This introspective method gives introspective data concerning visual phenomenology authority over theory. The expectation is that all theories of experience have to answer to it. In line with this, much current work on visual experience treats introspection as a source of fairly pre-theoretic knowledge of phenomenal character.3 3 See, e.g., Anscombe 1965, 173; Block 1980, 278; Martin 2000, 195–196; Levine 2001, 134 and Tye 2000, 28. For a more detailed discussion and critique of the introspective method, see Spener (ms).

384

Introspection and the Nature of Experience

To fulfill this function of introspective evidence in the debate about visual experience, I shall say that introspection-based claims about phenomenal character must have ‘plain status’. An introspection-based claim is plain if it captures pre-theoretic data about visual phenomenology that provides an independent vantage point from which to begin theorizing and to check one’s theorizing. Moreover, given that the introspective method is meant to yield an account of visual experience in general and not just about one’s own case, the introspective data is such that one can use it to check another’s theorizing. In other words, introspective data about phenomenal character is applicable to human experience in general and one can expect it to be available to all who enjoy normal visual experience and are capable of first-person reflection upon it. In this chapter I argue that the introspection-based claim that experience presents the objective world does not have plain status.

2. Apparent Mind-Independence What do philosophers mean when they say, on the basis of introspecting the phenomenal character of their experience, that visual experience presents things as part of the objective world? What is it for ‘[visual] phenomenology to [involve] a uniquely vivid directedness upon the world’ (Sturgeon 2000, 9), or for it to have ‘objective purport’ (McDowell 1998, 435)? A natural way to spell this out is in terms of the idea that objects and properties presented in visual experience appear to be mind-independent, that they appear to be independent of the perceptual episode itself. As Tim Crane puts it all (or almost all) serious theories of perception agree that our perceptual experience seems as if it were an awareness of a mind-independent world. One’s awareness of the objects of a perceptual experience does not seem to be an awareness of things which depend on that experience for their existence. (2006, 2.1.1) The thought that visual experience presents things as part of the objective world is also sometimes expressed in other, related terms. For instance, it is claimed that visual experience is as of ordinary public objects, or as of physical objects, or as of material objects and properties (see, e.g. Lycan 1995 and Peacocke 1993). Arguably, though, at a minimum such characterizations are meant to invoke the idea of perception-independent existence. While there are likely further aspects relevant to the visual appearance of things as physical, material, etc., such further aspects are not relevant to my discussion. In this chapter I concentrate on views that are committed at least to this much: Apparent Mind-Independence: Ordinary visual experience presents objects and properties as mind-independent.

Mind-Independence and Visual Phenomenology

385

The sense of mind-independence at issue here involves the possibility of perception-independent existence of the objects presented in experience.4 Apparent Mind-Independence says that the things presented in ordinary visual experience appear to be independent of the experiential episode itself. It is crucial to keep in focus that Apparent Mind-Independence is a claim about visual appearance, about the way experience visually presents things as. It is not a claim about what these presented things in fact are. I take Apparent Mind-Independence to be the basic articulation of the thought introduced in the beginning, that experience visually presents things as part of the objective world. Accordingly, I will speak of Apparent Mind-Independence from now on, instead of using the latter, more cumbersome, formulation.

3. The Challenge Suppose you see a red apple and you introspect what it is like for you to have that experience. In cases like this, there is almost universal agreement among introspecting subjects that redness (or some red-related feature) is part of what it is visually like for one to undergo such an experience. But about some features there is considerable disagreement as to whether introspection reveals them to be part of visual phenomenal character. One might think that if these features are part of phenomenal character, introspective awareness of them does not seem to be quite so straightforward as it is in the color case. A notorious example of this is causation: Hume famously denies that we are aware in experience of causation itself, claiming instead that we perceive only conjunction (Hume 1978, Bk I.3.xiv). Other philosophers and psychologists argue that we do experience certain events as causal and hence that causal features are part of visual phenomenal character (see, e.g., Michotte 1963, Beebee 2003, Siegel 2008). Such disagreements pose a prima facie difficulty for the introspective method outlined in the introduction. On the one hand, the introspective method presupposes that as an introspector, one is authoritative about what introspection reveals experience to be visually like. On the other hand, the method also presupposes that 4

This still allows for some dependence on human cognition. Dispositional properties that make ineliminable reference to human experience may nonetheless be considered mind-independent in this sense, so long as a thing can have them in the absence of being perceived. Another way to spell out the sense of mind-independence involved focuses on the independence of some property of objects from perspectives taken on them in visual experience, rather than on the independent existence of the object of experience itself. In this vein Siegel (2006) holds that in visual experience objects appear to be perspective-independent.

386

Introspection and the Nature of Experience

one can generalize from one’s own case to the visual experience of others. When faced with introspective disagreement about visual phenomenal character, then, a user of the introspective method cannot flatly insist that such disagreement arises because her opponents are wrong about their own experience. Nor can she say that disagreement arises because her opponents have different kinds of experience. These replies would undermine the introspective method itself. This is not to say that the method must be abandoned as soon as there is any introspective disagreement about phenomenal character. But when there is disagreement about the correct introspective characterization of what it is visually like to undergo experience, proponents of each side must make a case in defense of their view. They must tell some story that makes it plausible to hold that visual phenomenal character is as they say it is in introspection. Importantly, the story they tell must be consistent with the presuppositions of the introspective method. Of course, the particular instance of introspective disagreement itself has to be plausible in the first place. We would not be tempted into a dispute about whether introspection reveals that color or shape features are part of visual phenomenal character, for instance. Theorists may—and in fact do—disagree about the precise way in which introspection reveals them to be presented. Nonetheless, color and shape features are generally taken to be paradigms of visual phenomenal features and it is not at all controversial to assume that they are mentioned in an apt introspection-based characterization of one’s experience. To a large degree, the uncontroversial nature of this assumption surely rests on the absence of genuine introspective disagreement concerning the presence of apparent color and shape properties in visual phenomenal character. Near-universal agreement about the deliverance of introspection in a given case does not rule out introspective error concerning the features in question because there could be collective introspective mistakes. Yet, introspective agreement of this kind does nothing to raise or sustain such worries either. Absent any other reasons to doubt the deliverance of introspection (in these instances or in more global terms), faith in introspective judgment about visual phenomenal character that garners widespread agreement seems wellmotivated (see Bayne and Spener 2010 for discussion). By contrast, genuine disagreement about the deliverance of introspection raises the question whether the features at issue really are part of phenomenal character.5 This question becomes more forceful still if there are additional good reasons to

5

The fact of disagreement about the deliverance of introspection raises many other questions as well, e.g. about the epistemic significance of disagreement (see Spener 2011) and about the reliability of introspection in particular cases and more generally (see Schwitzgebel 2008 and 2011; Bayne and Spener 2010; Spener ms).

Mind-Independence and Visual Phenomenology

387

worry about the introspectability of a given feature being part of the phenomenal character of visual experience. In this circumstance, proponents of the view that the feature is part of visual phenomenal character must be prepared to defend their introspection-based claim in some way. Mere introspective foot-stamping is not enough. There has in fact been ample disagreement about whether introspection reveals Apparent Mind-Independence. Some of the quotations above show this. Strawson’s ‘given with the given’, for instance, is part of his case against Ayer’s conception of the character of sensory experience, according to which we are visually presented with a mosaic of colors and shapes (see Ayer 1973). Strawson argues that a person asked to describe what it is visually like for her to have her experience would not “start talking about lights and colours, patches and patterns”. (Strawson 1979, 43). Crane, too, suggests that introspection favors the claim that the character of experience involves Apparent Mind-Independence over the view that visual appearance is “an awareness of things which depend on that experience for their existence.” As he says, it ‘seems as if ’ this is so—I take it, that he means that it seems that way on the basis of first-personal reflection on one’s experience. The positive assertion of Apparent Mind-Independence is often made by juxtaposing it with a negative claim about what phenomenal character is not. This brings out something that has been suppressed so far, namely that proponents of the idea that experience presents things as objective typically have in mind something stronger than Apparent Mind-Independence. Strictly speaking, Apparent MindIndependence is compatible with experience also presenting (some) things or features as mind-dependent, or as neutral with respect to ontological status. The juxtapositions make clear that this is meant to be ruled out: Exclusive Apparent Mind-Independence: Ordinary visual experience presents objects and properties as mind-independent and not as mind-dependent or as neutral in this respect. It is this claim—Exclusive Apparent Mind-Independence—which is usually the bone of contention. That makes sense, of course, given its role as introspective evidence in the overall debate between subjectivist and world-directed views. Some subjectivist views, notably qualia views, can—and in fact do—accept Apparent Mind-Independence. But they hold that in addition experience presents some features as mind-dependent (or, at least, not as mind-independent). My focus from now on is Exclusive Apparent Mind-Independence. The existing introspective disagreement alone raises worries about Exclusive Apparent Mind-Independence. But I think there are further good reasons to be skeptical about it. When Exclusive Apparent Mind-Independence is put forward as

388

Introspection and the Nature of Experience

an introspective datum, it is part and parcel of the transparency thesis. The transparency thesis, recall, is a thesis about what is introspectible when one reflects on what it is visually like to undergo one’s experience. According to it, one is aware of what is presented in experience and of nothing else. And, according to the transparency thesis involving Exclusive Apparent Mind-Independence, what is presented in experience are just those objects and properties putatively seen or observed when visually perceiving one’s environment. The deliverance of introspection must therefore be characterized only in terms of those objects and properties putatively seen or observed when undergoing a visual experience. Bluntly put, my worry is this: mind-independence does not seem to be among the properties observed when visually perceiving one’s environment. But if it is not, in what sense is Exclusive Apparent Mind-Independence introspectible? Let me explain this worry in more detail. Suppose a normally sighted adult, looking at a red cube, wonders about which properties she sees it to have. She might hold that she sees the thing’s cube shape and redness, or its shininess, or something like that. But one would not expect her to claim that she sees the thing’s mind-independence. It would strike one as odd to list mind-independence as an observed property along with redness and shininess. On the face of it, then, there is a question about accounting for the alleged introspectability of Exclusive Apparent Mind-Independence in conjunction with the transparency thesis. In reply, one might argue that the problem does not get off the ground. For, the objects and properties we observe when visually perceiving our environment and by reference to which we must characterize the phenomenal character of a given visual experience are just ordinary objects and properties in the world, i.e., tables, chairs, their colors and shapes, and so on. Once it is understood that we characterize phenomenal character by reference to such ordinary objects and their properties the question vanishes—there is no additional property that we need to appeal to which accounts for Exclusive Apparent Mind-Independence. So it is no surprise (the reply goes) that nobody is going to list mind-independence alongside properties like redness and shininess when describing what he or she sees. By describing what is visually presented to one in terms of tables and chairs, one has already described it in terms conveying the mind-independence of the apparent objects. But this reply misses the point. In the previous section I said that the thought that visual experience presents things as part of the objective world is put in various ways, among them that such experience presents things as ordinary objects and properties in the world. The common, basic element in these different ways of putting the thought is Apparent Mind-Independence. So, yes, we do talk in terms of perceiving tables and chairs and, true, a chair is a mind-independent thing and in

Mind-Independence and Visual Phenomenology

389

describing something as a chair, we have described it in terms of an ordinary mindindependent object. But how does that relate to visual appearance? In describing something as a chair, we have also described it in terms of certain causal and functional properties. A chair is something to sit on, to get rest after physical exertion, to push up to a table, etc. Does the transparency thesis say that all these properties are also part of the phenomenal character of the experience? There are undoubtedly diverging opinions about the matter, and the line will be drawn differently by different people (see, e.g., Siegel 2010). But unless one holds an extremely abundant view of what sort of properties can be presented in visual experience there is an implicit restriction in the transparency thesis. The description of phenomenal character by reference to objects and properties observed concerns only properties that are visually available properties. Thus, it is by no means clear that mindindependence is part of what it is visually like to have a given experience if one characterizes it by reference to tables and chairs.6 One reason for concerns about Exclusive Apparent Mind-Independence might be that mind- or perception-independence involves a relation between the object of perception and the perceiving subject (or the subject’s state of awareness, or some such). But the latter is not among the objects straightforwardly presented in normal visual perception among the ordinary objects of perception. When seeing a green cup on the table, one does not perceive oneself, qua perceiving subject, out there seeing a green cup.7 Another reason for concern might be that there is a modal aspect to Exclusive Apparent Mind-Independence, i.e. it concerns the possibility that the objects of perception exist unperceived. Yet, we take visual experience to concern actual situations, namely things with properties as they are right there in front of one. Visual experience does not seem to present mere possibility (see McGinn 1996 and Siegel 2008 for related discussion). Both of these cases make it hard to see how Exclusive Apparent Mind-Independence can be introspectible as part of phenomenal character if the introspectible character of experience concerns solely what is presented in experience. Philosophers who hold that Exclusive Apparent Mind-Independence is revealed by introspection need to explain how this can be so. The property of mind-independence 6

I am assuming that there is an upper bound on which (high-level) properties can be presented in visual experience. As will become clear, though, someone who holds an extremely abundant view of the sort alluded to above, is not going to be too worried by my challenge. However, they still must explain the persistent introspection-based disagreement about Exclusive Apparent Mind-Independence. 7 Indeed, the issue of presence of the self (qua perceiving subject) in the phenomenology of experience is itself an area of fierce disagreement, especially in the phenomenologist tradition (see, e.g., Zahavi 2005).

