Intellectual Assurance: Essays on Traditional Epistemic Internalism 2015944559, 9780198719632

378 104 2MB

English Pages [274] Year 2016

Report DMCA / Copyright

DOWNLOAD FILE

Polecaj historie

Intellectual Assurance: Essays on Traditional Epistemic Internalism
 2015944559, 9780198719632

Table of contents :
Cover
Intellectual Assurance: Essays on Traditional Epistemic Internalism
Copyright
Dedication
Contents
Notes on Contributors
Acknowledgments
Traditional Internalism: An Introduction
1 The Aims of Traditional Internalism
1.1 Evidentialism
1.2 Conservatism
1.3 Traditional internalism
2 The Tenets of Traditional Internalism
2.1 Non-inferentially justified belief
2.2 Inferentially justified belief
2.3 Skepticism
3 The Problems for Traditional Internalism
3.1 Non-inferentially justified belief
3.2 Inferentially justified belief
3.3 Skepticism
4 Conclusion
References
Chapter Abstracts
Chapter 1: Confrontation Foundationalism, Peter Markie
Chapter 2: Acquaintance and Fallible Non-Inferential Justification, Chris Tucker
Chapter 3: Foundational Justification, Meta-Justification, and Fumertonian Acquaintance, Matthias Steup
Chapter 4: Staying Indoors: How Phenomenal Dogmatism Solves the Skeptical Problem without Going Externalist, Berit Brogaard
Chapter 5: Experience and Evidence Abridged, Susanna Schellenberg
Chapter 6: Principles of Inferential Justification, Trent Dougherty
Chapter 7: Inferential Appearances, Michael Huemer
Chapter 8: The Costs of Demon-Proof Justification, Sanford C. Goldberg
Chapter 9: Acquaintance and Skepticism about the Past, Ted Poston
Chapter 10: On Metaepistemological Scepticism, Duncan Pritchard and Christopher Ranalli
Chapter 11: How Our Knowledge Squares with Skeptical Intuitions Despite the Circle, Ernest Sosa
Chapter 12: The Prospects for Traditional Internalism, Richard Fumerton
PART I: Traditional Internalism and Non-Inferentially Justified Belief
Direct Acquaintance
Chapter 1: Confrontation Foundationalism
1 Confrontational Foundationalism: Classical and Neoclassical
2 The Limits of Acquaintance
3 Acquaintance and Identification Ability
4 Alternatives
5 Conclusion
References
Chapter 2: Acquaintance and Fallible Non-Inferential Justification
1 Introduction
2 Fumerton’s Account of Non-Inferential Justification
3 A Fixable Problem
4 Acquaintance with Contradicting Facts
5 Fallible NIJ without Contradicting Facts?
6 Hasan to the Rescue?
7 Conclusion
References
Chapter 3: Foundational Justification, Meta-Justification, and Fumertonian Acquaintance
1 Internalism vs. Externalism and the Expectation of Non-Accidentality
2 BonJour’s Anti-Foundationalist Argument
3 Two Foundationalist Responses to BonJour’s Argument
4 Two Conceptions of Foundational Justification
5 Fumerton’s Principle of Inferential Justification
6 Fumertonian Non-Inferential Justification
7 Three Versions of Foundationalism
8 Does Foundational Justification Require Acquaintance?
References
Perceptual Belief
Chapter 4: Staying Indoors: How Phenomenal Dogmatism Solves the Skeptical Problem without Going Externalist
1 The Threat of Skepticism
2 Externalism
3 Naive Realism and Disjunctivism
4 Why Naive Realism Does Not Solve the Skeptical Problem
5 Phenomenal Dogmatism
6 Bootstrapping
7 Conclusion
References
Chapter 5: Experience and Evidence Abridged
Prologue to ‘Experience and Evidence Abridged’
1 Perceptual Evidence and Introspective Evidence
2 The Phenomenal Evidence Argument
2.1 Premiss 2: sensory states and phenomenal evidence
2.2 Premiss 2a: sensory states and perceptual capacities
2.3 Premiss 2b: phenomenal evidence and systematic linkage
2.4 Coda
References
PART II: Traditional Internalism and Inferentially Justified Belief
Chapter 6: Principles of Inferential Justification
1 Fumerton’s PIJ
1.1 Fumerton’s case for PIJ
1.2 An alternative explanation
2 Huemer’s PIJ*
2.1 Huemer’s critique of Fumerton
2.2 Huemer’s account of the Astrologer case
2.3 Huemer’s own account flawed
3 Chisholming Away
3.1 A sketch of a theory of knowledge
3.1.1 PERSPECTIVAL JUSTIFICATION
3.1.2 A NOTE ON SEEMINGS AND THE SUBCONSCIOUS
3.1.3 KNOWLEDGE
3.2 On ‘inference’
3.3 ‘Inferential’ justification in light of the above
4 Conclusion
References
Chapter 7: Inferential Appearances
1 Justificatory Dependence
2 Conditions for an Adequate Theory of Inferential Justification
3 An Appearance-Based Theory of Inferential Justification
3.1 Inferential appearances
3.2 Conditions for explicit inferential justification
4 The Inferential Appearance Theory Satisfies the Conditions for a Theory of Inferential Justification
5 Revisiting Fumerton: Acquaintance vs. Appearance
6 Inexplicit Justificatory Dependence
7 Conclusion
References
PART III: Traditional Internalism and Skepticism
Responding to the Skeptic
Chapter 8: The Costs of Demon-Proof Justification
1
2
3
4
References
Chapter 9: Acquaintance and Skepticism about the Past
1 Skepticism about the Past
2 Fumerton’s Acquaintance Theory and Skepticism about the Past
2.1 Fumerton’s acquaintance theory
2.2 Fumerton’s replies to skepticism about the past
2.2.1 REPLY 1: SELF-REFUTATION
2.2.2 REPLY 2: ACQUAINTANCE WITH THE RELATION OF MAKING-PROBABLE
2.3 Difficulties with Fumerton’s response to memory skepticism
2.3.1 THE PHENOMENOLOGICAL PROBLEM
2.3.2 THE PROBLEM OF BACKGROUND EVIDENCE
2.3.3 THE CONTENT PROBLEM
3 Epistemic Conservatism and Skepticism about the Past
4 Conclusion
References
Skepticism and Circularity
Chapter 10: On Metaepistemological Scepticism
1 Introduction
2 Fumerton on Metaepistemology and Scepticism
3 Stroud’s Metaepistemological Scepticism
4 Fumerton and Stroud
References
Chapter 11: How Our Knowledge Squares with Skeptical Intuitions Despite the Circle
1 Preliminaries on Skepticism in General
1.1 Varieties of skepticism
1.2 Skepticism, sensitivity, and safety
1.3 Why the dream scenario is special
1.4 Skepticism and circularity
2 A More Specific Disagreement on Philosophical Skepticism
3 Conclusion
References
Afterword
Chapter 12: The Prospects for Traditional Internalism
1 Introduction
2 Noninferential Justification and Acquaintance
2.1 Acquaintance and mental states
2.2 Acquaintance and seemings
2.3 Speckled hens and acquaintance
2.4 Noninferential justification and assurance
3 Inferential Justification
4 Prospects for Traditional Internalism
4.1 Traditional internalism and skepticism
4.2 Will we find the future of traditional internalism by thinking about the past?
References
Index

Citation preview

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

Intellectual Assurance

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

Intellectual Assurance Essays on Traditional Epistemic Internalism

EDITED BY

Brett Coppenger and Michael Bergmann

1

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

3

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © the several contributors 2016 The moral rights of the authors have been asserted First Edition published in 2016 Impression: 1 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, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2015944559 ISBN 978–0–19–871963–2 Printed in Great Britain by Clays Ltd, St Ives plc Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

For Richard Fumerton: A model of intellectual honesty, philosophical rigor, and independence of mind

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

Contents Notes on Contributors Acknowledgments

ix xi

Traditional Internalism: An Introduction Brett Coppenger

1

Chapter Abstracts

15

Part I. Traditional Internalism and Non-Inferentially Justified Belief Direct Acquaintance 1. Confrontation Foundationalism Peter Markie

25

2. Acquaintance and Fallible Non-Inferential Justification Chris Tucker

43

3. Foundational Justification, Meta-Justification, and Fumertonian Acquaintance Matthias Steup

61

Perceptual Belief 4. Staying Indoors: How Phenomenal Dogmatism Solves the Skeptical Problem without Going Externalist Berit Brogaard 5. Experience and Evidence Abridged Susanna Schellenberg

85 105

Part II. Traditional Internalism and Inferentially Justified Belief 6. Principles of Inferential Justification Trent Dougherty

127

7. Inferential Appearances Michael Huemer

144

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

viii

CONTENTS

Part III. Traditional Internalism and Skepticism Responding to the Skeptic 8. The Costs of Demon-Proof Justification Sanford C. Goldberg

165

9. Acquaintance and Skepticism about the Past Ted Poston

183

Skepticism and Circularity 10. On Metaepistemological Scepticism Duncan Pritchard and Christopher Ranalli 11. How Our Knowledge Squares with Skeptical Intuitions Despite the Circle Ernest Sosa

205

224

Afterword 12. The Prospects for Traditional Internalism Richard Fumerton

239

Index

259

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

Notes on Contributors M ICHAEL B ERGMANN is Professor of Philosophy at Purdue University. B ERIT B ROGAARD is Professor of Philosophy at the University of Miami and Professor II at the University of Oslo. B RETT C OPPENGER is Assistant Professor of Philosophy at Tuskegee University. T RENT D OUGHERTY is Associate Professor of Philosophy at Baylor University. R ICHARD F UMERTON is F. Wendell Miller Professor of Philosophy at the University of Iowa. S ANFORD C. G OLDBERG is Professor of Philosophy at Northwestern University. M ICHAEL H UEMER is Professor of Philosophy at the University of Colorado. P ETER M ARKIE is Curators’ Teaching Professor of Philosophy at the University of Missouri. T ED P OSTON is Professor of Philosophy at the University of South Alabama. D UNCAN P RITCHARD is Professor of Philosophy at the University of Edinburgh. C HRISTOPHER R ANALLI is a Postdoctoral Research Fellow at the Universidad Nacional Autónoma de México. S USANNA S CHELLENBERG is Associate Professor of Philosophy at Rutgers University. E RNEST S OSA is Board of Governors Professor of Philosophy at Rutgers University. M ATTHIAS S TEUP is Professor of Philosophy at Purdue University. C HRIS T UCKER is Assistant Professor of Philosophy at the College of William and Mary.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

Acknowledgments We are grateful to Peter Momtchiloff at Oxford University Press and to each of the volume contributors—especially Trent Dougherty, Richard Fumerton, and Ted Poston—for their helpful input at various stages along the way in preparing this volume for publication.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

Traditional Internalism An Introduction Brett Coppenger

Ordinarily, people take themselves to know a lot. I know where I was born, I know that I have two hands, I know that two plus two equals four, and I also think I know a lot of other stuff too. However, the project of trying to provide a philosophically satisfying account of knowledge, one that holds up against skeptical challenges, has proven surprisingly difficult. Either one aims for an account of justification (and knowledge) that is epistemologically demanding, in an effort to offer an account that satisfactorily addresses skepticism, or one aims for an account of justification (and knowledge) that makes sense of our ordinary knowledge claims. However, the history of contemporary epistemology tells us that you cannot have both: the former results in skepticism, the latter in an unsatisfying response to skepticism. What we find, in the array of contemporary attempts to give accounts of knowledge and justification, are numerous views spread across the internalism/externalism spectrum that deal with the dilemma of the previous paragraph in different ways. Most of them are guided by the goal of accommodating our ordinary knowledge claims. This is especially true of externalist accounts, but many internalist accounts are guided by this same goal, though perhaps to a lesser degree. One kind of internalist view stands out for its insistence on providing philosophically satisfying accounts of knowledge and justification, even if doing so has skeptical implications. This is the traditional, old-fashioned, Cartesian-style internalism that was so prominent in the early 20th century and is now a minority position. Unlike competing versions of epistemic internalism, the guiding principle of traditional, Cartesian-style internalism (what I will henceforth call ‘traditional internalism’) is not to accommodate our commonsense views about the rationality of our ordinary beliefs. Instead, traditional internalism emphasizes rationality’s demand for philosophical assurance, on the basis of evidence that can withstand the strongest skeptical challenges, that our ordinary beliefs (perceptual and otherwise) are true. According to the traditional internalist, the philosopher, qua philosopher, ought

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



BRETT COPPENGER

to begin the epistemological project from the inside, placing a premium on satisfying our philosophical curiosity. Despite the relative unpopularity of traditional internalism the view can be taken to be worthy of attention for a variety of reasons. First, traditional internalism has great historical importance. Significant portions of the history of contemporary epistemology, and to a lesser degree philosophy in the modern era (roughly, Descartes to Kant), are grounded in a number of the intuitions that drive traditional internalism. Second, and perhaps even more importantly, traditional internalism serves as the source of a wide variety of criticisms for other, more prominent, contemporary epistemological views. In a number of debates the views of the traditional internalist are used to play a kind of Devil’s advocate. Finally, some of the ideas that ground traditional internalism have had a bit of resurgence of late.1 To put it simply, traditional internalism refuses to go away. In this introduction, I will do three things. First, I will situate traditional internalism among other competing versions of internalism by highlighting the ways in which traditional internalism differs from them (in particular, evidentialism and conservatism). Second, I will explain more carefully some of the central tenets of traditional internalism and what motivates them. Third, I will highlight some of the difficulties that threaten traditional internalism, and I will explain how the contributions to this volume interact with those difficulties.2 The goal of this volume is to test again the staying power of traditional internalism, to see if this once historically prominent view deserves another look, to see if traditional internalism is a legitimate contender providing useful criticisms of more prominent views, or, if instead it is time for traditional internalism to be left by the wayside.