390

Introspection and the Nature of Experience

as such does not seem to be straightforwardly among the objects and properties observed or seen. Yet these philosophers say that experience is transparent in introspection, by which they mean that only those objects and properties putatively observed when having the experience are introspectible. The challenge for proponents of Exclusive Apparent Mind-Independence, then, is to tell a story about how mindindependence can be revealed by introspection, consistent with their commitment to the transparency thesis. In the next section, I outline a general strategy about visually apparent properties that, on the one hand, seems a promising route to meeting the challenge, and that, on the other hand, is implicit in the literature concerning Exclusive Apparent Mind-Independence.

4. The Extraction Proposal Suppose that during a trekking adventure in the Andes you come to a narrow footbridge made of planks and ropes leading across a deep gorge. You hesitate stepping on the bridge since it seems rather rickety. Your impression of the bridge’s ricketyness is based on your visual perception of the bridge. But does the bridge visually appear rickety; can you literally see it is rickety? Being rickety is a causaldispositional property and as such has a modal aspect to it: the bridge seems disposed to collapse under light strain. The question is how this could be a visually presented property, how it could be an observable property of ordinary objects. One might suggest that the bridge’s appearance of being rickety is based on nonmodal properties the bridge is visually presented to have, such as the thinness of its planks and the tattered texture of the ropes holding it together. In particular, one might claim, the appearance of being rickety arises from, or is manifested in, the joint visual appearance of those other nonmodal properties. In essence, the idea is that the bridge’s appearance of being rickety is extracted from other properties straightforwardly visually presented in experience. Generalizing this idea, the result is what I call the ‘extraction proposal’: certain apparent properties can count as visually apparent properties because they can be extracted from a combination of distinct apparent properties that are more straightforwardly visually presented properties. Call the former kind of apparent properties ‘extracted properties’ and the latter ‘basic properties’. An extraction proposal has to make plausible that a given extracted property is bona fide visually apparent. The link between extracted and basic properties must be such that it allows us to think of the extracted property as somehow manifested in the visual presentation of the basic properties. This requires that the basic properties themselves fix or determine the extracted property to a sufficient degree, rather than, say, that the former merely

Mind-Independence and Visual Phenomenology

391

provide grounds for inferring the latter, significantly mediated by relevant background beliefs. Otherwise, the extracted property would still not count as visually apparent. To get a feel for the extraction proposal and what it requires, let us consider some specific cases. Large circle: Suppose you are looking at a mosaic of square glass pieces on a wall from an appropriate distance. It looks as if there is a large circle on the wall, although you can also still make out the square pieces of glass. There is a sense in which you see the large circle in virtue of seeing the small squares. The apparent properties of the small squares jointly determine the appearance of the large circle on the wall. But it seems odd to speak of the apparent property of being a circle as extracted from the apparent properties of the squares. Intuitively, in this case you simply see the circle: the squares’ color and geometrical arrangement produce the appearance of the circle on the wall because the squares form into a circle, which can be seen from a certain distance. Being a circle is an observational property par excellence. Plump tomato: Suppose you see a tomato before you. It looks a deep shade of red, evenly rounded, shiny and plump. Intuitively, the tomato’s appearance of plumpness is due to its appearance as round, shiny, and deeply red (and perhaps other apparent properties). Looking plump, though, is more than simply looking deeply-red-and-round-and-shiny. It is a further, distinct way in which the tomato appears to be. Still, the tomato looks plump because it looks shiny, round and deeply red. This manifestation of a distinct property in the joint appearance of others can be thought of as a case of extraction. The appearance of being plump is extracted from the appearance of being shiny, round, and deeply red. The latter apparent properties jointly manifest the tomato’s appearance of plumpness without apparent plumpness reducing to the conjunction of the former. Let us consider another case of extraction. Ripe pear: Suppose you see a pear on a table before you. The pear looks yellowishgreen, smooth and ripe. Intuitively, the pear’s appearance of being ripe owes much to its appearance as yellowish-green and smooth (and perhaps other apparent properties). Ripeness is a state of pieces of fruit and vegetable at which they are ready to eat. Having reached that state typically is a matter of color and surface texture (as well as internal consistency, smell, and taste, of course.) Looking ripe, however, is not merely the conjunction of looking yellowish-green and smooth, but it is a further property the pear appears to have. Insofar as appearing ripe is manifested in the pear’s joint appearance

392

Introspection and the Nature of Experience

of yellowish-green and smoothness, it does so without reducing to their conjunction. In that sense, the appearance of ripeness is extracted from other apparent properties. However, one also might reasonably think that the pear’s looking yellowish-green and smooth alone is not sufficient to give rise to the appearance of the pear’s being ripe. In particular, it seems that one also needs certain background beliefs connecting the surface texture and color of pears with their readiness to be eaten. Suppose we agree that such background beliefs are, in conjunction with the apparent properties of being yellowish-green and smooth, necessary and operative in the production of the pear’s appearance of ripeness. The role of background beliefs in the extraction of a given apparent property raises the question whether the extracted property is visually presented. Depending on the role the background beliefs play in producing the appearance of the extracted property, this question could get a positive or negative answer. For the purposes of this chapter, it is enough to note that in some cases there is an overall tendency to give a ‘yes’ answer, in some cases there is an overall tendency to give a ‘no’ answer and in some cases there is no strong tendency either way. The ripe pear case strikes me as one of the latter, whereas the plump case, insofar as background beliefs are involved, strikes me as one where there will be an overall tendency to a ‘yes’ answer. What about a case generating an overall tendency to give a ‘no’ answer? Let us consider the following case. Healthy dog: Suppose you see a dog running around in the park. The dog looks to have shiny fur and be bright-eyed and healthy. It does seem that the dog’s appearance of being healthy owes much to the dog’s appearance of having shiny fur and bright eyes. In dogs, shiny fur and bright eyes are markers of good health. Looking to have shiny fur and bright eyes, then, is a way the dog looks that indicates good health. Recognizing good health on this basis seems to require background beliefs connecting the shiny-furred and bright-eyed look with a certain underlying physiological state of dog. It is an open question whether the recognition of healthiness is a type of visual recognition, where the background beliefs are merely enabling conditions that make available the visual recognition of a specific property, or whether it is something more akin to an inference-based recognition, where the background beliefs work, together with how things look, in an inference to the effect that the dog is healthy. I think most people would tend toward the latter view, with the result that healthiness would not be considered a visually apparent property although extracted to some extent from other visually apparent properties. The more substantial the role of the background beliefs in the extraction of the apparent property in question, the more plausible the inference-based recognition becomes. A more substantial role of background belief shows that the extracted

Mind-Independence and Visual Phenomenology

393

property is connected less tightly to the basic properties. A specification of the former can be made without significant appeal to the latter (or similar ones). In turn, that means that the background beliefs have more work to do in forging a connection that can be exploited to extract a given extracted property. Plumpness, ripeness, and healthiness, for instance, contrast nicely in this way: a typical specification of each of them makes significant appeal to the relevant basic properties (or similar) to a varying degree—decreasingly so in the order I have listed them. But the weaker the role of the basic properties in producing the extracted property, the less plausible it is to think of the latter as observed rather than inferred. At the end of the last section I said that those who endorse Exclusive Apparent Mind-Independence on the basis of the transparency thesis face the challenge of explaining how mind-independence can be a visually observable property. In this section, I set out a general proposal about how to think of certain apparent properties as extracted from other apparent properties. This seems the most promising way for proponents of Exclusive Apparent Mind-Independence to meet the challenge given their other assumptions. Extraction is compatible with respecting the plain status of Exclusive Apparent Mind-Independence, i.e., the status of Exclusive Apparent Mind-Independence as fairly pre-theoretic introspective data about visual phenomenology. Once one rejects the plain status of introspection-based claims, there are other possibilities for defending Exclusive Apparent MindIndependence, but investigating these goes beyond the aim of this chapter. Any specific proposal about how to extract apparent mind-independence from other apparent properties should aim to be more like the plump tomato case and less like the healthy dog case. Which kind of properties visually present in experience could bring about the manifestation of apparent mind-independence in this way? I now want to explore three different suggestions about which properties could do the job.

4.1. The Spatial Character of Experience According to William Lycan, [S]hapes etc. are (represented as being) objects external to oneself, not as mere contents of one’s consciousness. When one looks into the peep-box, one sees shapes arranged in a complex design, but one sees them as residing inside the box, on the other side of the peep-hole, not in one’s head. They are external things. (1995, 95) Lycan endorses Exclusive Apparent Mind-independence here, but he uses spatial terminology to do so. He talks of shapes being represented as ‘external to oneself ’, where this is meant to contrast with being represented as subjective or mental. Such

394

Introspection and the Nature of Experience

use of spatial terminology is not at all unusual. We often say ‘external world’ or ‘outside world’ when we mean the objective, physical, or mind-independent world. Terms like ‘external’ may be taken to refer to aspects of space or spatiality or to objectivity and mind-independence (see, e.g., Bennett 1971 for relevant discussion). In this vein Sydney Shoemaker (1996a, 102), for instance, holds that the alleged phenomenological fact that features are ‘experienced as spatially located’ motivates antisubjectivist positions in the theory of visual experience. The use of spatial terminology to express Exclusive Apparent Mind-Independence could be held to be licensed by a close connection between experience’s having a certain spatial character, on the one hand, and the seeming mind-independence of visually presented objects, on the other. This close connection, in turn, may yield an account of how apparent mind-independence is extracted from other properties visually presented in experience. One starts with the uncontroversial introspective datum that experience presents entities as spatial in certain ways. Add to this the idea of an intimate link between one’s visual awareness of entities as spatial and one’s awareness of them as mind-independent. I have in mind here, for instance, the sort of link between our ideas of space and objectivity in the concept of experience that has been explored by Kant in the Critique of Pure Reason and then by Strawson (1959, 1966) and Evans (1985). The result is an extraction account of Exclusive Apparent Mind-Independence according to which some entities are presented to one as mindindependent in virtue of being visually presented as spatial in experience. Call this the ‘spatial character extraction proposal’. What sort of spatial character could serve in such an extraction proposal? There is a long tradition of debate among philosophers and psychologists about the correct description of the spatial character of visual experience. One view is that conscious visual awareness is most accurately described in terms of a three-dimensional visual array. Such phenomenal three-dimensionality is sometimes held to endow visual experience with intentionality: in virtue of having three-dimensional spatial character visual experience is directed at, or provides information about, objects and their visible properties. The thought is that such spatial character yields awareness of entities as apparently out there and distinct from oneself (e.g., Smith 2000). Three-dimensional spatial character therefore seems a promising candidate for the extraction strategy. If experience with three-dimensional spatial character involves the visual presentation of objects as out there and distinct from oneself, perhaps this produces their appearance as mind-independent objects. On reflection, though, it becomes clear that it cannot be the simple fact of apparent three-dimensionality alone manifesting apparent mind-independence. Consider visual experiences that are standardly considered not to involve the visual presentation of objects as mind-independent. Common examples of these are the