1 The Aims of Traditional Internalism The importance of the internalism/externalism controversy in contemporary epistemology is hard to overstate. And, as is the case with so many other central debates in philosophy, there does not seem to be a universally accepted way of distinguishing these different views. Instead, there are a number of different ways of carving up the debate. The different attempts to carve up the internalist/externalist controversy can help us to see that there are different kinds of internalism. These different kinds of internalism can in some cases be compatible with one another, but they also helpfully 1

See note 4 and Section 2 for a discussion of some of these ideas. Early versions of many of the chapters written for this volume (those by Huemer, Poston, Pritchard and Ranalli, Schellenberg, Steup, and Tucker) were presented at the 2014 Orange Beach Epistemology Workshop. A schedule for that event, along with a paper presented at the workshop by Richard Foley (2014) in honor of Richard Fumerton’s work in philosophy, can be found at: . 2

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

TRADITIONAL INTERNALISM : AN INTRODUCTION



illustrate the different fundamental aims of the internalists engaged in the various versions of the internalism/externalism controversy. Two of the most prominent contemporary internalist alternatives to traditional internalism are evidentialism and conservatism.3

1.1 Evidentialism The evidentialist maintains that justification is a matter of believing in accordance with one’s evidence. On this view, my belief that I have a cup of coffee on the table before me is justified only if that belief accords with the evidence that I possess. Conee and Feldman, two leading evidentialists, argue that, “the epistemic justification of a belief is determined by the quality of the believer’s evidence for the belief ” (2004: 83). They have famously defended an account of what they call evidentialist justification: Doxastic attitude D toward proposition p is epistemically justified for S at t if and only if having D toward p fits the evidence S has at t. (2004: 83)

Proponents of evidentialism sometimes go on to argue that what counts as evidence for a subject is that subject’s mental states. Thus, we end up with a version of mentalism saying that, “If any two possible individuals are exactly alike mentally, then they are alike justificationally, e.g., the same beliefs are justified for them to the same extent” (2004: 57). This kind of internalism appears to capture our commonsense intuitions about the similar justificatory status of two subjects who share all the same evidential states. In addition, by focusing attention on the internal mental states of the subject, the name ‘internalism’ makes perfect sense. Moreover, according to this view, it’s plausible to think that most of the beliefs we tend to think of as justified are justified.

1.2 Conservatism The conservatist maintains that there is an intimate connection between how things seem to a subject and the justified beliefs of that subject. On this view, I believe that I have a cup of coffee on the table before me because that visually seems to me to be the case; and that belief is justified because it is based on that visual seeming, with the same propositional content. Huemer has famously argued that, “when we form beliefs, our beliefs are based on the way things seem to us” (2007: 39). Huemer contends that beliefs based on seemings can be justified, and he defends phenomenal conservatism: If it seems to S that p, then, in the absence of defeaters, S thereby has at least some degree of justification for believing that p. (2007: 30) 3 For a discussion of evidentialism see Dougherty (2011), and for a discussion of conservatism see Tucker (2013). The present volume, taken together with the Dougherty and Tucker volumes, represent contemporary defenses of the most prominent versions of epistemic internalism.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



BRETT COPPENGER

Because seemings are a kind of internal state, conservatism, like evidentialism, fits nicely into the category of internalism. Conservatism aims at making sense of our commonsense assumption that most of our ordinary beliefs are justified. The goal for the conservatist is to defend and elucidate the epistemic connection between seemings and justified beliefs.

1.3 Traditional internalism What distinguishes traditional internalism from these other kinds of internalism? Richard Fumerton, who is perhaps the most prominent and systematic contemporary defender of traditional internalism, contends that, [T]he fundamental internalist concern is that having knowledge or justification in the externalist sense doesn’t seem to satisfy philosophical curiosity. It doesn’t seem to provide any assurance of the sort the philosopher seeks when wondering about the truth of various propositions. (2004: 75)

Thus, according to Fumerton, the essential feature of internalism is not its claim that justification depends on the internal states of the believer, nor any interest in accommodating our commonsense assumptions about the justification of our ordinary beliefs. Instead, what characterizes traditional internalism is the aim of emphasizing the need for philosophical assurance, on the basis of evidence that can withstand the strongest skeptical challenges, that our ordinary beliefs are true. And the key to obtaining this assurance is to be directly acquainted or confronted with facts to be known. According to the traditional internalist, the only way one could achieve this kind of assurance, this kind of robust answer to the skeptic, would be by seeing the connection of one’s belief to the truth. It is not enough that one has a true belief that happens to be true; what one needs in order to be justified is some way of directly seeing that one’s belief is true. On this kind of reading, traditional internalism can be traced back to Descartes and his search for indubitable belief from the first person perspective. What Descartes wanted was a belief that he could be confident in, a belief that rested on an indubitable foundation that left no room for error. Descartes’ project was one of trying to secure philosophical assurance that skepticism was wrong, and it was only in light of this assurance that Descartes could proceed with peace of mind. Of course, one cannot tell the story of Descartes’ traditional internalism without also mentioning the received view that Descartes’ search for an indubitable foundation for our beliefs failed. While Descartes believed he was able to find his indubitable foundation and avoid skepticism in the process, many of his critics thought the Cartesian starting point led to a skeptical outcome. Traditional internalism, it is argued, sets the standard for justified belief too high, and as a result, skepticism is inevitable. Traditional internalism failed to deliver on its primary goal of giving an adequate and philosophically interesting response to skepticism.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

TRADITIONAL INTERNALISM : AN INTRODUCTION



Despite the fact that many philosophers have written off traditional internalism, in the past several decades there has been new work in epistemology that has led to a kind of revival of the view.4 The aim of this volume is to pay special attention to traditional internalism in its contemporary forms, in an effort to investigate how well this view might be defended.

2 The Tenets of Traditional Internalism If one has the traditional internalist’s aim of successfully connecting belief with assurance of its truth, one can certainly understand Descartes’ interest in a foundation of certainty coupled with inferential beliefs that depend on a clearly valid deduction from that foundation. Contemporary defenses of traditional internalism share Descartes’ interest in finding a solid foundation on which to build. The contemporary traditional internalist’s account of justification, like so many other foundationalist accounts, is recursive. According to it, a complete account of justification requires both a base clause about non-inferential belief and a recursive clause about inferential belief.

2.1 Non-inferentially justified belief For the sake of specificity, let’s focus on the views of one of the more prominent contemporary defenders of traditional internalism, Richard Fumerton. Fumerton’s account of non-inferential justification depends on the relation of acquaintance. On this view, acquaintance is a sui generis relation that holds between a self and a thing, a property, or a fact. According to Fumerton: [O]ne has a noninferentially justified belief that P when . . . 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. (1995: 75)

If a subject S is acquainted with (1) the fact that P, (2) the thought that P, and (3) the relation of correspondence holding between the fact that P and the thought that P, then S has a non-inferentially justified belief that P. It is the relation of acquaintance that is doing the heavy epistemic lifting. The acquaintance theorist is convinced that with respect to some truths, one can be directly aware of both the truth-bearer and the truth-maker and the correspondence that defines truth. (Fumerton, 2001: 70)

Fumerton’s account of non-inferential justification attempts to ground justification in a direct confrontation with reality (1995: 83). When a subject has the direct acquaintance that Fumerton requires for non-inferentially justified belief she has 4

Other contemporary defenders of traditional internalism, besides Fumerton, include Laurence BonJour (see BonJour, 2003), Evan Fales (see Fales, 1996), and Timothy McGrew (see McGrew, 1995).

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



BRETT COPPENGER

everything that she could possibly want, from an epistemic perspective, with regard to the truth of that thought. Fumerton contends that: The very source of justification includes that which makes true the belief. In a way this is the idea that makes an acquaintance foundation theory so attractive. I have no need to turn to other beliefs to justify my belief that I am in pain because the very fact that makes the belief true is unproblematically before consciousness, as is the correspondence that holds between my thought and the fact. Again, everything one could possibly want or need by way of justification is there in consciousness. (2001: 15)

While the present discussion of traditional internalism has focused on the way that Fumerton has developed and defended the view, it would be misleading to characterize traditional internalism in such a way that Fumerton is the only contemporary advocate of the view. The fundamental feature of traditional internalism, as I have presented it, is the goal of grounding justification in direct confrontation with reality in a way that allows for a philosophically satisfying account of justification. A number of contemporary advocates of this kind of view each try to develop the account of direct confrontation with reality along similar lines.5 According to BonJour, for example, a subject can be aware of some experience, and while attending to that experience she can believe that the experience has some feature, and crucially, this allows her to apprehend the agreement between the experience and the content of the belief about that experience. Thus, “the meta-belief is a description of the very content involved in the constitutive awareness of content, so that by consciously having the constitutive awareness, I am in an ideal position to judge whether or not the description is true” (BonJour, 2003: 64).

2.2 Inferentially justified belief Of course, non-inferential justification is only part of the story. The other part is inferential justification. It is commonly acknowledged that if one is to get justification from some other belief, that other belief must itself be justified. But is this enough? Suppose that I believe that I am going to have a long and rewarding life. When friends ask why I believe that I am going to have a long and rewarding life, I tell them it is because I consulted a Magic 8 Ball and that’s what it reported.6 It may very well be true that I have excellent justification for my belief that a Magic 8 Ball indicated that I am going to have a long and rewarding life, but something still seems to be missing. My friends might reasonably inquire further, “Why think that a Magic 8 Ball’s saying that you will have a long and rewarding life makes it all likely that you 5

See note 4 for a list of contemporary advocates of traditional internalism. The ‘Magic 8 Ball’ was a toy that would answer questions. It consisted of an opaque ball with a small window. One would ask the 8 Ball a question, shake the ball, and then look into the window at a 20-sided die that would display an answer to the question (e.g. Question: Does Fumerton’s account of traditional internalism avoid skepticism? Answer: Signs point to ‘no’). 6

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

TRADITIONAL INTERNALISM : AN INTRODUCTION



will?” The problem with this particular effort at inferential justification is not that my belief is based on an unjustified belief, but that I have no reason for thinking that the truth of the belief that is my basis would make likely the belief it is supposed to justify. Looking again at the work of Fumerton for an example of the traditional internalist perspective, we find him saying that inferential justification must be in accord with the following principle of inferential justification (PIJ): To be justified in believing one proposition P on the basis of another proposition E, one must be (1) justified in believing E and (2) justified in believing that E makes probable P. (1995: 36)

When one’s inferential belief satisfies both clauses of PIJ one is justified in believing one’s evidence, and one is also justified in believing that the evidence makes one’s inferred belief likely to be true. The idea is that with these things in place, one thereby obtains (in the inferential case just as in the non-inferential case) everything that one could possibly want in terms of a guarantee that one’s belief is true.

2.3 Skepticism According to the accounts of non-inferential and inferential justification offered in sections 2.1 and 2.2, when one has a justified belief, one has assurance of the truth of that belief. It’s true that fewer beliefs will count as justified by the traditional internalist’s standards. But, when the traditional internalist engages a skeptical interlocutor about those beliefs that do satisfy those standards, the traditional internalist will have the resources to answer the ‘why?’ questions in a philosophically satisfying way.

3 The Problems for Traditional Internalism If the traditional internalist’s account of justification successfully answered skepticism, it would be hard to ignore the apparent advantages of the view. However, traditional internalism fell on hard times for a reason, and the challenges that faced Descartes’ version of traditional internalism can still be seen as threatening contemporary attempts to vindicate the view.