Mind-Independence and Visual Phenomenology

395

visual experience of seeing stars when receiving a blow to the head, or the visual experience one has when pressing one’s hands against closed eyes and certain kinds of after-image experiences. These sort of visual experiences are sometimes called ‘pure visual sensations’ (as opposed to visual perceptions). When undergoing pure visual sensations, whatever one is visually presented with does not seem to be mind-independent; presumably, even defenders of Apparent Mind-Independence would agree with this (see, e.g., Siegel 2006 and Smith 2000). Yet, pure visual sensations do have a spatial character, in particular they sometimes exhibit depth. William James, for instance, spoke of the ‘voluminousness’ that he supposed was part of the character of such sensations ( James 1950). Moreover, in addition to voluminousness, pure visual sensations also feature an egocentric perspective, a sense that the objects of experience are before one. Suppose you have an after-image experience in a pitch dark room, say, after a very bright light has been switched off. You are visually presented with a shape at a certain depth and location. This will be an apparent location with respect to you: an egocentrically specified spatial relation. You do not experience the shape as somehow contiguous with you or as part of you, but you are visually presented with it as located right in front, to the left, etc. This is a sense of out and distinct from one, yet it does not manifest the appearance of mind-independence of the entities presented. So the simple fact of threedimensional spatial character involving a sense of outness and distinctness is insufficient on its own for producing the appearance of mind-independence. In response, a proponent of Exclusive Apparent Mind-Independence could perhaps deny that there is a phenomenal difference between pure visual sensations and ordinary visual perceptions in the first place. She would hold that, given their shared three-dimensional spatial character, they both in fact present us with things as mindindependent. She could then offer an explanation of why, in the case of pure visual sensations, it does not seem as if one is presented with mind-independent objects in terms of background beliefs and expectations overlaying the visual appearance to the contrary. This reply is open to the proponent of Exclusive Apparent Mind-Independence, but I think it would be a strategic mistake, given the aim to secure Apparent MindIndependence as a plain introspective datum. The reply concedes that introspection of what experience is like in some cases does not deliver awareness of what experience is really like because the deliverance of introspection is infected by background expectation. Yet, that this is so is not detectable on the basis of the deliverance of introspection. What is more, the influence of these background beliefs is rather strong, and it does not seem possible to bracket them or ignore them in some way for the sake of introspecting the true—mind-independent—character of these experiences. Hence, in those cases, we ought not trust introspection to provide us

396

Introspection and the Nature of Experience

with data about visual phenomenal character. This stands in stark contrast to the confidence with which a proponent of Exclusive Apparent Mind-Independence uses introspection in the case of ordinary perceptual experience, or more polemically put, in the case where introspection serves up favorable sort of data. But once doubt about the suitability of introspection to deliver adequate evidence about visual phenomenal character is raised, it will be hard to restrict such doubt to desired cases in the larger context of the debate.

4.2. Perceptual Constancies Another suggestion about how to extract Exclusive Apparent Mind-Independence from other, more basic introspectible features of phenomenal character would be to focus on perceptual constancy phenomena.8 There are different types of perceptual constancies associated with vision. For example, there are color constancies, such as when a green tablecloth is seen to be of uniform color even though half of it is exposed to bright sunlight and half of it is in the shade; shape constancies, such as when a plate is seen to be round as one examines it from various different angles; and size constancies, such as when a tower is seen to be the same size as one walks toward it from some distance. The perceptual constancies proposal starts out with the observation that introspection reveals that perceptual constancy phenomena are part of the visual phenomenal character of ordinary perceptual experience. In particular, the putative introspective datum is that in ordinary perception some objects appear to be changing with respect to one’s viewing of them where this does not appear to be a change in the object itself: the object seen appears, at the same time, to have properties that are independent of one’s viewing of them. This characterization of how perceptual constancy is manifest in phenomenal character is a very promising candidate for proponents of Exclusive Apparent Mind-Independence in responding to the challenge. So characterized, the appearance of phenomenal constancy can reasonably be taken to ground the appearance of perception-independence and thus mind-independence. This is because the characterization in terms of properties dependent and independent of the subject’s viewing introduces a distinction between the subject’s experience and the object experienced, where the latter is taken to be independent of the former. That distinction is a way to formulate the mind-independence of the object of experience. Hence, if introspection reveals that perceptual constancy phenomena so characterized are part of phenomenal character, then this would provide a workable basis for an extraction proposal for Exclusive Apparent Mind-Independence. Indeed, the 8

Smith also discusses perceptual constancy phenomena (2002, 170–184).

Mind-Independence and Visual Phenomenology

397

case might be considered similar to the round circle case in that it is not clear that apparent mind-independence is extracted from other apparent properties rather than visually manifested simpliciter in the guise of perceptual constancy. But do we have to accept that introspection reveals perceptual constancies as part of the phenomenal character visual experience? To just outright deny that perceptual constancy phenomena are part of what it is visually like to undergo experience is not plausible. Let us grant then that introspection reveals that phenomenal character of experience involves perceptual constancy phenomena. However, these phenomena may be characterized in different ways. The characterization above is rather full-blooded. According to it, something appears to have both properties dependent on, and properties independent of, one’s viewing of it. A more austere characterization would be to say that when undergoing a course of perceptual experience, something can appear the same yet different. There is a pattern of stability and change in what it is like for one to undergo ordinary experience. This austere characterization is neutral on what kind of things appear the same and different, and it is neutral on what sort of dependencies underlie a given pattern of stability and change. One advantage of this austere formulation over the full-blooded one is that it captures well what is in common between different kinds of constancy phenomena, namely a pattern of stability and change (Hilbert 2005). The austere characterization therefore easily includes color constancy, which is, after all, not primarily due to different subject-dependent viewing conditions, but to different external lighting conditions. Moreover, the full-blooded characterization includes in its specification of perceptual constancy reference to the subject’s viewings. But the subject and her perceiving or viewings are not straightforwardly observable in the scene before one—the way they figure in the full-blooded characterization of phenomenal constancies suggests that they are merely implicit in what one does observe. How is that sort of implicit feature introspectible as part of phenomenal character then? The question opens up a role for background belief to supply these aspects of phenomenal constancy. In light of this, evaluating what sort of extraction case the perceptual constancy extraction proposal offers by comparing it to the round circle, plump tomato, ripe pear, and healthy dog cases becomes more difficult. In particular, it now seems that all four cases are still on the table. But that means, that it is not clear that apparent mind-independence, although extracted, can count as bona fide visually apparent. If the only decent characterization of apparent perceptual constancy available were the full-blooded one, then—given the assumption that introspection does reveal the phenomenon to be part of visual phenomenal character—the above

398

Introspection and the Nature of Experience

worry would not get going. But we have an alternative characterization of the datum: the austere characterization. It is this that opens up the space of possibilities about the sort of extraction at issue. With the austere characterization in hand, the assumption that introspection reveals perceptual constancy phenomena in visual phenomenal character does not in and of itself guarantee an extraction proposal successful in meeting the challenge for proponents of Exclusive Apparent MindIndependence. According to the austere characterization, the basic apparent properties manifest a pattern of stability and change. Further substantial assumptions need to be made to characterize the metaphysical status of the things exhibiting these patterns of stability and change in the light of the kind of dependencies involved. If these more full-blooded apparent properties are extracted, they need serious help from background belief.

4.3. Bodily Movement Another version of the extraction proposal for Exclusive Apparent Mind-Independence focuses on the interaction between after images (as well as other kinds of pure visual sensations) and proprioception, specifically, one’s sense of bodily movement. The strategy is to claim that there is a systematic difference between how certain proprioceptive information relates to kinds of visual experience. The difference is then exploited in an extraction of apparent mind-independence. The relevant contrast is this. On the one hand, self-initiated bodily movements do not seem to alter how things appear to one when undergoing a pure visual sensation, such as an afterimage experience. In particular, whenever one moves one’s body, the presented after image moves as well, retaining its fixed position in the visual field. On the other hand, such bodily movements do typically seem to alter how things appear to one in ordinary visual perception. When one moves one’s body, the objects presented in ordinary perception tend to change in certain ways, including how they are located in the visual field. One might then hold that these different ways in which visual phenomenal character is affected by bodily movement suffice to ground the appearance of mind-independence of what is presented in ordinary perception, as opposed to the apparent mind-dependence (or neutrality) of what is presented in pure sensation. The bodily movement extraction proposal makes a plausible case for what might be involved in one’s overall experience of objects as mind-independent. It also avoids a significant role for background belief in generating the extraction, and so it does not seem threatened by easy comparison with the ripe pear case or the healthy dog case. But the challenge is to make plausible how Exclusive Apparent MindIndependence can be introspectible as part of visual phenomenal character. It is not

Mind-Independence and Visual Phenomenology

399

clear that the bodily movement extraction proposal can do so. If the appearance of mind-independence is the result of interaction between proprioceptive experience involved in self-initiated movement and visual experience, then the appearance of mind-independence seems likely to be the joint upshot of how things are visually presented and one’s proprioceptive or kinesthetic awareness of one’s own movement. In this case, one’s introspective judgment that what it is like for one is that things appear as mind-independent reflects not only what one’s visual experience is like but also the character one’s proprioceptive experience and perhaps also certain expectations shaped by a process of learning from a systematic contrast within the course of one’s experiences. If anything, it seems that this proposal brings out just how wrongheaded the idea of Exclusive Apparent Mind-Independence as an introspectible part of visual phenomenology is in the first place. Many cognitive resources are involved in one’s experience of objects as mind-independent. The proponent of the bodily movement extraction proposal might suggest that it is not the actual proprioceptive awareness of movement from which apparent mind-independence is extracted. Rather, it is extracted from the associated dispositions had by objects perceived, e.g., its being such that if I move relative to the object, the object will not thereby move (for a proposal along these lines see Siegel 2006). But it is even less clear how this dispositional property could be plainly available to introspection.

5. Conclusion I began with a putative introspective datum about the phenomenal character of ordinary visual experience—Exclusive Apparent Mind-Independence—that is frequently used as introspective evidence by philosophers of perception in theorizing about the nature of experience. Due to worries arising from introspective disagreement and from reasons more specific to the property of mind-independence, these philosophers face the challenge of saying how apparent mind-independence can be an introspectible part of phenomenal character if mind-independence is not among the ordinary objects and properties observed when having the experience. I explored the idea of extracting apparent mind-independence from other, straightforwardly introspectively available, apparent properties. I considered three specific extraction proposals, but in each case, the extraction of mind-independence from the relevant basic apparent properties did not yield a compelling answer to the challenge. Ultimately, Exclusive Apparent Mind-Independence does not look like a simple introspectively available datum about the character of visual experience. Its current status as introspective evidence in theorizing about perceptual experience, however,

400

Introspection and the Nature of Experience

suggests that it is. Once worries about its introspectability are introduced, its role in theorizing about perceptual experience ought to be reconsidered. It is unclear that proponents of world-directed views will be able to recover the strength of some of their antisubjectivist arguments in light of this reevaluation of the introspective evidence. So, for instance, attacks on subjectivist views as phenomenally inadequate are not going to command the same force once it is recognized that the accusers’ arguments turn on introspective judgments about phenomenology whose etiology is murky at best. In this chapter I have focused on a particular putative introspective datum, Exclusive Apparent Mind-Independence. As explained in the beginning, its role in current theorizing about visual experience role is an instance of an introspective method many philosophers employ—typically not exclusively, of course—to theorize about experience. I also said that it is a central component of this method, i.e. of the use of introspective evidence to constrain an account of visual experience, that introspective data has plain status. The conclusion of my investigation in this chapter undermines this plain status for a particular putative introspective datum, Exclusive Apparent Mind-Independence. As just pointed out, it consequently diminishes the force of arguments appealing to Exclusive Apparent Mind-Independence as introspective evidence against subjectivist views and in defenses of world-individuating views. But it thereby also strengthens the case for skepticism about the introspective method in general. Such skepticism takes its cue from the persistent and radical disagreement about the deliverance of introspection that runs through the debate about the nature of perceptual experience and of which Exclusive Apparent Mind-Independence is a part. According to this skeptical perspective, the best explanation of why there are conflicting introspection-based claims about what it is like to undergo experience involves denying that phenomenal character is plain to one in introspection (see Spener ms). Instead, introspection-based claims about phenomenal character, such as Exclusive Apparent Mind-Independence, are largely a reflection of the introspective subject’s antecedent expectation about the metaphysics of phenomenal character (as well as expectations fixed by other relevant philosophical commitments and background belief). Let me emphasize that it is not the mere fact of expectations driving introspective judgment that is problematic in this context. All of our evidence might be held to be theory-laden in some sense or other and plainness ought not to require introspective judgments to form an exception on this point. But the notion of plainness is consistent with some expectations driving the judgments. What it demands is that in order for the introspective judgments to provide rational constraints on theorizing about experience, the expectations in question must not be shaped by antecedent theoretical commitments of the kind that are up for debate.