3.1 Non-inferentially justified belief To say that the account of non-inferentially justified belief required by Fumerton’s account of traditional internalism is controversial would be an understatement. Section 2.1 was aimed at motivating traditional internalism on the basis of its promise to deliver a philosophically satisfying account of non-inferential justification. However, as Fumerton is well aware, making a promise, and keeping a promise, are two very different things. The account of traditional internalism that Fumerton defends has been criticized as offering requirements for an account of non-inferential justification that are neither necessary nor sufficient for the kind of justification that interests the traditional internalist.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



BRETT COPPENGER

Following Bergmann, one can see why someone might think that Fumerton’s requirements on non-inferential justification are not sufficient because, even if one were to grant that one could satisfy each of the three criteria required on the account that Fumerton defends (i.e., one is acquainted with one’s thought that P, the fact that P, and the relation of correspondence holding between the thought that P and the fact that P), we could still imagine the following example of Jack: Suppose Jack is being appeared to redly and that he believes that he is being appeared to redly (call this belief ‘B2’) . . . He can be directly acquainted with the relation of correspondence holding between his thought that he is being appeared to redly and the fact that he is being appeared to redly even if he has no idea that the relation of correspondence holds between these two items. Thus, Jack’s belief that B2 can satisfy [the acquaintance theorist’s requirements] even if he conceives of his being appeared to redly as no more relevant to B2 than is the mild pain in his left knee. (Bergmann 2006: 28–30)

Given the possibility of this kind of scenario, Bergmann concludes that, “It is, therefore, exceedingly difficult to see how these direct acquaintances improve things from Jack’s subjective perspective” (2006: 30). If Bergmann is right, and the kind of scenario that Jack finds himself in can obtain, then it seems as if the traditional internalist has failed to deliver on the promise of philosophical assurance, or, as Fumerton puts it, providing everything that a subject could need or want by way of providing justification. Even if one were to grant that the conditions on non-inferential justification that Fumerton requires are sufficient for justification, one might still wonder whether those conditions are necessary. As Poston has pointed out (see Poston 2008) Fumerton allows for the possibility of non-inferentially justified but false beliefs. According to Fumerton, I might be noninferentially justified in believing that I seem to see twenty-five rectangular spots in my visual field by virtue of my direct acquaintance with what turns out to be twenty-six rectangular spots in my visual field. (2001: 74)

Thus, according to Poston, Fumerton has a difficult question to answer: “is acquaintance with the truth-maker of one’s belief a necessary condition for noninferential justification?” (2008: 371). In addition to the kind of problem that Poston raises for Fumerton, there are also related problems that arise in connection with the recently revived Problem of the Speckled Hen (see Chisholm, 1942). Consider an ordinary observation of a manyspeckled hen. The traditional internalist seems committed to maintaining that one is acquainted with the experience resulting from one’s observation of some determinate number of speckles (suppose the experience is, in fact, of 37 speckles). However, in the case of a subject’s observation of that determinate number of speckles, one might reasonably wonder whether the subject has everything that one could need or want by way of providing justification for the belief that the number of speckles is 37. The

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

TRADITIONAL INTERNALISM : AN INTRODUCTION



Problem of the Speckled Hen challenges a traditional internalist like Fumerton to make difficult decisions regarding how to classify perceptual experience and the things with which we are supposed to be acquainted. In Peter Markie’s contribution to this volume, ‘Confrontation Foundationalism,’ he discusses Fumerton’s account of non-inferential justification in light of the Problem of the Speckled Hen. He argues that the difficulties that arise give us reason to set aside the traditional internalist’s grounding of justification in the direct confrontation with reality. In ‘Acquaintance and Fallible Non-Inferential Justification,’ Chris Tucker raises concerns about acquaintance and non-inferentially justified but false beliefs and argues that while Fumerton is sympathetic to the possibility of non-inferentially justified but false beliefs, his view cannot accommodate the justification of false thoughts. In ‘Foundational Justification, Meta-Justification, and Fumertonian Acquaintance,’ Matthias Steup distinguishes between monistic and holistic kinds of foundationalism. Holistic, unlike monistic, foundationalism requires meta-justification. Steup argues that Fumerton’s traditional internalism is best understood as a kind of holistic foundationalism. Both Berit Brogaard and Susanna Schellenberg develop accounts of noninferentially justified perceptual belief that differ from the kind of account offered by the traditional internalist. In ‘Staying Indoors: How Phenomenal Dogmatism Solves the Skeptical Problem without Going Externalist,’ Brogaard attempts to distance herself from the kind of skepticism that worries Fumerton, and she argues for an account of Phenomenal Dogmatism where non-inferential perceptual justification is grounded in seemings. In ‘Experience and Evidence Abridged,’ Schellenberg highlights what she takes to be the shortcomings of the traditional internalist conception of evidence and presents an externalist view of perceptual experience as evidence.7

3.2 Inferentially justified belief Objections can also be pressed against what traditional internalists say about inferential justification. For example, one could argue that Fumerton’s account of inferential justification results in a vicious regress. Consider Lewis Carroll’s dialogue (see Carroll 1895) between the Tortoise and Achilles where Achilles is trying to prove that: (Z)

The two sides of this triangle are equal.

Achilles argues that (Z) follows logically from two premises: (A) Things that are equal to the same are equal to each other, and, (B) The two sides of this triangle are things that are equal to the same. 7

At the workshop mentioned in note 2, Schellenberg presented a paper based in part on her previously published paper “Experience and Evidence” (Schellenberg 2013). She had originally planned to develop that presentation into a new paper for this volume, but this didn’t work out. In light of that, we decided instead to include in the volume an abridged version of Schellenberg (2013). We are grateful to Schellenberg and to Mind for permission to reprint this abridged version here.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



BRETT COPPENGER

However, the Tortoise is also able to get Achilles to endorse: (C)

If (A) and (B) are true, then (Z) is true.

But once Achilles accepts that (C) is required in order to prove (Z), he is committed to an infinite regress. The Tortoise claimed that (A) and (B) by themselves do not imply (Z); one needs the further premise (C) that links (A), (B), and (Z). However, if (A) and (B) weren’t sufficient to infer (Z), then (A), (B), and (C) will not get the job done either. In addition, says the Tortoise, Achilles needs to accept: (D)

If (A), (B), and (C) are true, then (Z) is true.

Of course, (D) does not stop the regress. In addition to (A), (B), (C), and (D), Achilles is committed to: (E)

If (A), (B), (C), and (D) are true, (Z) is true.

The lesson for Achilles is that once he grants that (A) and (B) are not sufficient to imply (Z), he will never be able to finish writing down the required premises in order to prove (Z). The worry, for the advocate of PIJ, is that by appealing to what seems to be a higher-order requirement on inferential justification (like clause 2 of PIJ) a vicious regress results. If one must be justified in believing that E makes probable P in order to justifiedly believe P on the basis of E, then it would seem that in order to justifiedly believe P (on the basis of E and E makes probable P) one must also justifiedly believe that ‘P is made probable by the conjunction of E and E makes probable P.’ In short, an unhappy regress ensues. In ‘Principles of Inferential Justification,’ Trent Dougherty calls into question the motivation for Fumerton’s PIJ. According to Dougherty, inferential justification does not require the satisfaction of the skepticism-inducing conditions imposed by the kind of principle that Fumerton defends. In ‘Inferential Appearances,’ Michael Huemer offers an alternative to Fumerton’s acquaintance theory account of inferential justification. Huemer contends that his appearance account of inferential justification has important benefits that Fumerton’s account cannot provide.

3.3 Skepticism The starting point for investigation into knowledge by the paradigm traditional internalist, Descartes, was his consideration of beliefs about the external world. Yet, as Descartes famously pointed out, the possibility of dreaming or hallucination seems to immediately threaten the justificatory status of our beliefs about the external world. I believe that I have a cup of coffee on the table before me on the basis of my sensory experiences. However, because of the possibility of dreaming and hallucination, it seems that I could be having exactly that kind of sensory experience while at the same time my belief that I have a cup of coffee on the table before me is false. As a result, it must be admitted that the available evidence does not guarantee the truth of my external world belief. In addition, as Hume famously argued, one might

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

TRADITIONAL INTERNALISM : AN INTRODUCTION



also think that once one’s evidence is separated from the belief it is supposed to justify there is no legitimate means of bridging the logical gap that results. The Humean worry is that I simply do not have the resources available to justifiedly believe that my sensations are reliable indicators of reality. Of course, our beliefs about the external world should not be taken as a special case. The argument for skepticism in the case of external world belief can be generalized in a way that will call into question a number of different kinds of belief. If, in order to have a justified belief, I need to be justified in believing that my evidence makes likely my belief, then it might seem that more often than not, I am simply not in the position to have a justified belief. As a result of considerations like these, it is commonly maintained that traditional internalism has succeeded only at over-intellectualizing the requirements on justification in a way that fails to adequately answer skepticism. Surely, it is argued, a failure to respond to skepticism is a significant challenge for a view whose primary aim is the attempt to satisfactorily answer skepticism. In ‘The Costs of Demon-Proof Justification,’ Sanford Goldberg argues that the traditional internalist’s favorite refrain of appealing to the New Evil Demon Intuition does not actually result in the promising account of justification that the traditional internalist is hoping for. Instead, the costs associated with making one’s justification demon-proof are greater than it is usually assumed. In ‘Acquaintance and Skepticism about the Past,’ Ted Poston challenges Fumerton’s acquaintance theory and the PIJ on the basis of their skeptical consequences with regard to beliefs about the past. According to Poston, the acquaintance theorist is not in a position to avoid a far-reaching skeptical conclusion once the limitations of what we are acquainted with are realized. Of course, at this point the traditional internalist should be quick to remind us that even if traditional internalism cannot offer a successful solution to the problem of skepticism, taking skeptical challenges seriously, and then acknowledging the skeptical consequences of doing so still counts as a kind of answer, despite the fact that it may not be the answer one was hoping for. In addition to what many see as the straightforward skeptical consequences of traditional internalism, there is also the related problem of epistemic circularity. According to the traditional internalist, using one’s perceptual faculties to justify the reliability of one’s perceptual faculties seems problematic. But what, exactly, is wrong with using perception to justify perception, or memory to justify memory? Perhaps it is simply that, in establishing the efficacy of a belief-producing faculty, no genuine help can be provided by appealing to the faculty in question.8 Although the problem of epistemic circularity can be emphasized using perception, similar concerns can arise in connection with acquaintance. According to

8

Cohen (2002), for example, worries that all foundationalist accounts of justification face the problem of easy knowledge.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



BRETT COPPENGER

Fumerton’s account of non-inferential justification, when one has a justified belief, one has before one’s consciousness the very features that constitute the justification contributors of one’s beliefs. As we have seen, Fumerton thinks acquaintance is not by itself a justification-conferring relation; instead, justification results when we are acquainted with the appropriate objects of immediate experience. In light of the crucial role it plays, it seems reasonable to question the status of acquaintance. But surely the acquaintance theorist will have to appeal to her acquaintance with acquaintance in justifying the efficacy of acquaintance. How, then, is the acquaintance theorist’s position any less contaminated by epistemic circularity than the position of the person trying to justify perception by way of perception? In ‘On Metaepistemological Scepticism,’ Duncan Pritchard and Christopher Ranalli attempt to clarify Fumerton’s metaepistemology by comparing and contrasting it with Stroud’s. On their view, seeing the differences between Fumerton and Stroud helps to elucidate the role that rule-circular explanations play in our knowledge of the external world. In ‘How Our Knowledge Squares with Skeptical Intuitions despite the Circle,’ Ernest Sosa discusses traditional internalism, and in particular Fumerton’s view, in relation to the problem of epistemic circularity and the attempt to avoid skepticism. Sosa argues that one can take skepticism seriously while still avoiding both vicious circularity and skepticism.

4 Conclusion The goal of this introduction has been to clarify what traditional internalism is, how it can be usefully distinguished from the internalist competition, how it accounts for both non-inferential and inferential justification in a way that takes skepticism seriously, and what some of the most challenging problems it faces are. In ‘The Prospects for Traditional Internalism,’ the final chapter of the volume, Fumerton does some clarifying and reviewing of a similar nature. But in his case, the goal is to elucidate, revise, and defend his account of traditional internalism in light of the problems developed by the contributors of this volume. For more details on the contents of each of the chapters in the volume, see the list of chapter abstracts immediately following this introduction.9

References Bergmann, Michael. (2006). Justification without Awareness. Oxford: Oxford University Press. BonJour, Laurence. (2003). “Back to Foundationalism,” in Epistemic Justification: Internalism vs. Externalism, Foundations vs. Virtues. Malden: Blackwell, 60–96. Carroll, Lewis. (1895). “What the Tortoise Said to Achilles,” Mind, 4: 278–80.

9

I would like to thank Michael Bergmann for his helpful discussion on earlier versions of this chapter.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

TRADITIONAL INTERNALISM : AN INTRODUCTION



Chisholm, Roderick. (1942). “The Problem of the Speckled Hen,” Mind, 51: 368–73. Cohen, Stewart. (2002). “Basic Knowledge and the Problem of Easy Knowledge,” Philosophy and Phenomenological Research, 65: 309–29. Conee, Earl and Richard Feldman. (2004). Evidentialism: Essays in Epistemology. Oxford: Oxford University Press. Dougherty, Trent. (2011). Evidentialism and Its Discontents. Oxford: Oxford University Press. Fales, Evan. (1996). A Defense of the Given. Lanham, MD: Rowman and Littlefield. Foley, Richard. (2014). “Some Features of Richard Fumerton’s Philosophical Views.” Accessible online at: . Fumerton, Richard. (1995). Metaepistemology and Skepticism. Lanham, MD: Rowman and Littlefield. Fumerton, Richard. (2001). “Classical Foundationalism,” in Michael DePaul (ed.), Resurrecting Old-Fashioned Foundationalism. Lanham: Rowman and Littlefield, 3–20. Fumerton, Richard. (2004). “Achieving Epistemic Ascent,” in John Greco (ed.), Sosa and His Critics. Malden: Blackwell, 72–85. Huemer, Michael. (2007). “Compassionate Phenomenal Conservatism,” Philosophy and Phenomenological Research, 74: 30–55. McGrew, Timothy. (1995). The Foundations of Knowledge. Lanham, MD: Rowman and Littlefield. Poston, Ted. (2008). “Similarity and Acquaintance: A Dilemma,” Philosophical Studies, 147: 369–78. Schellenberg, Susanna. (2013). “Experience and Evidence,” Mind, 122: 699–747. Tucker, Chris. (2013). Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

Chapter Abstracts Chapter 1: Confrontation Foundationalism, Peter Markie Confrontation Foundationalism explains non-inferential justification as follows. CF: Non-inferential justification results from a direct confrontation with reality, in which the fact that makes true the belief is right there before consciousness. This chapter assesses the view, especially as developed by Richard Fumerton and Laurence BonJour. Objections to Confrontation Foundationalism based on the limits of our direct acquaintance and the speckled-hen problem are considered. It is argued that an adequate form of Confrontation Foundationalism must account for the difference between being directly acquainted with a truth-making fact and having such an acquaintance as assurance of one’s belief. That difference is best captured, not in terms of our awareness of a correspondence between a truth maker and a thought, as Fumerton and BonJour propose, but in terms of the role our direct awareness of a truth-making fact plays in a non-inferential identification ability.

Chapter 2: Acquaintance and Fallible Non-Inferential Justification, Chris Tucker Classical acquaintance theory is any version of classical foundationalism that appeals to acquaintance in order to account for non-inferential justification. Such theories are well suited to account for a kind of infallible non-inferential justification. Why is one justified in believing that one is in pain? An initially attractive (partial) answer is that one is acquainted with one’s pain. But since one can’t be acquainted with what isn’t there, acquaintance with one’s pain guarantees that one is in pain. What’s less clear is whether, given classical acquaintance theory, it’s possible to have noninferential justification to believe something false. Classical acquaintance theorists try to make room for such a possibility. The chapter argues, however, that the attempts of Richard Fumerton and Ali Hasan are inadequate, at least with respect to introspective justification.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



CHAPTER ABSTRACTS

Chapter 3: Foundational Justification, Meta-Justification, and Fumertonian Acquaintance, Matthias Steup Internalist foundationalism can be holistic or monistic. Holistic foundationalism says justification requires evidence providing meta-justification. Monistic foundationalism says it does not: a perceptual experience as of p gives, all by itself, a subject justification for believing p. Holistic foundationalists claim that such an experience gives a subject justification for believing p only if the experience is embedded in a body of evidence that justifies the subject in viewing the experience as reliable, thus putting the subject in a position to reject p-incompatible skeptical alternatives. This chapter discusses whether Fumerton’s internalism foundationalism is monistic or holistic. In light of his principle of inferential internalism, it initially seems Fumerton is holistic about inferential justification and monistic about foundational justification. However, given his commitment to internalism, it turns out that Fumerton’s acquaintance-based construal of foundational justification has a strong kind of metajustification built into it and thus must be understood as holistic.

Chapter 4: Staying Indoors: How Phenomenal Dogmatism Solves the Skeptical Problem without Going Externalist, Berit Brogaard Many naive realists and theorists who hold related positions (e.g., disjunctivism) are primarily motivated by epistemological considerations. They argue that unless we accept naive realism or related forms of direct realism, we cannot overcome the problem of skepticism. This chapter looks closer at this concern and attempts to figure out why naive realism is sometimes thought to be the only reasonable account of perception if one wants to avoid the skeptical problem. It is then argued that the naive realist’s response to the skeptical challenge is ultimately unsatisfactory and that phenomenal dogmatism, a form of epistemic internalism that is reconcilable with a representational view of perception, provides an adequate response to the skeptical problem.

Chapter 5: Experience and Evidence Abridged, Susanna Schellenberg It is argued that perceptual experience provides us with both phenomenal and factive evidence. To a first approximation, one can understand phenomenal evidence as determined by how the environment sensorily seems to the subject when she is experiencing. To a first approximation, one can understand factive evidence as necessarily determined by the environment to which the subject is perceptually

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

CHAPTER ABSTRACTS



related such that the evidence is guaranteed to be an accurate guide to the environment. It is argued moreover that the rational source of both phenomenal and factive evidence lies in employing perceptual capacities that subjects have in virtue of being perceivers. By showing that both kinds of evidence have the same rational source, one can achieve a unified account of perceptual evidence and its rational source in perceptual experience.

Chapter 6: Principles of Inferential Justification, Trent Dougherty This chapter presents a version of internalist foundationalism within the phenomenal conservative tradition—adopting the phenomenal conception of evidence/reasons— that gives a novel but commonsense account of the role of inference in human cognition covering both epistemic justification and knowledge. It argues, essentially, that there really is no such human activity as inference. Moves from premises to conclusions are taken to be sub-personal computational processes that are sometimes consciously initiated by the agent in question. However, the results of such ‘unconscious inferences’ are immediately justified for the agent, since the computation is at the sub-personal level. This approach avoids many problems traditionally associated with conceptions of non-basic belief within the tradition of internalism and foundationalism.

Chapter 7: Inferential Appearances, Michael Huemer Under what conditions is one justified in believing P on the basis of E? According to Richard Fumerton’s ‘inferential internalism,’ one must have justification for believing that E makes probable P. This chapter proposes that this condition is satisfied by a special sort of appearance, an ‘inferential appearance,’ by which it seems to one that P must be or is likely to be true in the light of E. This account explains the sense in which one must ‘see the connection’ between premises and conclusion; it avoids Lewis Carroll’s problem of Achilles and the Tortoise, it avoids the author’s own earlier objections to inferential internalism, and it avoids skepticism about inferential justification. The appearance-based theory is superior to rival theories that appeal to acquaintance, in part because it correctly allows that one can have inferential justification even when no genuine logical relation exists.

Chapter 8: The Costs of Demon-Proof Justification, Sanford C. Goldberg Internalist justification is standardly motivated by appeal to the New Evil Demon intuition. In this chapter it is argued that the Demon-proofing of justification comes

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



CHAPTER ABSTRACTS

at a great cost. Assuming that if S’s belief that p enjoys a ‘Demon-proof ’ justification, then the strength of S’s epistemic position is no greater than the epistemically worstoff of S’s doppelgangers, the burden of the argument is to show that the Demon can wreak havoc on the epistemic robustness of S’s doppelgangers. In this way the trope of the Demon can be used to show how weak internalist justification actually is, when it is formulated so as to meet the constraint of the New Evil Demon intuition. This (admittedly familiar) point will be brought out in a distinctly metaepistemological fashion; hopefully, this will bring new light to a traditional topic, and will honor the seminal epistemological work of Richard Fumerton.

Chapter 9: Acquaintance and Skepticism about the Past, Ted Poston This chapter examines the problem of skepticism about the past within Richard Fumerton’s acquaintance theory of non-inferential justification. Acts of acquaintance occur within the fleeting present. But if the data for non-inferential justification are limited to the fleeting present then the options for avoiding a far-reaching skepticism are quite limited. This chapter considers Fumerton’s responses to skepticism about the past and argues that his acquaintance theory is not able to stave off skepticism about the past. Furthermore, it is argued that the bounds of skepticism about the past overflow to the fleeting present by limiting the kind of content knowable within the present ‘now.’ Finally, this chapter defends an epistemic conservative response to this stark skeptical problem by arguing that it is a theoretically economical account of the justification for beliefs about the past.

Chapter 10: On Metaepistemological Scepticism, Duncan Pritchard and Christopher Ranalli Fumerton’s distinctive brand of metaepistemological skepticism is compared and contrasted with the related position outlined by Stroud. It is argued that there are at least three interesting points of contact between Fumerton and Stroud’s metaepistemology. The first is that both Fumerton and Stroud think that (1) externalist theories of justification permit a kind of non-inferential, perceptual justification for our beliefs about non-psychological reality, but it’s not sufficient for philosophical assurance. However, Fumerton claims, while Stroud denies, that (2) direct acquaintance with facts is sufficient for philosophical assurance. And this is important because Stroud thinks that (3) we have direct perceptual access or acquaintance with facts about the external world, while Fumerton denies this. What is key to understanding the metaepistemological scepticism offered by Fumerton and Stroud is an appreciation of the role that rule-circular explanations could potentially play in an account of how our knowledge of the world is possible.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

CHAPTER ABSTRACTS



Chapter 11: How Our Knowledge Squares with Skeptical Intuitions Despite the Circle, Ernest Sosa This chapter takes up a general epistemological and skeptical problematic, which is found in Pyrrhonism and is also important in Descartes’ epistemology, where it takes the form of the notorious Cartesian Circle. Epistemic circularity is the focus of the chapter, as part of the Agrippan trilemma that gives rise to the deepest skepticism that we owe to the schools of Ancient Greece. The question that needs to be considered, and that is considered in the chapter, is that of whether we can attain a general epistemological account of our cognitive lives without an inevitable, and an inevitably vicious, epistemic circularity.

Chapter 12: The Prospects for Traditional Internalism, Richard Fumerton In light of many of the comments and criticisms developed by the authors of the other chapters contained in this volume, this chapter clarifies, revises, and defends a version of a very traditional foundationalism. The chapter begins with an attempt to ground noninferential justification in the idea of a direct acquaintance with truth bearers, facts, and a truth-making relation. It is argued that the account developed succeeds in satisfying the internalist’s desire for a strong connection between having noninferential justification and having philosophically relevant assurance of truth. The chapter then moves on to a discussion of intellectually satisfying inferential justification and it concludes with reflections on the prospects this view has for meeting the challenge of skepticism.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PART I

Traditional Internalism and Non-Inferentially Justified Belief

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

Direct Acquaintance

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

1 Confrontation Foundationalism Peter Markie

What justifies a non-inferentially justified belief? Confrontation Foundationalism offers one answer. CF: Non-inferential justification results from a ‘direct confrontation’ with reality, in which the fact that makes true the belief is right ‘there before consciousness.’1 Confrontation Foundationalism isn’t popular—some of its proponents acknowledge as much (Fumerton 2006a: 689)—but it has a distinguished history, and its current defenders include such talented and thoughtful epistemologists as Richard Fumerton (1995, 2005, 2006a, 2009, 2010) and Laurence BonJour (2003, 2006).2 My aim is to assess Confrontation Foundationalism. I shall consider different versions of the view, but focus my attention on those of Fumerton and BonJour, with Fumerton’s account at center stage. Much of my assessment will be negative, as I present substantial objections to the versions I consider. I shall end though by exploring ways to avoid the objections, while preserving the insight in CF. I come as a sympathetic critic, trying to appreciate the challenges involved in developing Confrontational Foundationalism and the chance of meeting them.