Mind-Independence and Visual Phenomenology

401

The trouble is that participants in the debate mistakenly suppose that their introspection-based judgments are primarily driven by the phenomenal character of the experience when really they are expectation-driven in a way that undermines their role as fairly independent constraints on theorizing. Introspection does not itself clarify the extent to which introspection-based judgment about experience is driven by an introspector’s expectation about that nature rather than by the phenomenal character of experience itself. Hence, the radical and resilient disagreement about the nature of experience among philosophers in the debate who employ the introspective method is itself best explained by the following conjunctive hypothesis: first, introspection-based views about the nature of experience are driven by expectation rather than by the phenomenal character of experience; second, participants in the debate about the nature of experience and phenomenal consciousness harbor wildly differing expectations about that nature and it is those and other closely related expectations that drive their introspective judgments. The expectation-driven nature of introspective reflection is a good reason to reject the assumption that phenomenal character is plain to introspection—at least when it comes to aspects allegedly suggestive of the metaphysics of phenomenal character. This means that introspective evidence about the character of experience does not have the status of pre-theoretic data in the relevant sense; it cannot provide a neutral—or even fairly neutral—arbiter between theories about the nature of visual consciousness.9

References Anscombe, Elisabeth. 1965. “The Intentionality of Sensation: A Grammatical Feature.” Pp. 158–190 in Analytical Philosophy, second series, edited by R. J. Butler. Oxford: Basil Blackwell. Armstrong, David M. 1968. A Materialist Theory of the Mind. London: Routledge & Kegan Paul. Ayer, A. J. 1973. The Central Questions of Philosophy. London: Weidenfeld. Bayne,Tim, and Maja Spener. 2010.“Introspective Humility.” Philosophical Issues 20: 1–22. Beebee, Helen. 2003. “Seeing Causing.” Proceedings of the Aristotelian Society 103: 257–280. Bennett, Jonathan. 1971. Locke, Berkeley, Hume: Central Themes. Oxford: Clarendon Press. Block, Ned. 1980. “Troubles with Functionalism.” Pp. 268–305 in Readings in Philosophy of Psychology,Vol. 1, edited by N. Block. Cambridge, Mass.: Harvard University Press.

9 I am grateful to Tim Bayne, David Chalmers, Michael Martin, and Ram Neta for their helpful comments. Special thanks to Susanna Siegel.

402

Introspection and the Nature of Experience

Broad, C. D. 1965. “Some Elementary Reflexions on Sense-Perception.” Pp. 27–48 in Perceiving, Sensing and Knowing, edited by R. Swartz. New York: Anchor Books. Originally published in Philosophy 27 (1952). Byrne, Alex. 2001. “Intentionalism Defended.” Philosophical Review 11: 49–90. Crane, Tim. 2005 “The Problem of Perception.” Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta. Dretske, Fred. 1995. Naturalizing the Mind. Cambridge, Mass.: MIT Press. Evans, Gareth. 1985. “Things Without the Mind.” Pp. 249–290 in his Collected Papers. Oxford: Clarendon Press. Harman, Gilbert. 1990. “The Intrinsic Quality of Experience.” Pp. 31–52 in Philosophical Perspectives,Vol. 4, edited by J. E. Tomberlin. Atascadero, Calif.: Ridgeview. Hilbert, David. 2005. “Color Constancy and the Complexity of Color.” Philosophical Topics 33: 141–158. Hume, David. 1978. A Treatise of Human Nature. 2d ed. Edited by P. H. Nidditch. Oxford: Clarendon. James, William. 1950. The Principles of Psychology,Vol. 2. New York: Dover Press. Kind, Amy. 2003. “What’s so Transparent about Transparency?” Philosophical Studies 115(3): 225–244. Levine, Joseph. 2001. Purple Haze. Oxford: Oxford University Press. Lycan, William. 1995. “Layered Perceptual Representation.” Pp. 81–100 in Philosophical Issues, Vol. 7, edited by E. Villanueva. Atascadero, Calif.: Ridgeview Publishing Company. Martin, M. G. F. 1997. “The Reality of Appearances.” Pp. 81–106 in Thought and Ontology, edited by M. Sainsbury. Milano, Italy: Franco Angeli. ———. 2000. “Beyond Dispute: Sense-Data, Intentionality and the Mind–Body Problem.” Pp. 195–231 in History of the Mind–Body Problem, edited by T. Crane and S. Patterson. London: Routledge. ———. 2002. “The Transparency of Experience.” Mind & Language 17: 376–425. McDowell, John. 1998. “Lecture I: Sellars on Perceptual Experience.” Journal of Philosophy 95: 431–450. McGinn, Colin. 1996. “Another Look at Colour.” Journal of Philosophy 93: 537–553. Michotte, Albert. 1963. The Perception of Causality. London: Methuen. Peacocke, Christopher. 1993. “Intuitive Mechanics, Psychological Reality, and the Idea of a Material Object.” Pp. 162–176 in Spatial Representation, edited by N. Eilan, R. McCarthy, and B. Brewer. Oxford: Basil Blackwell. Schwitzgebel, Eric. 2008. “The Unreliability of Naive Introspection.” Philosophical Review 117: 245–273. ———. 2011. Perplexities of Consciousness. Cambridge, Mass.: MIT Press. Shoemaker, Sydney. 1996a.“Qualities and Qualia:What’s in the Mind?” Pp. 97–120 in his The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press.

Mind-Independence and Visual Phenomenology

403

———. 1996b. “Self-Knowledge and ‘Inner Sense.’ Lecture III.” Pp. 246–268 in his The First-Person Perspective and Other Essays. Cambridge: Cambridge University Press. Siewert, Charles. 2004.“Is Experience Transparent?” Philosophical Studies 117(1–2): 15–41. Siegel, Susanna. 2006. “Subject and Object in the Contents of Visual Experience.” Philosophical Review 115(3): 355–388. ———. 2008. “The Visual Experience of Causation.” Philosophical Quarterly 59: 1–22. ———. 2010. The Contents of Experience. New York: Oxford University Press. Smith, A. David. 2000. “Space and Sight.” Mind 103: 481–518. ———. 2002. The Problem of Perception. Cambridge, Mass.: Harvard University Press. Spener, Maja. 2011. “Disagreement about Cognitive Phenomenology.” In Cognitive Phenomenology, edited by T. Bayne and M. Montague. Oxford: Oxford University Press. ———. ms. “Introspective Evidence and Phenomenal Character.” Manuscript. Stoljar, Daniel. 2004. “The Argument from Diaphanousness.” In Language Mind and World: Special Issue of the Canadian Journal of Philosophy, edited by M. Escurdia, R. Stainton, and C.Viger. Strawson, P. F. 1959. Individuals. Oxford: Oxford University Press. ———. 1966. Bounds of Sense. Oxford: Oxford University Press. ———. 1979. “Perception and its Objects.” Pp. 41–60 in Perception and Identity: Essays Presented to A. J. Ayer, edited by G. F. MacDonald. Ithaca, N.Y.: Cornell University Press. Sturgeon, Scott. 2000. Matters of Mind. London: Routledge. Thau, Michael. 2002. Consciousness and Cognition. New York: Oxford University Press. Tye, Michael. 1995. Ten Problems of Consciousness. Cambridge, Mass.: MIT Press. ———. 2000. Consciousness, Color, and Content. Cambridge, Mass.: MIT Press. ———. 2003. “Representationalism and the Transparency of Experience.” Pp. 31–43 in Privileged Access, edited by Brie Gertler. Aldershot: Ashgate. Zahavi, Dan. 2005. Subjectivity and Selfhood: Investigating the First-Person Perspective. Cambridge, Mass.: Bradford Books/MIT Press.

This page intentionally left blank

15 Introspection about Phenomenal Consciousness: Running the Gamut from Infallibility to Impotence Terry Horgan

Phenomenal consciousness comprises those aspects of one’s mental life such that there is “something that it is like” for the experiencing agent to undergo them. The phenomenal character of experience is intrinsic, qua mental; hence, there is more to it than its functional role in the agent’s cognitive economy—since functional role is nonintrinsic, being entirely a matter of typical-cause relations between a given state type and other state types instantiable either by the agent or by the agent’s wider environment. The intrinsicness of phenomenal character is the reason why one can conceive of physical/functional duplicates of ourselves who are phenomenal inverts—creatures who are physically, and hence functionally, just like ordinary humans but in whom certain kinds of phenomenal character (e.g., colorexperience phenomenal character) are systematically inverted relative to ourselves. The intrinsicness of the phenomenal is also the reason why one can conceive of zombies—creatures who are physically, and hence functionally, just like ordinary humans but whose internal states lack intrinsic phenomenal character altogether. (Whether such inverts or zombies are metaphysically possible is, of course, a highly

406

Introspection and the Nature of Experience

tendentious issue in philosophy of mind. But nowadays the conceivability of such creatures is usually conceded, even by many who are materialists about mentality.) Phenomenal character is, by its very nature, self-presenting to the experiencing agent. That is, the what-it’s-like-ness of phenomenal consciousness is something immediately given in experience. As this point is often put, in the case of phenomenal consciousness there is no gap between appearance and reality, because the appearance just is the reality: how the phenomenal character seems, to the agent, is how it is. Because of the self-presentingness of phenomenal character, and the consequent lack of an appearance/reality gap, a prima facie plausible thought is that introspection, insofar as it seeks to ascertain facts about current phenomenal character itself, is an extremely reliable belief-forming process, perhaps even an infallible one. One simply attends to some aspect of the phenomenal character of one’s present experience—some aspect of the appearance-that-is-the-reality—and then forms a belief about the intrinsic phenomenal nature of that self-presented phenomenal character by directly “reading off ” its self-presented nature. My project in this chapter is to subject this thought to critical scrutiny. I will argue that there are (at least) three kinds of phenomenal beliefs—where by ‘phenomenal belief ’ I mean a belief that is entirely about intrinsic phenomenal aspects of one’s current experience. First are phenomenal beliefs that can be especially reliably produced by introspection and are especially well justified when so produced, but which nonetheless possess a certain specific, narrow, kind of fallibility. Second are phenomenal beliefs that are outright infallible when introspectively produced; they do not possess even the limited form of the fallibility that accrues to introspectively produced phenomenal beliefs of the first kind. Third—and perhaps somewhat surprisingly—are phenomenal beliefs that cannot be reliably produced on the basis of introspection alone, and for which introspection alone does not provide good evidential support. Much of the chapter will be devoted to an example involving beliefs of the third kind—an example pertaining to the phenomenology of agency. I will argue that introspection by itself is virtually impotent as a way of forming certain sorts of beliefs concerning the intrinsic character of agentive phenomenology, and I will offer a proposed explanation of this impotence. Before proceeding, let me make some clarificatory remarks about how I am here construing the self-presentingness—i.e., the givenness—of phenomenal character. First, I take self-presentingness to be an especially intimate form of direct acquaintance between the experiencing subject and the phenomenal character of some aspect of the subject’s current state of phenomenal consciousness; the state’s

Introspection about Phenomenal Consciousness

407

appearing a certain way, acquaintance-wise, is constitutive of the state’s actually being that way. Second, I do not construe the state’s appearing a certain way, in the relevant sense of ‘appearing’, as a matter of the subject’s having some kind of belief about the state; on the contrary, on my view the way the phenomenal state appears to the subject is distinct from—and typically is the epistemological justification for—an appearance-based belief about the intrinsic phenomenal character of the experience. Third, I mean to be neutral here about certain issues concerning the nature of phenomenal self-presentingness—e.g., whether it is a matter of phenomenal states representing themselves secondarily while also having primary representational content of a different kind, or whether instead it is a matter of the experiencing subject being in a certain sort of higher-order intentional state that is directed toward the phenomenal state, or whether instead the self-presentation of phenomenal character is somehow importantly different from representation (either self-representation or higher-order representation). I am presuming that the reader has enough of a pre-theoretic grip on the notion of self-presentingness to be able to understand and evaluate the discussion that follows, and that the issues I will be addressing are largely orthogonal to issues about the structure of selfpresentation itself.