1 Confrontational Foundationalism: Classical and Neoclassical Developing CF requires answering two questions. What is it to directly confront a fact, to have it ‘there before consciousness’? Why does our direct awareness of the fact that makes a belief true epistemically justify the belief for us? The questions come with a restriction. To end the regress of inferential justification, our state of direct I take the phrases, ‘direct confrontation’ and ‘there before consciousness’ from Fumerton’s characterization of the position (1995: 83). Some may associate CF with the title ‘Classical Foundationalism,’ but I prefer not to use that title as it often covers other claims beyond CF, such as that our non-inferentially justified beliefs do not extend to beliefs about the external world. 2 Earl Conee (1998) may also adopt CF for the case of a priori non-inferential justification. 1

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



PETER MARKIE

awareness, whatever its precise nature, must support epistemic justification without itself requiring it.3 Fumerton answers the first question in terms of a relation of acquaintance. To directly confront a fact is to be acquainted with it. Taking acquaintance to be, as he puts it, “a sui generis relation that holds between a self and a thing, property or fact” (1995: 74), he doesn’t define it but ostends it by a thought experiment. He imagines a pain that enters, leaves, and reenters consciousness as he is momentarily distracted from it during a conversation. Acquaintance is that relation I had to my pain (something other than belief) that was present, temporarily ceased during my conversation, and began again when the conversation ended. (2005: 123)

We can deepen our understanding of acquaintance by considering some additional points. First, as Fumerton’s pain example indicates, acquaintance isn’t necessarily a relation to some conceptual or propositional content. One can be acquainted with a property or fact without even possessing the conceptual resources to represent that fact in thought, and certainly without possessing the ability to linguistically express that fact. (1995: 74)

A dog or a young child can be acquainted with its sharp pain without having any conceptual resources by which to conceive of it as a pain or as sharp. Second, acquaintance is “not by itself an epistemic relation” (1995: 75); in particular, instances of acquaintance are not open to evaluation as being epistemically justified or unjustified. Third, acquaintance “cannot be reduced to non-epistemic concepts, particularly the nomological concepts upon which all externalists build their analyses” of justification (1995: 83).4 Fourth, the range of our acquaintance is limited: “it is always a mental state or feature of a mental state with which we are acquainted” (1995: 79), and our acquaintance with an experience is restricted to its non-relational properties. So any defensible form of radical foundationalism claims only that we are sometimes directly acquainted with some of the properties of an experience. The relevant properties are, presumably, non-relational properties characterizing the intrinsic nature of the experience. (2009: 208; Fumerton’s emphasis) Looming in the background is what Sosa aptly terms, “The Dreaded Sellars Dilemma.” [E]ither the foundational conscious states have propositional content, in which case they would seem to require justification in turn, and could not after all function as a foundation; or else they have no propositional content of their own, in which case it is hard to see how they could possibly provide epistemic justification for any belief founded upon them (2003: 212). 4 Since acquaintance is a non-epistemic relation, the concept of acquaintance is presumably a non-epistemic concept. When Fumerton claims that the concept of acquaintance “cannot be reduced to non-epistemic concepts,” his point is presumably one about the concept’s irreducibility. He does not intend to imply that the concept is epistemic after all. 3

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

CONFRONTATION FOUNDATIONALISM



We can be acquainted with our pain and its property of being sharp; we cannot be acquainted with its property of being caused by a malignant tumor in our brain.5 So much, for now, for our first question; what of the second: Why does our being acquainted with the fact that makes a belief true make the belief justified for us? To appreciate the question, consider what I’ll call ‘Classical Confrontation Foundationalism.’ CCF: S has non-inferential justification for believing that P, if S is acquainted with the fact that P. If we are acquainted with the fact that we are in pain, then we have—in the sense that we are acquainted with it—something that assures us of the truth of the belief that we are in pain, and, if we have assurance of the belief ’s truth, we have justification for it. This all sounds plausible enough, but it won’t do. We can be acquainted with our pain but have no sense, no idea, no awareness, no clue that our pain is what makes our thought that we are in pain true. It may, from our perspective, be no more relevant to the truth of our thought than any other experience with which we are acquainted. The fact that we are in pain makes our belief that we are in pain true, and when we are acquainted with it, we have—in the simple sense that we are acquainted with it—something that ensures the truth of our belief. Yet, we don’t have that fact as assurance of our belief ’s truth, unless we have an appropriate awareness of the fact as supporting the belief ’s truth. CCF needs an additional condition, one that captures our need to be appropriately aware, not just of our pain, but of our pain as supporting the truth of our belief.6 This condition can’t involve our judging that our pain makes the content of our belief true, for again, the conditions for non-inferential justification mustn’t include any state open to evaluation as justified or unjustified. Fumerton addresses the issue by adding another acquaintance condition to the Classical view. Classical acquaintance theorists like Russell appropriately emphasized the role of acquaintance with particulars, properties, and even facts in grounding justification. But a fact is not a truth, and what one needs to end a regress of justification is a direct confrontation with truth. To 5 Fumerton (1995: 79) notes that other acquaintance theorists may have a more expansive view of the range of acquaintance, allowing for acquaintance with the fact that the surface of a physical object exemplifies a certain property, or with mind-independent universals or with mind-independent, nonoccurrent states of affairs and logical relations between them. Coupled with Confrontation Foundationalism, such different views on the range of acquaintance may well have different implications for the range of our non-inferentially justified beliefs. For more on the limits of non-inferential justification, see Fumerton (1995: especially 184–7, and 2006a). 6 Compare Fumerton’s position (1995: 85–9) on what is required for us to have inferential justification. Having inferential justification for believing that P requires more than having justification for believing another proposition, Q, that makes P probable. We must also have justification for believing that Q makes P probable. Another analogy: Being aware of a sign that tells us where we are in words we don’t understand doesn’t constitute having an awareness of the sign as assurance of where we are, so too simply being acquainted with the fact that makes a belief true isn’t yet having an awareness of that fact as assurance that the belief is true.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



PETER MARKIE

secure that confrontation, one needs to be directly aware of not just a truth-maker (a fact to which a truth corresponds) but also a truth bearer (a thought) and the correspondence that holds between them. (1995: 75; Fumerton’s emphasis)

He puts his proposal, which I’ll call ‘Neoclassical Confrontation Foundationalism,’ as follows. But how does acquaintance give us non-inferential justification? My suggestion is that one has a non-inferentially justified belief that P 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. No single act of acquaintance yields knowledge or justified belief, but when one has the relevant thought, the three acts together constitute non-inferential justification. When everything that is constitutive of a thought’s being true is immediately before consciousness, there is nothing more that one could want or need to justify belief. (Fumerton 1995: 75)

In developing his position, Fumerton reminds us of the distinction between having justification for a belief (sometimes called ‘propositional justification’) and the belief ’s being justified for us (sometimes called ‘doxastic justification’).7 In each case, we have the required basis for the belief; in the latter case, we actually hold the belief on that basis (1995: 91). We thus have two principles: NCFP: S has non-inferential justification for believing that P, if S is acquainted with the fact that P, the thought that P, and the relation of correspondence between the thought that P and the fact that P.8 NCFD: S is non-inferentially justified in believing that P, if (1) S believes that P, (2) S is acquainted with the fact that P, the thought that P, and the relation of correspondence between the thought that P and the fact that P, and (3) S bases his belief that P on his acquaintance with these facts.9 As Fumerton tells us, When one is directly acquainted with, say, one’s pain as one entertains the thought that one is in pain and one is directly acquainted with the correspondence between the thought and the pain, one has non-inferential justification for believing that one is in pain. And when one bases one’s belief on that justification, one’s belief is justified. (2006a: 689)

The situation is a bit more complex than the initial insight (CF) and the Classical view (CCF) suggest, but, in the end, non-inferential justification derives from our direct awareness of the fact that makes our belief true. We gain justification by an

7

See Bergmann (2006b: 3–9) for a helpful discussion of propositional and doxastic justification. Although Fumerton does not explicitly present his account as concerned with prima facie justification, I assume that it is. The conditions given in these principles are open to epistemic defeat. 9 See Fumerton (2009: 207) for the explicit inclusion of the basing condition. 8

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

CONFRONTATION FOUNDATIONALISM



acquaintance with that fact and its truth-making relation to the content of our belief, and we believe with justification when we believe on the basis of that acquaintance. BonJour (2003: 193) gives a similar account, though he doesn’t employ the concept of acquaintance. Where being e-aware of an experience is just having it among the contents of one’s consciousness, his view is captured by the following conditions. S is non-inferentially justified in believing that his experience has a certain character, if S is e-aware of an experience with that character, attends to the relevant aspect of the experience, believes the experience to have that character, and apprehends the agreement or fit between the aspect of the experience and the conceptual description given by the belief.10

When we meet these conditions, we have, according to BonJour, “a kind of reason for thinking that the description [in our belief] is true or correct” (2003: 72; his emphasis). Although he makes no reference to acquaintance and writes instead of our “apprehending or recognizing” the agreement between our experience and a description of it, BonJour (2003: 193) describes his account as “essentially the same view that Fumerton has in mind.” Focusing our attention back on Fumerton’s two principles, it is important to note three points. First, the explanation in NCFD of what it is to be non-inferentially justified in believing a proposition is parasitic on that in NCFP of what it is to have non-inferential justification. The former is simply the latter plus two psychological conditions, that we believe the proposition and base our belief on the relevant states of acquaintance. Fumerton thus takes NCFP to contain the meat of his account. It seems obvious that in the context of normative epistemology, philosophers qua philosophers should be interested primarily, or perhaps even exclusively, in questions as they relate to having justification. Once one settles whether or not one has justification for believing some proposition, the question whether the belief is justified becomes an empirical question concerning causal connections, an empirical question that goes beyond the scope of philosophical investigation. (1995: 92)

The question of what it is to have justification is primary, and its answer does not involve any nomological conditions about the belief ’s formation. Fumerton’s account of non-inferential justification is thus designed to avoid what he (1995: 84) takes to be the “fundamental mistake of externalism [which] is its attempt to reduce the epistemic to the nomological.”11 We’ll soon take a closer look at whether we can explain what it is to have justification without any appeal to nomological conditions. Second, note that Fumerton’s principles offer sufficient, but not necessary, conditions for non-inferential justification. Other sources of non-inferential justification

10

This is my summary of the conditions BonJour presents (2003: 193). Not all normative epistemologists share Fumerton’s emphasis on having justification as opposed to being justified in believing. Certainly, some externalists don’t. Consider Bergmann (2006b: 4): “Let me begin my attempt to isolate the sort of justification on which I’ll be focusing by saying that it is a form of doxastic justification, not of propositional justification” (his emphasis). 11

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



PETER MARKIE

exist. We might be in pain, never form, and so never be acquainted with, the thought that we are in pain, and yet still have non-inferential justification for believing that we are in pain. We can also have, as Fumerton acknowledges, non-inferentially justified false beliefs. I also argue that it may be possible on an acquaintance theory to have non-inferential justification that does not entail the truth of what is believed. Specifically, I have argued that one might be acquainted with a fact very similar to the fact that makes P true, and such acquaintance might give one a justified but false belief that P. (1995: 77)

Fumerton’s comment indicates a strategy for extending Confrontation Foundationalism to non-inferentially justified false belief: Locate the source of justification in our acquaintance with a fact very similar to the one that would make our belief true. The details are obscure, however. What is the relevant relation of similarity between the fact that would make our thought true and the fact with which we are acquainted? With what relation between that fact and our thought are we acquainted, given that it is not the truth-making relation of correspondence? Third, consider how, relative to these principles, Fumerton answers our second question: Why does our having certain states of acquaintance support epistemic justification? He tells us the following. When everything that is constitutive of a thought’s being true is immediately before consciousness, there is nothing more that one could want or need to justify belief (1995: 75; Fumerton’s emphasis). I stub my toe and I believe that I am in excruciating pain. What justification do I have for thinking that I’m in pain? How do I know that I’m in pain? My answer is that I am directly aware of the pain itself—the very truth maker for my belief. The pain is ‘there’ transparently before my mind. The thought that is about the pain and the pain that is its object are both constituents of the conscious mental state that I call acquaintance. When all this is so, we are in a state that is all that it could be by way of satisfying philosophical curiosity. What more could one want as an assurance of truth than the truth-maker there before one’s mind? (2006b: 189)

Fumerton’s final rhetorical question (“What more could one want as an assurance of truth than the truth-maker there before one’s mind?”) isn’t quite right. We’ve already seen we need more. We must have that truth maker before our mind as assurance of our belief ’s truth. We must, in Fumerton’s terms, be acquainted, not only with the fact that makes the belief true, but also with the truth-making correspondence between the fact and the belief ’s content. Still, when all this is the case, what more could we want as assurance of the truth of our belief? There is no denying the question’s pull. It is hard to figure what additional mental state we could properly want as assurance, given that we are already acquainted with our pain, our thought that we are in pain and the truth-making correspondence between the two. An awareness of other contents of thought that represent our pain to

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

CONFRONTATION FOUNDATIONALISM



us, say the sound of our low moaning voice or the doctor’s office image of a teary, frowny face, adds nothing to our epistemic support; such contents are but indirect significations of the pain already before our consciousness. A Dogmatist/Phenomenal Conservativist may want to add a seeming experience, its seeming to us as if we are in pain. Yet, such a seeming experience also appears to add nothing of significance; we are already acquainted with the reality of which it indirectly takes note, the truthmaking relation between our pain and our thought that we are in pain. Even as we feel the pull of Fumerton’s question, we should resist it. It is not clear we can have all the states of acquaintance it assumes. Even if there’s nothing more to add to our experience to better assure us of the truth of our belief, perhaps having justification isn’t just a matter of the contents of our experience. For reasons we are about to see, Neoclassical Confrontation Foundationalism doesn’t solve the problem with the Classical view. It doesn’t capture the fact that having justification requires having the truth maker before our mind as assurance of our belief ’s truth.

2 The Limits of Acquaintance The Neoclassical view seeks to improve on the Classical position by adding another condition to the account of non-inferential justification. We must be acquainted with the truth-making fact and with the correspondence between that fact and the thought that is the content of the associated belief. Yet, adding this condition leads to an anomalous result that is cause to wonder whether we are on the right track. According to Neoclassicism, when we are in pain and think we are, our justification consists in our acquaintance with our pain, with our thought that we are in pain, and with the truth-making correspondence between the two. What about when we are in pain but happen not to form the thought that we are? We still have justification for believing that we are in pain, but, according to the Neoclassicist, our justification consists in something else. This is at best unexpected. Our thinking that we are in pain makes no difference for whether we have justification, so it should—one would think—make no difference for the nature of our justification. A more serious problem is that the Neoclassical view overestimates the range of our acquaintance. Can we really be acquainted with the truth-making relation of correspondence between our experiences and thoughts about them? Fumerton doesn’t defend his claim that we can, and his own account of acquaintance actually implies otherwise. If we are acquainted with the truth-making relation of correspondence between our pain and our thought that we are in pain, we are, presumably, acquainted with the fact that one makes the other true. Yet, recall: The range of our acquaintance is limited to our mental states. Our pain and our thought are mental states; the fact that our pain makes our thought true most certainly is not. Perhaps, the object of our acquaintance is not the fact that our pain makes our thought true, but something else; it is our pain and its property of making our thought true, or perhaps, it’s our thought and its property of being made true by our pain. Yet, recall

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



PETER MARKIE

again: our acquaintance only extends to the non-relational properties of our experiences. Making-true and being-made-true-by are relational properties. This can, of course, be sorted out. Fumerton might plausibly claim that we can, after all, be acquainted with some relational facts about our mental states and with some relational properties of our experiences. When we have two simultaneous pains, we can be acquainted with the fact that one is sharper than the other. It is plausible (Sosa 2003: 130) that when we experience two triangular images, we are acquainted with each and also acquainted with their shape-sameness. Perhaps, it is equally plausible to say that we can be acquainted, not only with our pain and the thought that we are in pain, but with the relational fact that the one makes the other true. Plausible, perhaps, but not true. Let’s risk channeling Hume for a moment. What do we find, when we enter ‘most intimately’ into ourselves? For my part—and I suggest the same for you—I am acquainted with mental states and some of their properties. I am sometimes acquainted with my pain as being sharp. I am sometimes acquainted with my thought that I have a sharp pain. Yet, I am never acquainted with any of my pains or other experiences as making true a particular thought, and I am never acquainted with any of my thoughts as being made true by a particular experience. I am, of course, sometimes acquainted, and I suppose you are too, with thoughts to the effect that a particular experience, a pain, makes a particular thought, that I have a sharp pain, true. Yet, being acquainted with that thought about the truth-making relation between my pain and the thought that I am in pain is not being acquainted with the reality that it is about, with the fact that my pain stands in a truth-making relation of correspondence to the thought that I am in pain. The following general principle is, I suggest, true: If we are acquainted with a particular aspect of a content of consciousness, then that aspect is part of how we experience that content. When we are acquainted with the sharpness of our pain, we experience our pain as sharp. We are not acquainted with our pain as being caused by cancer, because we don’t experience our pain as being caused by cancer. Consider then a pain’s relational property of making a particular thought true. We may experience the pain in many ways, as sharp, dull, throbbing, etc., but we do not experience it as making a particular thought true. The argument is simple enough. 1. We are acquainted with a particular aspect, e.g. the sharpness, of a content of consciousness, e.g. our pain, only if we experience the content as having that aspect. 2. We do not experience any content of consciousness, e.g. our pain, as making true a particular thought. 3. Therefore, even if some contents of consciousness make certain thoughts true, we are not acquainted with that aspect of them. As I understand him, Fumerton rejects the second premise. I’m not sure how to convince him or anyone else of his error, except to issue the Humean invitation. Turn

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

CONFRONTATION FOUNDATIONALISM



your attention inward, take a look at what’s there, and be careful to distinguish between a thought to the effect that a particular experience makes a particular thought true, with which you might be acquainted, and the actual aspect of an experience that is its making that thought true, with which you are not. BonJour’s Neoclassical position encounters the same problem. Consider his description of a case of non-inferential justification. Suppose that I am looking directly at a large, multicolored, abstract painting, under good lighting conditions, with normally functioning vision and no interfering factors, and at a distance at which the image produced by the painting roughly fills my visual field. I come . . . to entertain the propositional claim that within my rather complicated visual field there is a sharp green, approximately equilateral, triangular shape . . . I come to have the corresponding belief, and, according to my account, this belief is justified—justified by my awareness of the descriptive fit between the content of the proposition I believe and the relevant aspect of the content of the experience. (2006: 744; my emphasis)

BonJour explains his non-inferential justification as stemming from his “awareness,” not just of his visual experience and his thought, but of the “descriptive fit” between the two. He gives a similar description elsewhere (2003: 193) writing of his “seeing that” his experience fits a particular description. Where being e-aware of an experience is just having it among the contents of one’s consciousness, he tells us: When I am e-aware of an experiential feature, it is genuinely present in my consciousness. Because many other things are present there as well, I may have to selectively focus in order to pick out that feature and come to see that its character fits a particular description, but the awareness of it that allows me to do these things is just the original e-awareness itself. To insist that a distinct cognitive act, requiring an independent justification, was involved would be in effect to say that e-awareness in itself has no cognitive significance at all, that I can be genuinely conscious of the feature in question but be unable to do anything further with that consciousness. (2003: 193)

No doubt BonJour, and the rest of us, can be acquainted with, or, as he puts it, e-aware of, an experience such as a pain, as well as of the thought that we have the experience. He is also right that we can do some things with this acquaintance; perhaps we can even come to, as he puts it, “see that” the thought fits the experience. Yet, that ‘seeing that’ is not itself another case of acquaintance, for the reasons I have given. Indeed, BonJour’s use of ‘seeing that’ at this point suggests as much. ‘Seeingthat’ is generally understood to be a form of judging. We can certainly judge that a particular experience makes a particular thought true. Our judgment isn’t an instance of acquaintance, and it’s open to evaluation as justified or unjustified.12 12 Bergmann (2006a) raises a closely related, though different, concern about BonJour’s account. If the ‘seeing-that’ condition does not involve judging and is not open to epistemic evaluation as justified or unjustified, the account violates the internalist insight (shared by Confrontation Foundationalists) that justification has an awareness requirement. According to Bergmann, the very cases of unjustified belief that internalists use to illustrate the fatal absence of an awareness requirement in such externalist accounts of justification as Reliabilism can be applied with similar effect against BonJour’s account.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



PETER MARKIE

The situation worsens for Neoclassicism in the case of false non-inferentially justified beliefs. In this case, we are presumably acquainted, not with a truth-making correspondence between our experience and the content of our belief, but with some other relationship. What relationship? Perhaps, the requirement is that we be acquainted with the ‘almost-like-what-makes-for truth’ relationship between an experience and a thought. Yet, we clearly are not acquainted with any such relationship, even when one exists.

3 Acquaintance and Identification Ability Perhaps I am mistaken. We can be acquainted with the fact that a particular experience makes a particular thought true. Neoclassicism then faces another problem. The conditions in NCFP and NCFD are not sufficient for justification. To return to Fumerton’s rhetorical question, it is not that we need more conscious states to gain justification. It is that justification isn’t just a matter of our conscious states. The case of the speckled hen is often used to make the point (see Sosa 2003, Markie 2009a, and Fumerton 2009). Suppose we visually experience a 3-speckled hen, and we are able to non-inferentially distinguish images of 3-speckles from images of other numbers of speckles. We form the non-inferentially justified belief that we are having a 3-speckles experience. According to the Neoclassical view, our justification for our belief stems from our acquaintance with our 3-speckles experience, with our thought that we are having a 3-speckles experience, and with the truth-making correspondence relation between the experience and thought. Having these states of acquaintance gives us non-inferential justification, and in basing our belief on them, we believe with justification. Fair enough. Now suppose we visually experience a 48speckled hen. We are unable to visually distinguish an image of 48-speckles from images of other numbers of speckles, say 47 or 50. We have no non-inferential justification for believing that we are having a 48-speckles experience. Yet, Neoclassicism implies otherwise. We can be acquainted with our 48-speckles experience, with the thought that we are having a 48-speckles experience, and with the truthmaking relation of correspondence between the two. We can thus have justification for believing that we are having a 48-speckles experience. If we then believe on the basis of these acquaintance states, our belief, according to Neoclassical Confrontation Foundationalism, is non-inferentially justified. That’s surely not right.13

13 I present the case as one where the subject bases his belief on his states of acquaintance. Others, for example, Sosa (2003: 137), present the case as one where the subject guesses in forming his belief. So presented, the case is a counterexample to NCFP, but not to NCFD, since the subject never meets the latter’s condition that he base his belief on his states of acquaintance. As Sosa notes, the subject also doesn’t know that he has a 48-speckles experience. Confrontation Foundationalists can agree with this point. Not all true justified beliefs are knowledge, especially ones in which the subject lands on the truth accidentally, as happens here.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

CONFRONTATION FOUNDATIONALISM



The general features of the speckled-hen case can be duplicated to produce a host of similar counterexamples to Neoclassicism. Sosa (2003: 129) presents several, including a case where we see a well-lit white decagon against a black background in an otherwise darkened room. We can be acquainted with the shape, with our thought that it is a decagon and with the correspondence between the two, but, given our assumed inability to identify decagons as such, we have no non-inferential justification for believing that it is a decagon. The list goes on. I shall refer to the initial case involving our visual experiences of speckled hens as ‘the original speckled hen case.’ I shall refer to cases of this general sort as ‘speckled-hen cases.’ As counterexamples to Confrontational Foundationalism, speckled-hen cases have two components. First, there’s the assumption that the subject has all the required states of acquaintance and bases his belief on them. The bare bones nature of acquaintance seems to allow for this much. Second, there’s our intuition that the subject lacks a non-inferentially justified belief, because he lacks a required identification ability. The cases rely on the following principle. Ability: We have a non-inferentially justified belief that our experience is of a certain character, only if we form that belief through an (innate or learned) ability to non-inferentially identify experiences as having that character. Ability sure seems correct. Suppose someone claims to have non-inferential justification for his introspective belief that his experience has a certain character, even as he correctly admits that he has no ability to introspectively identify which of his experiences have that character and which do not. He tells us, “I just introspect, and I am non-inferentially justified in believing that my experience is C, though, of course, I’ve no ability to non-inferentially distinguish C experiences from non-C experiences.” Surely, he is confused. His correct admission that he has no noninferential ability to identify which of his experiences is which undercuts his claim to be non-inferentially justified in his belief. If he has no non-inferential identification ability, any justification he has for his belief must have come by an inference from other justified beliefs.14 What is the basis for the link between non-inferential justification and an associated identification ability? I think Sosa (2003: 138) gets it right. [I]f the judgment (with its content) is to be rationalized by the experience (with its relevant character), then the former must be appropriately responsive to the latter, in such a way that variations in the latter would have led to corresponding variations in the former. Or perhaps it will suffice for appropriate responsiveness that one might easily not have believed this without that belief ’s being then accompanied by one’s experiencing in a corresponding way.

14 Fumerton (1995: 85–6) gives somewhat similar considerations in support of the principle of inferential justification that “one’s belief in some proposition E can justify one in believing another proposition P only when one’s belief that E is itself justified and one has justification for thinking that E makes P probable.”

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



PETER MARKIE

If we are to have an experience—in this case one of acquaintance—as assurance for a belief—if, in Sosa’s words, the experience is to “rationalize” the belief for us—the belief must be appropriately responsive to the experience. In the absence of a logical support relation between their propositional contents, they must be linked by an appropriate belief-forming disposition. There are lots of ways to understand the relation of appropriate responsiveness. Sosa suggests some here. Proper functionalists will suggest another: the experience and belief are to be connected by a design plan specifying the proper function of a reliable cognitive faculty. We need not canvass the options. The important point is one they all share. Having an experience as assurance for a belief requires the exercise of a non-inferential identification ability, an appropriate disposition to form beliefs of that sort in response to experiences of that sort. Classical Confrontational Foundationalists overlook this point when they take our mere acquaintance with a fact to be sufficient for justification. Neoclassicists overlook it when they try to fix the Classical view by just adding more states of acquaintance. The additional acquaintance states don’t ensure the presence of the required identification ability. We might expect Neoclassicists to respond to speckled-hen cases by rejecting Ability and claiming that the subject in speckled-hen cases gains non-inferential justification from his acquaintance states, despite lacking a non-inferential identification ability. Neither Fumerton nor BonJour does so, however (and insofar as I know, neither does anyone else). They accept the intuition that the subject lacks justification, and, insofar as I can tell, they challenge the other component of speckled-hen cases, the assumption that the subject has the required states of acquaintance. Their argument isn’t always clear, but interestingly enough, their point seems to be that the subject lacks the acquaintance states because he lacks the required identification ability. The exercise of a non-inferential identification ability is required in these cases, not as a component or source of justification, but as a necessary condition for the states of acquaintance that are the source of justification. Fumerton (2005, 2009, 2010) presents three ways to develop this response, tied to the three conditions presented in NCFP for non-inferential justification. To defend Neoclassicism against the original speckled-hen case, one might argue that the subject doesn’t have, and so isn’t acquainted with, a 48-speckles experience. So at least some philosophers would question the presupposition that there is some determinate number of speckles characterizing the visual experience. You have the property of being appeared to many-speckled-ly, but you simply don’t have the property of being appeared to n-speckledly where n represents some particular number. (2010)

One might instead argue that the experience is a 48-speckles experience but the subject is not acquainted with that aspect of it. Alternatively, one might allow that an experience can be characterized determinately but deny that one typically has acquaintance with the experience’s having that particular determinate character. (2010)