1. Super-Reliable Introspection Consider a phenomenal belief expressible by saying “I am now having a reddish experience”—where the word ‘reddish’ is used to describe the color phenomenology of the experience itself (and not to attribute the property of redness to something in one’s ambient environment). This belief has two notable characteristics. First, if one possesses the phenomenal concept of reddishness and is competent in deploying it, then one is capable of accurately and spontaneously applying this concept (both affirmatively and negatively) to one’s current experience, just on the basis of attending introspectively to the experience itself—without reliance on collateral information or evidence. Second, because there is no appearance/ reality gap as regards the phenomenal character of one’s current experience, evidently this belief is not susceptible to a form of fallibility—call it appearance/ reality fallibility—possessed by certain other kinds of beliefs, notably external-world beliefs that arise from one’s current sensory experience. Consider, for instance, a belief expressible as “There is something red in front of me,” formed by a visualmode sensory experience as-of the presence of a red object in front of oneself. This external-world belief has appearance/reality fallibility, by virtue of the possibility that the representational content of one’s present perceptual experience is

408

Introspection and the Nature of Experience

highly deceptive and highly nonveridical—say, because one is the victim of a Cartesian evil deceiver, or because one is an envatted brain. By contrast, one’s introspection-based belief that one is now having a reddish experience cannot be nonveridical in that way, because this belief is about the intrinsic phenomenal character of one’s present experience itself—not about some further matters that one’s experience represents as being the case (e.g., that there is something red in front of oneself ). Although the belief that I am now having a reddish experience lacks appearance/ reality fallibility, it is not infallible tout court. In order to appreciate why not, it will be helpful to distinguish two kinds of introspection concerning one’s current experience. On one hand, there is attentive introspection: paying attention to certain aspects of one’s current experience. On the other hand, there is judgmental introspection: the process of forming a judgment about the nature of one’s current experience, and doing so spontaneously just on the basis of attending to the aspects(s) of one’s current experience about which one is judging—without any reliance on collateral information or evidence. ( Judgmental introspection thus deploys attentive introspection, while also generating a judgment about what is being attended to.) One’s belief that one is now having a reddish experience is not appearance/reality fallible, because there is no appearance/reality gap here. It does possess a limited kind of fallibility nonetheless. For, there is a possibility (however remote) that in the present case, one’s judgmental faculties have not operated properly during the process of judgmental introspection. For instance, perhaps the experience one is now classifying under the category phenomenally reddish is not suitably type-similar to other experiences that one has so classified or that other people so classify. (Maybe one is having a phenomenally purplish experience, but one is misremembering how to apply the predicative category phenomenally reddish.) This kind of fallibility stems not from any appearance/reality gap concerning the object of one’s judgment (viz., the phenomenal color-character of one’s current experience), but rather from the possibility that one is conducting judgmental introspection not by exercising full-fledged conceptual competence but rather in a way that embodies a lapse in conceptual competence—a performance error. I will refer to this kind of fallibility as labeling fallibility. The idea behind this terminology is that even though the phenomenal color-character of one’s current experience is self-presented in a way that does not allow for any appearance/reality gap, and even though one is able to correctly categorize this current experience under the pertinent phenomenal concept just by exercising one’s competence with that concept while attending to the experience itself, nevertheless there remains the possibility (however remote) that one is mislabeling that immediately given experience, in one’s judgment and/or in one’s overt linguistic expression of one’s

Introspection about Phenomenal Consciousness

409

judgment, by classifying it as phenomenally reddish when it is really phenomenally purplish. That is, there remains the possibility (however remote) of a lapse in conceptual competence.1 I take it that numerous phenomenal beliefs that are formed by judgmental introspection have the features lately noted: (1) humans can form these beliefs just by exercising competence with phenomenal concepts2 while introspectively attending to current experience, (2) the beliefs are not appearance/reality fallible, but nevertheless (3) they do possess the limited kind of fallibility I am calling labeling fallibility. For beliefs of this kind, judgmental introspection is an extraordinarily reliable beliefforming process, even though it falls short of outright infallibility. Indeed, judgmental introspection can appropriately be called super-reliable relative to beliefs of this kind, by virtue of its extremely high reliability when forming such beliefs and its immunity from appearance/reality fallibility. (Because the absence of an appearance/ reality gap renders judgmental introspection super-reliable in relation to so many phenomenal beliefs, virtually nobody finds intuitively gripping the idea of radical skepticism concerning such beliefs. The contrast with radical external-world skepticism, which grips the philosophical imagination strongly because of the appearance/ reality gap it brings into focus, is striking.3)

2. Infallible Introspection Uriah Kriegel and I have recently argued that there are certain kinds of introspective judgments concerning one’s current phenomenology that are outright infallible (Horgan and Kriegel 2007). In this section I will briefly summarize our 1

There is even a remote possibility that one has not only committed a performance error, but that one has actually lost—or has never really possessed—competence with the concept of phenomenal reddishness and/or the expression ‘phenomenally reddish’. 2 How rich is the stock of phenomenal concepts that most people possess, and what does it take to be fully competent with such concepts? I myself am inclined toward liberalism about these questions. As regards sensory-perceptual experience, for example, I would claim that the key thing is just the (commonly possessed) ability to be able to articulate the contents of such experience via the “bracketing” use of qualifier expressions like ‘It seems that . . .’—where what is bracketed is the question of whether things are as they appear to be. (A yet more restrictive form of bracketing is described in section 2 below.) 3 For most of us, radical “internal-world skepticism” about what one is now thinking also gets no grip on the philosophical imagination. I would maintain that the best explanation for this fact—and the only adequate explanation—is that it is grounded in an appreciation (however unacknowledged) of self-presenting cognitive phenomenology—the “what it is like” of occurrent thought. See Horgan, Tienson, and Graham 2006.

410

Introspection and the Nature of Experience

position and one key aspect of its rationale. The infallibility thesis we advance is restricted to a special class of beliefs about phenomenal consciousness—viz., phenomenal beliefs that we call “singular, present, prototypical, and bracketed” (SPPB phenomenal beliefs). Such a belief is singular in this sense: it predicates a phenomenal property (e.g., phenomenal reddishness) of a token mental state of oneself. (Alternatively, one could let such a belief be one that predicates a phenomenal property of oneself—say, the property experiencing redly.) The token phenomenal state is being deployed in the belief as a singular thought-constituent that is its own mode of presentation and denotes itself; and the token phenomenal state is also being deployed as a predicative thought constituent that is its own mode of presentation—this time to pick out a phenomenal property that the state itself tokens. An SPPB belief is present in the sense that it concerns the phenomenal character of one’s current experience. The belief is prototypical in the sense that it does not arise in extraordinary circumstances where one’s prior beliefs could distort the selfpresentingness of the token experience’s phenomenal character (as some think could happen, for instance, if one is blindfolded and experiences an ice cube touching one’s cheek when one was expecting to feel one’s cheek being cut by a razorblade). And the belief is bracketed in the sense that its being true does not require the truth of any fallible presuppositions—e.g., the presupposition of a pertinent type-similarity between one’s current experience and others that one has classified, or would classify, as being phenomenally reddish. Letting the notation ‘Bnow’ express the idea that the belief is indexically temporally directed to the present moment (the very moment at which the belief is occurring), Kriegel and I formulate the proposed infallibility thesis as follows: Necessarily, if a subject S has a singular phenomenal belief Bnow[e, p] at a time t whose singular and predicative constituents are the bracketed phenomenal modes of presentation e and p, respectively, and if there is a time t that (1) is roughly simultaneous with t, and (2) is such that S has the belief Bnow[e, p] at t, then Bnow[e, p] is true at t. SPPB phenomenal beliefs share two features also possessed by the phenomenal beliefs discussed in section 1—viz., (1) they cannot be appearance/reality fallible because there is no appearance/reality gap that pertains to them, and (2) humans can form these beliefs just by exercising conceptual competence with phenomenal concepts while directing attentive introspection to the phenomenal character of present experience. But in addition, the bracketing aspect of SPPB phenomenal beliefs also removes the possibility of labeling fallibility; for, the bracketed presuppositions are precisely the ones whose possible falsity is constitutive of labeling fallibility. The upshot is that SPPB phenomenal beliefs are outright infallible.

Introspection about Phenomenal Consciousness

411

How might one articulate an SPPB phenomenal belief using public language if one wants one’s articulation itself to be bracketed? That is a delicate matter, since public-language expressions like ‘phenomenally reddish’ evidently are governed by the kinds of presuppositions that get bracketed in the belief. Plausibly, about the best one can do is to express the SPPB phenomenal belief using indexical terms, as in “This experience has this feature”—where ‘this experience’ is used singularly to verbally correspond to the token experience qua singular belief-constituent and ‘has this feature’ is used predicatively to correspond to the token experience qua predicative belief-constituent. It might be objected that this sort of belief involves a trivial infallibility and, moreover, a kind of infallibility that has nothing distinctively to do with phenomenal consciousness. A belief expressible as “This experience has this feature,” says the objection, is comparable to a belief expressible as “I am here”—comparable in having very thin, purely indexical, content, comparable in being infallible by virtue of the extreme thinness of its content, and comparable in being infallible solely by virtue of this indexical thinness (rather than because of anything distinctive about phenomenal character). The cases are not analogous, however. When one thinks “I am here” in a way that makes one’s belief trivially infallible, one does not have a substantive conception of what “here” stands for.4 One thinks of “here” as simply “the place in which this thought occurs, whatever it may be.” But when one thinks “This experience has this feature,” one does have a substantive conception of the referents of one’s indexical expressions. One thinks of “this feature” not simply as “the phenomenal character of this experience, whatever it may be,” but as the distinctive phenomenal character with which one is presented; and one thinks of “this experience” as a specific current experience with that very character. Because one’s SPPB phenomenal belief deploys bracketed modes of presentation of the experience and its feature, one must use indexical language in order to verbalize the belief in a way that is itself comparably bracketed; the singular and predicative modes of presentation thereby expressed do not readily admit of descriptive articulation.5 But the belief itself is not a purely indexical belief. In other words, “I 4

There may be beliefs that are naturally expressed as “I am here” that do involve a substantive conception of the subject’s location. But such beliefs are not infallible (let alone trivially). 5 Is an SPPB phenomenal belief communicable? Well, not by saying “This experience has this feature.” That language will not communicate the content of the experiencer’s belief to anyone else. But the experiencer might say, for instance, “I am now undergoing a bracketed-greenish experience.”This will successfully communicate the content of the speaker’s belief to those hearers who have the phenomenal concept expressed by ‘greenish’—provided that the presuppositions governing the unbracketed phenomenal concept expressed by ‘greenish’ are true of the bracketed

412

Introspection and the Nature of Experience

am here” not only expresses a belief indexically but also expresses a purely indexical belief. It expresses a belief that employs a thin, purely indexical, mode of presentation of the subject’s location. By contrast, “This experience has this feature” expresses a belief indexically, but does not express a purely indexical belief. The belief it expresses does not use a purely indexical singular mode of presentation of the experience, or a purely indexical predicative mode of presentation of the experience’s phenomenal character. Rather, it uses rich and nonempty singular and predicative modes of presentation, which happen to be bracketed modes of presentation.6

3. Impotent Introspection In light of the fact that phenomenal character is self-presenting, one might be inclined to think that judgmental introspection is always at least super-reliable in the formation of phenomenal beliefs (and is outright infallible in the formation of SPPB phenomenal beliefs). But I will now argue that this is not so at all. On the contrary, I maintain that there are some questions about the intrinsic phenomenal nature of current experience that judgmental introspection is virtually impotent to answer. Let me begin by introducing some terminology and making some distinctions. Let a purely phenomenological question be a question that (1) is about some aspect of the intrinsic phenomenal character of one’s present experience, and (2) is such that the answer is entirely determined just by the intrinsic phenomenal character of one’s present experience. (The point of clause (2) is to exclude questions that bring in some extrinsic aspect while still being in some sense “about” intrinsic phenomenal character—e.g., “Am I now undergoing an experience with the phenomenal character that I was writing about last Tuesday?”) Let a purely phenomenal question be conceptual-competence amenable (for short, CC amenable) just in case it can be correctly answered by simply introspectively attending to one’s current experience and then spontaneously exercising one’s conceptual competence with the pertinent phenomenal concepts. By contrast, let a purely phenomenal question be conceptual-competence transcendent (for short, CC transcendent) just in case it cannot be correctly answered this way. mode of presentation that the speaker is now expressing with the locution ‘bracketed-greenish’. The speaker is justified in believing that this so (and the hearers are too), but this belief is fallible— for the same reasons that the belief expressible by saying “I am now undergoing a greenish experience” is fallible. 6 On these matters, see also Chalmers 2003.