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

CONFRONTATION FOUNDATIONALISM



And, one might argue that the subject is not acquainted with the truth-making correspondence between the experience and the thought that it’s a 48-speckles experience. Alternatively, the acquaintance theorist might deny that one can have the thought that something has some determinate number of speckles, at least in way [sic] that would allow one to be acquainted with a correspondence between that thought and the fact that makes it true. (2010)

One way or another, the subject in a speckled-hen case lacks the required acquaintance states. Yet, why can’t the subject have all the required states of acquaintance? Fumerton offers different answers based on different theories of experience, but they all assume that it’s ultimately because the subject lacks the associated non-inferential identification ability. If he should ever gain that ability, he’ll then be able to have the states of acquaintance to support non-inferential justification. Consider in this regard, Fumerton’s explanation of how Neoclassicism allows for the fact that the range of our non-inferential justification can increase as we develop new identification abilities. [O]ne can, through a learning process, get oneself in a position to know (non-inferentially) features of musical experience that were previously epistemically hidden. Now I suggested above that the classical foundationalist can accommodate this phenomenon. You will recall that the direct acquaintance theorist holds only that some properties of experiences can be discovered non-inferentially. If there are both determinate and determinable properties, I don’t see why one can’t be acquainted initially only with determinable properties and later, perhaps as a result of learning, become acquainted with more determinate properties. (2009: 4)

When we learn to identify features of a musical experience that we couldn’t identify previously, we extend the range of properties with which we can be acquainted, and that extension in the range of our acquaintance accounts for an extension in the range of our non-inferentially justified beliefs about our experience. Presumably, the same goes for subjects in speckled-hen cases. If they develop the requisite identification ability, they will have the requisite states of acquaintance. They will then have justification for their beliefs. BonJour may well have a similar view. With the original speckled-hen case among others in mind, he seems to discard the first two alternatives suggested by Fumerton and to settle on the third. [O]ne can have an experience with a certain feature, understand completely what an experience has to be like to be correctly described as having that feature (and so possess all the relevant concepts), attend to the relevant aspect or aspects of the experience, and still sometimes be unable to tell whether or not the description is actually correct, where at least the most obvious sort of case in which this happens is where the description is sufficiently complex as to make it hard to tell whether it applies even when the relevant feature is being attended to. Here the image of the speckled hen . . . can serve as [an example]. (2003: 192)

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



PETER MARKIE

The subject in the original speckled-hen case can have an experience of 48-speckles and attend to that aspect of his experience, even as he lacks the ability to identify his experience as being 48-speckled. He cannot, however, “apprehend or recognize the agreement or fit between the aspect of the experience being attended to and the conceptual description given by the belief ” (2003: 193). Put in the fashion of NCFP and NCFD, he lacks the required acquaintance with the truth-making correspondence between his experience and his thought. Why? BonJour doesn’t give an explanation, but he refers us to the subject’s lack of the associated identification ability. The case is one “where the description is sufficiently complex as to make it hard to tell whether it applies even when the relevant feature is being attended to.” This reply to the speckled-hen problem accommodates Ability, without sacrificing the view that states of acquaintance are a sufficient basis for non-inferential justification. One can maintain, as Fumerton (1995: 83–5) in particular does, that noninferential justification does not have any nomological components. There is no need to add an ability condition to NCFP or analyze acquaintance in nomological terms to capture the speckled-hen cases. Having non-inferential justification requires having an appropriate identification ability, but only because having such an ability is necessary to having the states of acquaintance that are the real basis for justification. The states of acquaintance, which are gained in the course of developing associated identification abilities, are the source of non-inferential justification. Yet, must the subjects in speckled-hen cases lack the states of acquaintance specified in NCFP, given that they lack the associated identification ability? Fumerton (2005: 209) claims they must; others (Sosa 2003: 130–40, Markie 2009a) offer hypothetical cases where they find the acquaintance states to be present and the identification ability absent. We can settle the issue once we recall a simple point. The principles of Confrontation Foundationalism are presumably offered as necessary truths. The speckled-hen counterexamples need only be possible to refute them. Ability, which Neoclassicists appear to accept, is also a necessary truth. The defense of Neoclassicism requires then that it be impossible that a subject in a speckled-hen case has all the acquaintance states NCFP presents as sufficient for justification but lacks the associated identification ability. This just isn’t so. Subjects can lack the appropriate identification ability—and with it any associated non-inferential justification— but still have the relevant states of acquaintance. Having an ability to use some states of acquaintance in forming beliefs is one state of affairs and having those states of acquaintance on a particular occasion is another. It is possible that the latter occur without the former. Recall that states of acquaintance are non-epistemic, psychological states of direct awareness. They do not require any judgment or conceptualization of content. It is at least possible that a subject in a speckled-hen case has some brain states that produce the associated states of acquaintance without having the associated identification ability. Pick your favorite possible world scenario. Even as a subject lacks the required identification ability, an evil demon manipulates his brain to produce the states of acquaintance. If you prefer, a brain tumor does it. Better still,

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

CONFRONTATION FOUNDATIONALISM



it’s Edna with a ray gun. One way or another, it is possible that something causes a subject, who lacks the required identification ability, to have the specified acquaintance states.15 Confrontation Foundationalists may, of course, claim that the states of acquaintance really are necessarily linked to the development of related identification abilities. It is a necessary truth that someone who has the required states of acquaintance has the required identification ability. This response is baldly ad hoc, however, unless it is accompanied by an explanation of the necessary connection. It is hard to see what that explanation might be, given how sparse the concept of acquaintance is, and even harder still, given that acquaintance is not to be understood in nomological terms in particular.

4 Alternatives Classical Confrontation Foundationalism misses the distinction between being acquainted with the experience that makes a belief true and having the experience as assurance of the belief ’s truth. Neoclassical Confrontation Foundationalism tries, but fails, to capture the distinction by adding more acquaintance states to the conditions for non-inferential justification. Should we give up on Confrontation Foundationalism? Maybe not. I want to briefly consider two ways to develop the view. While the second is more promising than the first, I don’t propose either. I present them to show how my objections to Confrontation Foundationalism don’t close the door on the view. Neoclassicists might remind us of the distinction between having justification for a belief, treated by NCFP, and being justified in a belief, treated by NCFD. Speckledhen cases, they might claim, are not counterexamples to NCFP. The subject in the original case has justification for believing that he has a 48-speckles experience, by virtue of having the acquaintance states specified. His lack of an ability to noninferentially identify 48-speckles experiences does not keep him from having those acquaintance states or from having justification. It instead keeps him from basing his belief on his states of acquaintance in a way that supports justified belief. Insofar as he forms the belief that he has a 48-speckles experience, he forms it on a basis other than his states of acquaintance. Perhaps, he forms it as a guess. As NCFD implies, he is, therefore, not justified in his belief. Speckled-hen cases pose no problem, and Ability is true: An appropriate non-inferential identification ability is necessary for justified belief. 15

Fumerton (1995: 71), following Ayer, argues in a similar way for the claim that we can believe that we are in pain when we actually are not: “It is a Humean sort of argument that proceeds from the simple observation that in the vast majority of cases, the belief that P is one state of affairs and P’s being the case is a different state of affairs. If these really are two distinct facts, then why couldn’t one have the one without the other?”

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



PETER MARKIE

Successfully developing this response requires correctly explaining both the kind of basing relation and the kind of ability required for non-inferentially justified belief so that the former requires the latter. Perhaps that can be done. The response also retains NCFP, which I have argued is deficient because we are not acquainted with relations of truth-making correspondence between our thoughts and our experiences. If I’m right about that, we need to keep looking. Here is another option. NCFP*: S has non-inferential justification for believing that P, if S is acquainted with the fact that P, and S has an appropriate ability to non-inferentially identify the fact that P on the basis of such an acquaintance. NCFD*: S is non-inferentially justified in believing that P, if (1) S believes that P, (2) S is acquainted with the fact that P, and (3) S bases his belief that P on his acquaintance with the fact that P through the exercise of an appropriate identification ability. The new condition in both principles of having/exercising an appropriate identification ability needs explanation. Nonetheless, this form of Confrontation Foundationalism has potential. The distinction between having the fact that P in our awareness and having the fact that P as assurance for believing that P, which bedeviled both Classical and Neoclassical Confrontation Foundationalism, is captured by the inclusion of an appropriate identification ability. The problem of the speckled hen is avoided and Ability is endorsed. Moreover, unlike both the alternative just considered and Neoclassical Confrontation Foundationalism, these principles don’t include acquaintance with a relation of truth-making correspondence, and, as an extra bonus, they cover cases in which we have non-inferential justification for believing that P without forming the thought that P. Still some may find this alternative a step in the wrong direction. If, as Fumerton (1995: 84) tells us, “the fundamental mistake of externalism is its attempt to reduce the epistemic to the nomological,” don’t NCFP* and NCFD* qualify as a classic blunder with their condition regarding a non-inferential identification ability? Perhaps not. They retain the still unanalyzed/unreduced relation of acquaintance to explain justification. Consider Fumerton’s (1995: 83) comment. In the case of a non-inferentially justified belief, the internalist wants the fact that makes true the belief ‘there before consciousness.’ The externalist can pay lip service to these desires by giving an externalist analysis of being confronted with a fact or having a fact before consciousness, but the internalist is convinced that no attempt to explain that immediacy in terms of nomological relations will succeed.

If the concern is to ground all justification in a direct confrontation with reality that cannot be reduced to nomological terms, NCFP* and NCFD* honor it. Retaining the sui generis relation of acquaintance as an element in non-inferential justification, they simply acknowledge the fact that, to support justification, such states of acquaintance

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

CONFRONTATION FOUNDATIONALISM



must play a certain role in our cognitive system. We must be responsive to them in the right way in the formation of beliefs, if they are to give us assurance of the truth of our beliefs. Moreover, even fans of Neoclassicism are likely to have to accept some nomological conditions for having justification. Recall that the view does not cover the fact that we can have non-inferential justification for believing that we are in pain, when we are in pain but don’t have the thought that we are. The most natural way to extend Neoclassicism to this case is to invoke an acquaintance disposition: If we were to have the thought, we would then be acquainted with it and the truth-making relation between it and our experience. Some may worry that NCFP* and NCFD* give so much away to externalism as to be unmotivated. Why even require the condition that the subject be acquainted with the fact that makes his belief true? Let’s just forget about what it is to have justification—pace Fumerton, that relation isn’t the primary topic for normative epistemologists—and explain justified belief as belief formed by the exercise of the right sort of identification ability. Lots of friendly, engaging externalists will show us the way. This serious concern is not unique to NCFP* and NCFD*. Any form of Confrontation Foundationalism, whether or not it contains some nomological conditions, is open to the challenge that its acquaintance condition(s) is unmotivated. Indeed, Bergmann (2006a) challenges BonJour’s Neoclassical view on just this basis.16 It is a serious challenge that needs to be met, but NCFP* and NCFD* don’t seem to have any special handicaps in responding to it.

5 Conclusion There may be some truth to Confrontation Foundationalism. We have seen though that to mine it successfully, we must acknowledge the distinction between having a direct awareness of a truth-making fact and having such an awareness as assurance for our belief. That distinction is not correctly captured in terms of an awareness of the truth-making correspondence between a truth maker and a thought. It is to be captured in terms of the role that our direct awareness of the truth-making fact plays within a non-inferential identification ability. If Confrontation Foundationalism can be developed within these boundaries, it may well be correct.