Introspection about Phenomenal Consciousness

413

I maintain that some purely phenomenal questions are CC transcendent. Although they are about the intrinsic character of one’s current phenomenology and their answer is entirely determined by that intrinsic phenomenal character, and although this phenomenology is self-presenting and is not subject to any appearance/reality gap, nonetheless these questions are too subtle to be answerable just by exercising one’s conceptual competence with phenomenal concepts while introspectively attending to that phenomenology. Conceptual competence will not suffice to enable one to just “read off ” the answer to such a question from one’s current experience while introspectively attending to that experience. I now want to give an example of a purely phenomenological question that (I claim) is CC transcendent. The question concerns certain aspects of agentive phenomenology—the what-it’s-like of experiencing oneself as actively generating one’s behaviors.7 One very salient feature of much ordinary agentive phenomenology is the aspect of freedom—the experiential aspect as-of being able to freely choose among a range of alternative potential courses of action, and as-of being such that one could act otherwise than one is in fact acting. Agentive phenomenology, with its aspect of freedom, is richly intentional: it represents oneself, to oneself, as being a freely choosing agentive source of one’s bodily motions.8 Being intentional, such phenomenology has satisfaction conditions: certain ways-things-might-be are in accord with, and others are not in accord with, the intentional content of free-agency phenomenology. Consider, then, the following three pairwise-incompatible claims about these satisfaction conditions. 1. The freedom aspect of agentive phenomenology has satisfaction conditions that (a) are fully fixed by intrinsic phenomenal character alone, and (b) are incompatible with determinism. 2. The freedom aspect of agentive phenomenology has satisfaction conditions that (a) are fully fixed by intrinsic phenomenal character alone, and (b) are compatible with determinism. 7 Largely for simplicity of exposition, I will formulate the question as being about agentive phenomenology in general, and about specific aspects thereof. Strictly speaking, however, what I mean to treat as a CC transcendent question is the corresponding one about one’s own currently tokened agentive experience. Only the more specific version of the question counts as a purely phenomenological question, in the sense defined above. 8 For more on this theme, see, for instance, Horgan and Tienson 2005; Horgan, Tienson, and Graham 2003; and Horgan 2007a, b, forthcoming. One key point stressed in these papers is that the phenomenal character of agentive phenomenology—the what-it’s-like of self-as-source of bodily motions—is not the what-it’s-like of one’s bodily motions being caused by psychological states of oneself.

414

Introspection and the Nature of Experience

3. The freedom aspect of agentive phenomenology has satisfaction conditions that (a) are not fully fixed by phenomenal character alone, (b) instead are fully fixed by phenomenology in combination with extra-phenomenological facts about the experiencing agent’s cognitive architecture, and (c) are such that their compatibility or incompatibility with determinism depends upon those cognitive-architecture facts. Claims (1) and (2) both construe the freedom aspect of agentive phenomenology as having “purely narrow” referential purport that lacks any constitutive externalistic elements, whereas claim (3) construes it has having “wide” referential purport that incorporates certain constitutive externalistic elements. For the freedom aspect to have wide referential purport is for its reference-relation to its referent-property (if it has a referent-property) to depend constitutively not merely on the intrinsic character of the phenomenology itself but also upon certain phenomenology-external facts about the nature of the experiencing agent—according to claim (3), facts about the agent’s cognitive architecture. (The pertinent facts presumably concern the nature of the cognitive-architectural choice-generating and behavior-generating mechanisms that are normally operative in situations where the experiencing agent undergoes free-agency phenomenology.) Claim (3) thus construes the freedom aspect of agentive phenomenology as being analogous to the thought-constituents expressible in public language by natural-kind terms like ‘water’: the freedom aspect is treated as referring to a mode of internal behavior-generation whose essence is not phenomenally given in experience, but instead depends upon the nature of human cognitive architecture. (Whether free agency is compatible with determinism thus depends upon the nature of human cognitive architecture; the answer is not fixed by free-agency phenomenology alone.) Claims (1) and (2), on the other hand, construe the freedom aspect as referring, in the experience of all actual and possible creatures who are phenomenal duplicates of one another, to one and the same property—regardless of any differences in the cognitive architectures of different phenomenal duplicates. The essence of the property that constitutes free agency is entirely fixed by the intrinsic phenomenal character of free-agency experience alone. Claim (1) says that this phenomenologically fixed property is incompatible with determinism, whereas claim (2) says that it is compatible with determinism. Consider now the following question, which pertains entirely to the intrinsic phenomenal character of agentive experience and whose answer depends only on that phenomenal character—and which is therefore a purely phenomenological question: (q) Which (if any) of the pairwise incompatible claims (1)–(3) is correct?

Introspection about Phenomenal Consciousness

415

For present purposes, the issue I want to focus upon is not what the answer is to question (Q), but rather this: whether or not one can reliably ascertain, just via judgmental introspection, what the answer is. I claim that one cannot do so, and that the reason why not is that (Q) is a CC transcendent question. On what basis do I make these claims, and on what basis do I urge you the reader to agree with me? Well, in part via (higher-order) judgmental introspection—mine in my own case, and yours in your case. When I engage in higher-order attentive introspection, directed upon my first-order attempt to ascertain via judgmental introspection which (if any) of claims (1)–(3) is true, I find myself forming the higher-order judgment that I cannot really tell, just on the basis of this first-order attempt at introspective judgment, which (if any) of claims (1)–(3) is true. I urge you the reader, likewise, to undertake the same attempt to form a first-order introspective judgment about which (if any) of claims (1)–(3) is true, while also undertaking the same higher-order introspective attentive monitoring of that first-order introspective attempt. Don’t you yourself form the higher-order introspective judgment, when you do this, that you cannot successfully pull off the attempt to use first-order judgmental introspection to answer question (Q)? Furthermore, this higher-order introspective judgment looks to have the default epistemic status of being one of those highly reliable, albeit fallible and defeasible, introspective judgments about the character of one’s own present phenomenal consciousness. Prima facie, it has much the same kind of especially strong epistemic warrant as do introspective judgments like “I am now experiencing phenomenal reddishness.”Thus, the higher-order introspective judgment that one cannot answer question (Q) just via judgmental introspection constitutes excellent evidence in support of the claim that one really cannot answer (Q) that way.9 Two explanatory tasks need addressing at this point, however. First is the task of explaining credibly why it should be that (Q) is a CC transcendent question. Since claims (1)–(3) all concern only the phenomenal character of free-agency experience, 9

I have assumed in this paragraph that there is rich cognitive phenomenology, and that introspective attention to what one is thinking is therefore a matter of attending to that phenomenology. Although the existence of cognitive phenomenology is (sadly) quite controversial, I have defended it elsewhere, sometimes collaboratively—for instance, Horgan and Tienson 2002; Horgan, Tienson, and Graham 2004; and Horgan forthcoming. See also Goldman 1993; Siewert 1998; Pitt 2004; and Strawson 1994. But even those who deny the existence of cognitive phenomenology should concede that people often have very good introspective access to their current conscious-as-opposed-to-unconscious thought processes. My main claim in the text is that one can tell, by higher-order introspection, that one cannot reliably answer question (Q) by first-order judgmental introspection. This higher-order introspection thus provides good evidence that one really cannot answer question (Q) that way—even if the higher-order introspection is not a matter of attending to cognitive phenomenology.

416

Introspection and the Nature of Experience

and since phenomenal character is self-presenting to the experiencing agent, something needs saying about why human agents are nonetheless unable to “read off ” the answer to question (Q) just by directing their attentive introspection upon their own free-agency experience and then exercising their conceptual competence with concepts like the concept of determinism and the phenomenal concept of free-agency phenomenal character. Second is the task of explaining why some folks (including some sophisticated philosophers) think that one can introspectively “read off ” from one’s free-agency phenomenology the answer to (Q), and that the read-offable answer is that claim (1) is correct. (I know there are sophisticated philosophers who think so, since they have said so to me in philosophical discussion. And I confess to experiencing some temptation to think so myself, as I suspect you the reader do too—a temptation that needs explaining.) I will take up these two explanatory issues in turn.10 How can one explain why judgmental introspection alone is unable to ascertain which (if any) of claims (1)–(3) is true? To begin with, it is important to acknowledge that understanding these claims, and exercising robust introspection to form judgments about them, requires considerable conceptual sophistication. Perhaps not all humans could even understand these three claims; but let us restrict our attention to those who can. Such people do possess a suitable conceptual repertoire to be able to address question (Q), and are able to try to answer it by judgmental introspection. Nevertheless, I have claimed, they cannot reliably arrive at an answer that way. Why not? Because answering question (Q) solely via judgmental introspection would require a degree of cognitive skill in the deployment of the pertinent concepts that far exceeds what is required for the conceptually competent use of these concepts. Even though (a) the answer to question (Q) is entirely determined by the intrinsic phenomenal character of agentive experience, (b) agentive phenomenology itself is self-presenting to the experiencing agent, and (c) the agent has conceptual mastery over all the concepts that figure in claims (1)–(3), nonetheless the agent lacks the capacity to determine, just via exercising conceptual competence while introspectively attending to the freedom aspect of one’s agentive experience, how those concepts would correctly apply to that self-presented phenomenology. Normal human conceptual competence is often mainly a matter of being able to correctly apply a given concept to a concrete case—or more precisely, do so correctly modulo one’s available evidence. Typically one does this automatically and spontaneously; normally it is not done via conscious rehearsal of, and conscious 10 The remainder of this section is adapted, with suitable modifications, from Horgan and Kriegel 2007.

Introspection about Phenomenal Consciousness

417

deployment of, the concept’s satisfaction conditions. Since competent conceptwielding is so much a matter of direct and appropriate application of concepts to concrete cases, conceptual competence alone is apt to be fairly limited as a basis for answering abstract general questions about the nature of satisfaction conditions. What typically happens is that (1) some general claim about the semantics of a given concept or class of concepts is put forward as a hypothesis, and (2) various data are adduced as evidence for or against the hypothesis; in effect, the argument is that the data would be well explained (or would fail to be well explained) by the hypothesis. (Thus, in effect, the argumentation is a matter of inference to the best explanation.) One especially salient form of data, of course, involves spontaneous intuitive judgments about whether or not to apply a given concept to a given concrete scenario—e.g., whether or not to apply the concept knowledge to a given Gettier case, whether or not to apply the concept water to the substance XYZ on Twin Earth, and so forth. Thus, normally the intuitive judgments are about the concrete cases, whereas the reasoning about general hypotheses concerning satisfaction conditions is a matter not of direct intuitive judgment, but rather of abductive inference—inference to the best explanation.11 In light of these remarks, consider what is apt to happen when one introspectively attends to one’s agentive phenomenology with the goal in mind of forming an introspective judgment about question (Q). Claims (1)–(3) are general hypotheses about satisfaction conditions—albeit in this case, hypotheses about the satisfaction conditions that constitute the intentional content of self-presented phenomenology that is not susceptible to any appearance/reality gap. So, given that competent judgmental deployment of concepts is so largely a matter of their correct use in forming concrete categorization judgments, should we expect a conceptually competent person to be able to tell directly, just by exercising conceptual competence while introspectively attending to one’s own free-agency phenomenology, which (if any) of the general hypotheses (1)–(3) is correct? No, we should not. It does not matter that the answer to question (Q) is entirely determined by intrinsic features of experience that are self-presenting and are not subject to an appearance/reality gap. It does not matter because the task envisioned here is to form conceptually sophisticated judgments about this self-presented phenomenology—and, moreover, sophisticated judgments concerning various general hypotheses concerning the satisfaction conditions of agentive experience.

11

On the idea of “ideology” as a form of abductive inquiry in which intuitive judgments about concrete hypothetical scenarios figure as empirical data, see Horgan and Graham 1994. For further discussion that classifies this form of inquiry as “low grade a priori,” see Henderson and Horgan 2000, 2002, 2011.

418

Introspection and the Nature of Experience

That is, the envisioned task is the direct, intuitive, formation of judgments about such general satisfaction-condition hypotheses—just on the basis of introspection itself. It is unreasonable and unwarranted to expect one’s capacity for conceptwielding to be that splendid when it is directed at general hypotheses concerning the intentional content of agentive phenomenology, just as it is unreasonable to expect it to be that splendid when it is directed at general hypotheses concerning the satisfaction conditions for concepts themselves. When it comes to reliability and accuracy of direct, intuitive, judgments, the forte of conceptual competence is concrete judgments about specific cases. General hypotheses about satisfaction conditions are a matter for abductive inference—even when these hypotheses concern facts about the intentional content of self-presenting phenomenal character, facts that are fully fixed by that phenomenal character itself. Let me add some points of clarification. I claim that (Q) is a CC transcendent question—which means that it cannot be answered just by exercising one’s conceptual competence while introspectively attending to the freedom aspect of agentive experience. I also claim that for most humans, including myself— and probably all humans—it is not possible to ascertain the answer to (Q) just via judgmental introspection; i.e., judgmental introspection is impotent as a means for answering (Q). It is important to appreciate, however, that the fact that (Q) is CC transcendent does not itself entail that (Q) cannot be answered by judgmental introspection. For, the possibility remains open that some humans might possess, or might be able to acquire, the ability to ascertain the answer to (Q) by a form of judgmental introspection that deploys an unusual degree of cognitive skill that far exceeds mere conceptual competence. A somewhat suggestive analogy here is the fact that some humans have “perfect pitch”: they can immediately and reliably ascertain, when hearing a single-note musical sound, what note it is (e.g., c-sharp). Most humans who are fully competent with the pertinent musical concepts (e.g., musical-scale concepts) cannot do this, even though the phenomenal character of a given single-note sound-experience does fully determine what note it is. Likewise, it is possible (although I think unlikely) that some humans are, or could come to be, “judgmental introspection savants” vis-à-vis questions like (Q); such people would indeed be able to spontaneously “read off ” the answer to (Q) just by introspectively attending to their agentive phenomenology while pondering the question. But as I say, I doubt that any real humans are capable of this. I turn now to the second of the above-mentioned explanatory tasks. Why might one think—or anyway, be tempted to think—that one can reliably ascertain the answer to question (Q) just via judgmental introspection, and that the answer is that claim (1) is true? I have a two-part debunking explanation to offer. (The two parts are distinct but mutually reinforcing. One could accept either part alone even if

Introspection about Phenomenal Consciousness

419

one rejected the other part.) First, the idea that one can tell introspectively that claim (1) is the correct answer to (Q) reflects a fairly subtle form of introspective confabulation. It is one thing for agentive experience to have this feature: a. Not presenting one’s behavior as causally determined by prior conditions. It is quite is another thing for agentive experience to have this distinct feature: b. Presenting one’s behavior as not causally determined by prior conditions. Feature (a) does seem directly accessible to attentive introspection, via the reliable form of judgmental introspection. (One introspectively ascertains that one’s freeagency experience lacks the causal-determination aspect that is sometimes present in other experiences—e.g., an experience of a billiard-ball collision and its aftermath.) But to think that attentive introspection directly reveals feature (b) is to be guilty of conflating the introspectable feature (a) with the distinct feature (b). Indeed, when one bears the distinction in mind, while also attentively introspecting in a suitably careful way, it seems that the following higher-order feature of agentive presentational content is itself introspectively accessible: being such that the presence or absence of feature (b) is not introspectively accessible. Turning to the second part of my proposed debunking explanation, I maintain (1) that the concept of free agency is governed by implicit, contextually variable, semantic parameters, (2) that everyday default settings on these parameters render free-agency attributions compatible with determinism, and (3) that explicitly posing the problem of freedom and determinism tends to drive the implicit parameters away from their default setting and toward a setting under which freedom attributions become incompatible with determinism (Horgan 1979; Horgan and Graham 1994). Suppose that this contextualist, compatibilism-friendly orientation is right with respect to the concept of free agency. What, then, should be said about attempts to introspectively ascertain the answer to (Q), a question pertaining to the free-agency intentional content of agentive experience? Well, when one introspectively attends to the freedom aspect of one’s agentive phenomenology, while simultaneously seeking to form a spontaneous judgment concerning question (Q), the implicit contextual parameters governing the free-agency concept are apt to be driven away from their default setting and toward an extreme, determinism-incompatible setting. Introspective inquiry directed at question (Q) is therefore prone to the following kind of subtle error. When one undertakes to answer (Q) on the basis of introspective attention to the free-agency aspect of one’s agentive experience, one’s judgmental deployment of the concept of free agency is apt to be unwittingly infected from the very start by the unnoticed changes in the contextual settings of the implicit parameters governing this concept. That is, one is apt to mistakenly think that it is introspectively self-evident that the intentional

420

Introspection and the Nature of Experience

content of one’s free-agency phenomenology is incompatible with causal determinism—when one’s tendency to think so has actually arisen because the judgmental aspect of one’s introspective endeavor has unwittingly been prodded into a nondefault parameter-setting by the posing of the very question one is seeking to answer, viz., (Q). Judgmental introspection thus gets perverted by an unnoticed context effect—even though the intentional content of agentive experience itself is probably not subject to contextually variable implicit parameters.12 Although one seems to oneself to be introspectively “reading off ” from one’s phenomenology that claim (1) is the answer to (Q), what one is really doing is forming a judgment about that phenomenology using an introspective process whose judgmental dimension has been unwittingly tripped up by a subtle context-effect. That unnoticed confusion could easily make one think that one is reliably ascertaining the answer to (Q) via the kind of judgmental introspection that deploys just attentive introspection plus conceptual competence, when in fact one is doing no such thing.

4. Conclusion The intrinsic phenomenal aspects of one’s current experience are epistemically special because they are self-presenting and are not subject to any appearance/reality gap. Some phenomenal beliefs that arise by judgmental introspection, viz., SPPB beliefs, are outright infallible. For many kinds of phenomenal belief, judgmental introspection is a super-reliable belief-forming process: it operates by exercising one’s conceptual competence with phenomenal concepts while directing one’s introspective attention to phenomenal aspects of one’s present experience, and the resulting phenomenal beliefs are subject only to labeling fallibility and not to appearance/reality fallibility. There are some purely phenomenal questions, however, that are CC transcendent and cannot be reliably answered by means of judgmental introspection; although the answers to these questions are fully determined by the intrinsic, self-presenting, phenomenal character of experience, the answers cannot be simply “read off ” from experience just by attending to it and then directly deploying one’s conceptual competence. An example of a CC transcendent question is (Q), which concerns subtle aspects of the satisfaction conditions that constitute the intentional content of agentive experience. There is little 12

One reason, inter alia, to doubt that agentive experience is subject to contextually variable implicit parameters is that it is plausible that certain nonhuman creatures (e.g., dogs) have agentive experience, whereas implicit contextual parameters seem to require cognitive sophistication that far exceeds what dogs possess and also seems intimately linked to the use of language.

Introspection about Phenomenal Consciousness

421

reason to expect to be able to answer such a question just by exercising one’s conceptual competence while attending introspectively to one’s own agentive experience, since the forte of conceptual competence is the capacity to make classificatory judgments in concrete cases, not the capacity to produce direct intuitive answers to abstract general questions about satisfaction conditions.

References Chalmers, D. 2003. “The Content and Epistemology of Phenomenal Belief.” Pp. 220– 272 in Consciousness: New Philosophical Perspectives, edited by Q. Smith and A. Jokic. Oxford: Oxford University Press. Goldman, A. 1993.“The Psychology of Folk Psychology.” Behavioral and Brain Sciences 16: 15–28. Henderson, D., and T. Horgan. 2000. “What Is A Priori and What Is It Good For?” Southern Journal of Philosophy, Spindel Conference Supplement on the Role of the Empirical and the A Priori in Philosophy 38: 51–86. ———. 2002. “The A Priori Isn’t All That It’s Cracked Up to Be, But It Is Something.” Philosophical Topics 29: 219–250. Issue honoring Alvin Goldman. ———. 2011. The Epistemological Spectrum: At the Interface of Cognitive Science and Conceptual Analysis. Oxford: Oxford University Press. Horgan, T. 1979. “‘Could’, Possible Worlds, and Moral Responsibility.” Southern Journal of Philosophy 17: 345–358. ———. 2007a. “Agentive Phenomenal Intentionality and the Limits of Introspection.” Psyche 13(2): 1–29. ———. 2007b. “Mental Causation and the Agent-Exclusion Problem.” Erkenntnis 67: 183–200. ———. Forthcoming.“From Agentive Phenomenology to Cognitive Phenomenology: A Guide for the Perplexed.” In Cognitive Phenomenology, edited by T. Bayne and M. Montague. Oxford: Oxford University Press. Horgan, T., and G. Graham. 1994. “Southern Fundamentalism and the End of Philosophy.” Philosophical Issues 5: 219–247. Reprinted pp. 271–292 in Rethinking Intuition: The Psychology of Intuition and Its Role in Philosophy, edited by M. DePaul and W. Ramsey. New York: Rowman and Littlefield, 1998. Horgan, T., and U. Kriegel. 2007. “Phenomenal Epistemology: What Is Phenomenal Consciousness That We May Know It So Well?” Philosophical Issues 17: 123–144. Horgan, T., and J. Tienson. 2002. “The Intentionality of Phenomenology and the Phenomenology of Intentionality.” Pp. 520–533 in Philosophy of Mind: Classical and Contemporary Readings, edited by D. Chalmers. Oxford: Oxford University Press. ———. 2005. “The Phenomenology of Embodied Agency.” Pp. 415–423 in A Explicacao da Interpretacao Humana: The Explanation of Human Interpretation, edited by M. Saagua

422

Introspection and the Nature of Experience

and F. de Ferro. Proceedings of the Conference Mind and Action III—May 2001. Lisbon: Edicoes Colibri. Horgan, T., J. Tienson, and G. Graham. 2003. “The Phenomenology of First-Person Agency.” Pp. 323–340 in Physicalism and Mental Causation:The Metaphysics of Mind and Action, edited by S. Walter and H. D. Heckmann. Exeter, UK: Imprint Academic. ———. 2004. “Phenomenal Intentionality and the Brain in a Vat.” Pp. 297–317 in The Externalist Challenge, edited by R. Schantz. Berlin: Walter de Gruyter. ———. 2006. “Internal-World Skepticism and the Self-Presentational Nature of Phenomenal Consciousness.” Pp. 41–61 in Self-Representational Approaches to Consciousness, edited by U. Kriegel and K. Williford. Cambridge, Mass.: MIT Press. Pitt, D. 2004. “The Phenomenology of Cognition: Or What Is It Like to Think That P?” Philosophy and Phenomenological Research 69: 1–36. Siewert, C. 1998. The Significance of Consciousness. Princeton: Princeton University Press. Strawson, G. 1994. Mental Reality. Cambridge, Mass.: MIT Press.

INDEX

access peculiar, 5–6, 16, 225–226, 260, 263, 277, 299, 307n24 privileged, 5–6, 16, 30–31, 52–57, 164, 225, 260–261, 263, 277 acquaintance, 11, 56–57, 93–125, 157, 406–407 action (cf. agency), 174, 178, 211–219, 221–227, 231–235, 277–278, 406, 413–421 Anscombe, G. E. M., 169n1 Anton’s syndrome, 80, 194n16 appearances (cf. seemings) epistemic vs. phenomenal, 56n7, 106–109 first- vs. higher-order, 136–139, 143–147, 152–154, 362–363 Armstrong, D. M., 12–13, 78n9, 132, 199, 262 attention, 11–12, 24–25, 134, 141, 145–148, 152–153, 265, 360–366, 383, 406–421 Austin, J. L., 72–75 authority, 52–55, 57–59, 212, 217, 359, 385–386 availability, 22, 88, 245–248, 254–256 belief activity vs. passivity of, 174, 178, 214, 232–235 background beliefs, 108–111, 307–308, 392–398 dispositional account of, 274–275, 298 higher-order, 242–257 regress problem for, 177–179, 248, 313 phenomenal beliefs, 406–410 SPPB phenomenal beliefs, 410–421 standing beliefs, 21, 239, 245, 251, 254–257, 270, 295–301, 308–309 blindsight, 23, 133n5, 135, 158, 179–180, 194n17, 271–273 “inner blindsight,” 145 “superblindsight,” 245n5, 271–272, 281 Block, Ned, 20, 245, 271 Boër, Steven, 54n6 BonJour, Laurence, 112–114 Boyle, Matthew, 231n20 broad perceptual model, 224–225, 232n21, 240, 271n39 Broome, John, 80n12

Burge, Tyler, 142n8, 252n3, 363, 377n21 Byrne, Alex, 13–15, 69, 224–227, 240–243, 304n12, 306n17, 307 Carruthers, Peter, 130n3, 138n6, 146n10, 154–155 Chalmers, David, 106, 267n12, 269n17 Chomsky, Noam, 20, 270 clairvoyant case, 280, 285, 290 cognitive phenomenology, 276–277, 415n9 concepts conceptual competence, 160–162, 178–180, 409–421 conceptual poverty, 265–266 dispositionalism about, 116–123 phenomenal, 25, 116–124, 409n3 recognitional, 155–159 consciousness, 3, 19–23, 297, 323–324 access, 20–22, 82, 245, 271–300 higher-order perception theories of , 130n3, 138n6, 157n21 higher-order thought theories of, 20–21, 51, 130n3, 138n6, 143, 272 phenomenal, 20–23, 82, 131, 265, 270–275 self-presentational theories of, 11, 25, 102–104, 406–409, 412–421 constitutivism, 16–19, 245–257, 264–269, 311–323 content externalism, 32, 121–124, 278, 355n4, 414 contextualism, 186, 334n10, 419–420 Crane, Tim, 384, 387 Davidson, Donald, 54n5 defeasibility, 80–81, 106–107, 114–115, 161, 190, 261–262, 302–303 demonstratives, 148–152, 161, 267 perceptual vs. introspective, 11, 104–107, 116–120 Dennett, Daniel, 162–163 Descartes, René, 20, 30–31, 109, 296–297, 308, 329n1, 363 Difference Thesis, 4–6 discrimination, 24, 133, 158, 330–348, 375–378 disjunctivism, 23–24, 68, 353–378, 382–383

424

Index

Donnellan, Keith, 104–105 Dretske, Fred, 8–9, 67, 71, 84–85, 184–187, 217n6 emotion, 35–37, 42 epistemic rules, 13–15, 189–192, 225–231, 240, 243, 307 Evans, Gareth, 187–188, 190n14, 203–205, 299–302, 362–364 evidence, 80–81, 85–88, 186–195, 225–231, 256, 274–275, 306, 382–384, 399–401 Evidence Argument, 9, 65–88 fallibility (cf. infallibility), 19, 22, 159–160, 262, 267, 278, 295–299, 318–323, 370–371, 406, 407–410, 420 Feldman, Richard, 112–114 Fish, William, 368n18 Fodor, Jerry, 7, 206 fraternity initiation case, 108–110, 146n10, 265, 268, 410 Freud, Sigmund, 20, 81 Fumerton, Richard, 112–114 functionalism, 43, 271–273 Gertler, Brie, 11, 157n20 Gettier case, 203, 261–263, 304, 417 Glüer, Kathrin, 206n38 Greenough, Patrick, 19 Harman, Gilbert, 382–383 Hohwy, Jakob, 36 Horgan, Terry, 11, 25, 111, 267n12, 413n8 Huemer, Michael, 282n40 Hume, David, 201–202, 356–358, 385 Husserl, Edmund, 154n17, 157n21 illusion, 23–24, 145–146, 159, 215, 262–278, 353–358, 365–367, 372–388 café wall illusion, 202 imagery, 33–39, 201–202 imagination, 70, 276n32 indexicals, 65n8, 97n4, 149–151, 410–412 introspection inferential theories of, 13–15, 31, 86, 154–155, 188–193, 207, 227n15, 231–232, 304, 330n3 inner sense theories of , 10–12, 16, 59, 100, 129–165, 224, 240, 261, 282, 299, 305, 322–323 module for, 5–7, 205 pluralist theories of, 7–8, 29–32, 40–43, 330n3 primitivism, 15–16, 280, 283 reductionism, 16, 280, 283

Simple Theory of, 16, 259–290 skepticism about, 6–10, 50–62 subject matter, 6, 19–21 259–262, 277–278 judgment, 32, 29–44, 160–163, 273–277, 302–309, 323–324 conscious vs. non-conscious, 253–256, 295, 298 empirical, 99–102, 112–115, 124 introspective, 32–35, 102–107, 358–378, 399–401, 408–421 truthmakers of, 96–98, 103–115, 121, 124, 157n20, 175, 177 justification, 98–99, 100–101, 112–115, 285–290, 295–309 doxastic vs. epistemic, 264–266, 301, 341n24 externalism, 227, 284 foundationalism, 112–113, 124, 178–179, 289 immediate vs. mediate, 19, 278n34, 298, 308–311, 321–322 internalism, 18, 99, 263, 283–290 introspective, 13, 18–22, 177, 261–290, 298–323, 339–344 latitudinarianism, 175 mentalism, 99, 279, 288–289 regress problem for, 177–180 Kemmerling, Andreas, 57–58 Kriegel, Uriah, 111, 157n21 Loar, Barry, 97, 155 Locke, John, 265n9, 329n1 lottery case, 334, 339n22 Ludlow, Peter, 123 luminosity, 19, 260n1, 331–348 anti-luminosity, 313–316, 331–348 margin for error principle, 315–316, 331–348 Lycan, William, 54n6, 130n3, 146n10, 147–148, 153–154, 393 Martin, Michael, 24, 67n2, 253, 353–378, 383 McGinn, Colin, 244 memory, 38–41, 193, 201–202, 254 Moore’s paradox, 18, 239–241, 249–253, 285–286, 302–303, 320n39 Moran, Richard, 14–15, 23, 61n14, 244, 246–247, 305n16, Müller-Lyer illusion, 12, 170–172, 205–206 naive realism, 23–24, 355–378 Neta, Ram, 261n2 Nichols, Shaun, 30, 39 NICS cases, 254–255, 270n21

Index omniscience, 18–19, 268, 284, 289–290, 307, 343–345 pain, 96–97, 104, 119, 136, 142, 149, 151n14, 174, 195, 214–219, 260, 265, 282–283, 295n1, 321n41, 330, 344 Peacocke, Christopher, 15–16, 57n10, 61, 248–250, 253–256, 268n16, 270n21, 274–276 perception belief-dependence of, 205–207 diaphaneity of, 69 factivity of, 181–182, 374n20 perceptual constancy, 133, 140–148, 370, 396–398 personal vs. subpersonal levels, 5–7, 41, 179, 244 phenomenal adequacy, 24–25, 383–384, 400 phenomenal consciousness see consciousness, phenomenal phenomenal conservatism, 282–283 Pitt, David, 273, 276–277 Price, H. H., 206 Prinz, Jesse, 31, 39 proprioception, 33, 36–37, 132, 134–136, 141, 144, 186–187, 277–278, 398–399 Pryor, Jim, 9n6, 86n16, 88

425

self-blindness, 77n8, 194n17, 241–243, 262n3 self-deception, 17, 239, 245, 248–252 self-intimation, 241, 244–257, 261, 312–318, 322, 341–342 Shah, Nishi, 220–224, 303n11 Shoemaker, Sydney, 18, 22, 143n9, 224–225, 268n14, 302n10, 317n35, 320n39, 342n27, 344n31, 361–363 Siegel, Susanna, 355n5, 374–375, 385n4 Siewert, Charles, 11–12, 364n13 Silins, Nico, 14, 18–19, 22, 245n4, 247n6, 279 skeptical hypotheses, 109–110, 185–187, 310–311 Smithies, Declan, 16, 18, 22, 113n20, 151n14, 311n24, 341n24 sorites paradox, 313–314, 331–332, 336n14, 340–347 Sosa, Ernest, 12–13, 23, 117–118, 213n4 speckled hen problem, 13, 103, 107–111, 176n7, 178, 265n9 Spener, Maja, 24–25, 366n15 Stalnaker, Robert, 94, 120–124 Stich, Stephen, 39, 270n19 Stoljar, Daniel, 9–10, 18, 109 Strawson, P. F., 381, 387 subdoxastic states, 270–277

qualia, 57, 132–133, 184, 383, 387 rationality, 12–13, 14–15 17–19, 57–58, 76–88, 109n38, 174, 177–181, 206n38, 227–232, 244, 256–257, 268, 289–290, 376 Raz, Joseph, 219n8 reflection (cf. critical reflection), 30, 43, 143–146, 151–165, 178–181, 286–288, 355–360, 371–372 reliabilism, 12–13, 51n3, 280–281 reliability, 12–13, 51, 100, 117, 176–179, 269, 281, 284, 287, 331–334, 344 super-reliability, 407–412, 420 representation higher-order, 130n3, 147, 157, 272, 407–408 representationalism, 23–24, 353–378, 382–383 self-representation, 138, 157–158, 164 Rorty, Richard, 271 Rosenthal, David, 21n12, 133 Russell, Bertrand, 94–100, 124–125 Ryle, Gilbert, 6, 21, 183, 186–188, 195–196, 265n9 safety, 203, 304, 314–316, 332–339 Schwitzgebel, Eric, 5, 7–8, 21, 100–101, 106–107, 151n14, 269n18 Searle, John, 196–197, 274n29, 310n22 seeming see appearance self-ascription, 30–31, 43, 188, 239–257, 265–267, 305, 310

Titchener, E. B., 35 transparency, 15, 18–19, 186, 211, 224–235, 297–310, 323–324, 382–383, 388–390 Twin Earth, 122–123, 279, 417 Tye, Michael, 153n16, 164n24 Velleman, David, 220–224, 303n11 veridicality, 23–24, 180, 278, 353–378, 418 virtue theory, 12–13, 23, 169n1 vision, 194–205 visual experience, 32–39, 103, 115, 132–159, 164n24, 174–182, 188, 196–199, 273, 310n22, 353–378, 381–401 visual objects, 14, 23–24, 32–35, 95, 137–158, 194–199, 354–366, 372–375 mind-independence of, 23–24, 187, 354–355, 372, 384–385, 387–401 voluntarism, 213, 234–235 Williamson, Timothy, 19, 81n14, 260n1, 263n6, 313–316, 330–340 Williford, Kenneth, 157n21 Wittgenstein, Ludwig, 115, 185n5, 223n11, 250, 370 Zagzebski, Linda, 263n5 zany argument, 18, 241–243 Zimmerman, Aaron, 23–24, 309n20, 318n38 zombies, 8–9, 65, 67, 184–187, 204n17, 405–406