References Bergmann, Michael. (2006a). “BonJour’s Dilemma,” Philosophical Studies, 131: 679–93. Bergmann, Michael. (2006b). Justification without Awareness. Oxford: Oxford University Press. BonJour, Laurence. (2003). Epistemic Justification: Internalism vs. Externalism, Foundations vs. Virtues (with Ernest Sosa). Oxford: Oxford University Press. 16

Bergmann’s challenge is an application of his more general challenge to access internalism in (2006b). For a reply to that challenge, see Markie (2009b).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



PETER MARKIE

BonJour, Laurence. (2006). “Replies,” Philosophical Studies, 131. Conee, Earl. (1998). “Seeing the Truth,” Philosophy and Phenomenological Research, 58: 847–57. Fumerton, Richard. (1995). Metaepistemology and Skepticism. Lanham, MD: Rowman and Littlefield. Fumerton, Richard. (2005). “Speckled Hens and Objects of Acquaintance,” Philosophical Perspectives, 19: 121–39. Fumerton, Richard. (2006a). “Direct Realism, Introspection and Cognitive Science,” Philosophy and Phenomenological Research, 73: 680–95. Fumerton, Richard. (2006b). “Epistemic Internalism, Philosophical Assurance, and the Skeptical Predicament.” In Knowledge and Reality: Essays in Honor of Alvin Plantinga, ed. Thomas M. Crisp, Matthew Davidson, and David Vander Laan. Dordrecht: Kluwer, 179–91. Fumerton, Richard. (2009). “Markie, Speckles and Classical Foundationalism,” Philosophy and Phenomenological Research, 79: 208–12. Fumerton, Richard. (2010). “Foundationalist Theories of Epistemic Justification,” Stanford Encyclopedia of Philosophy (Summer 2010 Edition), Edward N. Zalta (ed.), at . Markie, Peter. (2009a). “Classical Foundationalism and Speckled Hens,” Philosophy and Phenomenological Research, 79: 190–206. Markie, Peter. (2009b). “Justification and Awareness,” Philosophical Studies, 146: 361–77. Sosa, Ernest. (2003). Epistemic Justification: Internalism vs. Externalism, Foundations vs. Virtues (with Laurence BonJour). Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

2 Acquaintance and Fallible Non-Inferential Justification Chris Tucker

1 Introduction Acquaintance is a kind of direct awareness. “When one is acquainted with a fact, the fact is there before consciousness. Nothing stands ‘between’ the self and the fact” (Fumerton 1995: 76). Classical foundationalists, whatever else they are, are rather prudish when it comes to non-inferential justification. Since acquaintance puts you in direct contact with reality, it is one of the few things that may satisfy their exacting standards. Let us say that (classical ) acquaintance theory is any version of classical foundationalism that appeals to acquaintance in order to account for NIJ (noninferential justification). Richard Fumerton is arguably the leading proponent, but Evan Fales (1996), Ali Hasan (2013), and Michael Tooley (2013: 325–7), among others, also think of themselves as classical acquaintance theorists. Acquaintance theories are well suited to account for a kind of infallible justification. Why am I justified in believing that I’m in pain? An initially attractive (partial) answer is that I’m acquainted with my pain. But since I can’t be acquainted with what isn’t there, acquaintance with my pain guarantees that I’m in pain. What’s less clear is whether acquaintance theories can account for fallible NIJ. In particular, it’s unclear whether acquaintance theories leave room for misleading NIJ, NIJ to believe something false. I’ll focus on introspective justification, but similar issues arise for a priori justification as well. It would be unfortunate if acquaintance theorists can’t make room for misleading NIJ, for it is plausible that the following cases are possible:

• • •

My sensation has exactly eight speckles, but I have some non-inferential justification for thinking it has exactly seven. The shade of red on the left side of my visual field is ever so slightly distinct from the shade of red on the right side; but I have some non-inferential justification for thinking that they are the same shade. My sensation is an itch that is ever so close to being pain but isn’t, and I nonetheless have some degree of non-inferential justification for thinking that it’s pain.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



CHRIS TUCKER



The line in my visual field is curved ever so slightly, but I have some noninferential justification for thinking that it’s straight.

Fumerton and Hasan see the force of these examples and try to accommodate misleading NIJ from within classical acquaintance theory. I argue that these efforts fail.1 I focus on Fumerton’s work. In section 2, I present Fumerton’s account of noninferential justification. I argue that he can accommodate something like fallible NIJ for true thoughts. In the subsequent three sections, I assess whether he can accommodate fallible NIJ for false thoughts. In section 3, I show that Fumerton’s account of fallible NIJ needs refinement and suggest how it can be refined. In sections 4 and 5, I argue that Fumerton cannot accommodate misleading NIJ.2 In section 6, I argue that Hasan cannot account for misleading NIJ while remaining a classical foundationalist.

2 Fumerton’s Account of Non-Inferential Justification Fumerton’s account of non-inferential justification has two components, one concerning infallible justification and one concerning fallible justification. Fumerton has consistently endorsed the following account for infallible NIJ: (INIJ) S has infallible non-inferential justification for P iff S is acquainted with the fact that P, the thought that P, and the (perfect) correspondence between the fact and thought. If I have infallible NIJ that I’m in pain, it’s because I’m acquainted with the fact that I’m in pain, the thought that I’m in pain, and the (perfect) correspondence between the thought and the fact. At times, Fumerton has assumed that the infallible non-inferential justification is the only kind there is.3 At others, he endorses something like the following account of fallible NIJ: (FNIJ) S has fallible non-inferential justification for P iff the fact that P* is highly similar to the fact that P, and S is acquainted with the fact that P*, the thought that P, and a very high degree of correspondence between the fact that P* and the thought that P.4 1 Fales (1996) also tries to accommodate misleading non-inferential justification from within classical foundationalism, but I don’t have the space to address his views at length. I do, however, briefly discuss his views in note 15. 2 Poston (2010) argues that the possibility of misleading NIJ is in tension with what Fumerton says about the kind of assurance that justification is supposed to provide. I have sympathy with this criticism, but it shows only that Fumerton can’t get everything he wants. I argue that Fumerton fails to make sense of misleading NIJ, regardless of what he ultimately decides concerning assurance. 3 Poston (2010: 370–1) discusses some relevant passages, and Fumerton (2010: 379–80) acknowledges that he has been inconsistent in allowing for fallible justification. 4 Sometimes in connection with fallible justification Fumerton additionally or instead requires that one’s justification for P arise from a fact that P*, where that fact is easily or justifiably confused with the distinct fact that P. He stresses the “easily confused” language in his 2001: 75, but when he elaborates, how easily things are confused seems to boil down to how similar they are (75–6).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND FALLIBLE NON - INFERENTIAL JUSTIFICATION



For illustrative purposes, consider a marginal (painless) itch that is almost a pain. If I’m acquainted with the fact that the sensation is a painless itch, the thought that I’m in pain, and the near correspondence between the fact and the thought, then I have misleading NIJ that the sensation is pain. On the days when Fumerton allows for fallible non-inferential justification, he’s always thought that one has non-inferential justification iff one satisfies either INIJ or FNIJ. I do think that Fumerton’s acquaintance theory can accommodate one type of fallible NIJ. Contrast a searing pain with a marginal one. Fumerton holds that I have stronger justification that I’m in pain when the pain is searing than when it’s very mild, and he claims that in the latter case my justification “is not very strong” (2010: 381). Nonetheless, “There is a sense, of course, in which my justification for believing that I am in pain is infallible when it consists, even in part, of my direct awareness of pain, however marginal that pain might be” (381, emphasis in original). Here Fumerton wants to hold on to two ideas: my justification in the marginal pain case is weak but nonetheless includes my direct awareness of my pain. FNIJ can accommodate both ideas. Since similarity is compatible with identity (identity’s being the limit of similarity), let the fact that P* and the fact that P both be that the sensation is pain. Assume that while Fumerton is acquainted with a high degree, D, of correspondence between the fact that the sensation is pain and the thought that the sensation is pain, he’s not acquainted with the perfect correspondence. FNIJ entails that he has fallible NIJ that the sensation is pain. Calling this justification ‘fallible’ is inappropriate in one sense, because the sensation’s being pain guarantees the truth of my thought. In another sense, however, it arguably counts as fallible justification. What made the pain satisfy FNIJ’s conditions was not that it is pain but only that it’s similar to pain. If a very similar painless itch also corresponds to the pain thought to degree D, then there’s a sense in which he could have been mistaken. Don’t be distracted by whether this justification is best thought of as fallible or infallible (I don’t really care one way or the other). Fumerton wants to allow that there is a lower grade of justification than the one identified by INIJ. I’m granting that he can account for a lower grade of non-inferential justification for true thoughts by allowing that one is acquainted with a high (but not perfect) degree of correspondence between the fact that P* and the thought that P. What I deny is that Fumerton’s acquaintance theory can accommodate a grade of non-inferential justification for false thoughts.

3 A Fixable Problem Recall Fumerton’s account of fallible non-inferential justification: (FNIJ) S has fallible justification for P iff the fact that P* is highly similar to the fact that P, and S is acquainted with the fact that P*, the thought that P, and a very high degree of correspondence between the fact that P* and the thought that P.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



CHRIS TUCKER

In the previous section, we focused on a case that plausibly involved fallible NIJ for a true thought. In this section, we focus on a case in which we might have fallible NIJ for a false thought. Consider again a marginal itch that is very similar to pain without being a pain. Now suppose that: A.

I’m acquainted with the fact that my sensation is a painless itch, my thought that the sensation is a pain, and the very high degree of correspondence between the fact and the thought.

Since I’ve satisfied the conditions of FNIJ, Fumerton holds that A is sufficient for having (fallible) non-inferential justification that I’m in pain. Perhaps A can contribute to my having NIJ in some circumstances. What I argue in this section is that it’s not sufficient: there are some circumstances in which A is true but I lack NIJ for thinking that I’m in pain. While the painless itch and the thought that this sensation is pain do correspond to a very high degree, the correspondence isn’t perfect. Suppose that, in addition to A, it’s also true that: B.

I’m acquainted with the fact that my painless itch fails to have perfect correspondence with my thought that the sensation is pain.

Does B’s truth prevent A from providing me with NIJ to believe that the sensation is pain? I don’t think so. At first glance, it may seem that to be acquainted with imperfect correspondence between a thought and a fact is to be acquainted with the thought’s falsehood. But that’s not quite right. Suppose I’m acquainted with the fact that my sensation has either 13 or 14 speckles, and I’m not acquainted with any fact that is more determinate. The thought that it has exactly 13 speckles will correspond to a high degree to the disjunctive fact. In such a case, I might be acquainted with the failure of perfect correspondence, but it doesn’t follow that I’m acquainted with the thought’s falsehood. We’ll revisit imperfect correspondence in section 5. I’ll argue that it isn’t capable of supporting misleading NIJ, but the problem has nothing to do with the possibility of being acquainted with the failure of perfect correspondence. Acquaintance with imperfect correspondence may not be a problem, but my sensation doesn’t just imperfectly correspond to the thought that it’s pain; it also perfectly corresponds to the thought that it’s not pain. Consider the possibility that A is true and so is: C.

I’m acquainted with perfect correspondence between the fact that the sensation is a painless itch and the thought that the sensation is not pain.

Even if A can provide me with fallible NIJ in some circumstances, it’s not obvious that it can provide me with fallible NIJ when C is also true.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND FALLIBLE NON - INFERENTIAL JUSTIFICATION



We can make matters worse. Suppose that A is true and so is: D.

I’m acquainted with the fact that my sensation is a painless itch, my thought that the sensation is not pain, and the perfect correspondence between them.

According to INIJ, I have infallible NIJ that the sensation is not pain. When both A and D are true, Fumerton’s account of non-inferential justification entails that I have (undefeated) fallible justification that the sensation is pain and (undefeated) infallible justification that the sensation is not pain. This result is counterintuitive, in part because it violates: General Rule: it’s impossible for one to have undefeated justification for believing P while also having undefeated justification for believing ~P. While General Rule may have exceptions, alleged exceptions are guilty until proven innocent. Philosophers of language provide one way to argue that some exceptions are innocent. Assume Russellianism about proper names, i.e., assume that the meaning of a proper name just is its referent. On this view, the following two propositions are identical, because ‘Mark Twain’ and ‘Samuel Clemens’ have the same referent: (MT) Mark Twain wrote Huckleberry Finn, and (SC) Samuel Clemens wrote Huckleberry Finn. Given Russellianism about proper names, it doesn’t seem crazy at all to think that one might have (undefeated) justification to believe MT and (undefeated) justification to believe ~SC. Yet, since MT and SC are the same proposition, we have a case in which it is possible to have undefeated justification to believe P but also undefeated justification to believe ~P. Given Russellianism about proper names (or, more precisely, the analogous thesis concerning mental content), I think it’s plausible that there is an important class of exceptions to General Rule. This Russellian strategy provides cases that intuitively seem to be exceptions to General Rule, and it provides a theoretical explanation of how the rule could be rationally violated, namely that a subject might reasonably fail to know that certain propositions are identical. Fumerton, however, rejects Russellianism about proper names (2013: ch. 5). On his view, “If you really are using ‘a’ and ‘b’ as pure referring terms, then the statement “a=b” will strike you as utterly trivial” (2013: 187). Since Mark Twain is identical to Samuel Clemens is not trivial, these proper names aren’t pure referring terms. Perhaps Fumerton will argue that the cases I’m interested in—e.g., cases in which both A and D are true—are nonetheless legitimate exceptions to General Rule. If so, he needs to follow the example of the philosophers of language and find some way of motivating the idea that these cases really are exceptions. Unless Fumerton can motivate an exception to General Rule, he needs to revise his account of fallible NIJ. What our discussion reveals is that satisfying the conditions identified by FNIJ isn’t sufficient for NIJ. Satisfying those conditions doesn’t provide

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



CHRIS TUCKER

NIJ for P when one also has infallible NIJ for ~P. This problem shows that FNIJ needs refinement, not that it needs to be abandoned. One simple fix is to add a no defeater condition to FNIJ. I’m fine with this fix, but I don’t expect Fumerton to greet it with enthusiasm (e.g., 1985: 60–1; 2010: 385; personal correspondence). Whatever Fumerton ultimately decides, I’ll assume that he can augment FNIJ in a way that addresses the problem raised in this section.

4 Acquaintance with Contradicting Facts I now begin to criticize Fumerton’s FNIJ in earnest. In this section, I argue that FNIJ must be understood in a way that respects a certain constraint. This constraint concerns which facts can contribute to non-inferential justification. In the next section, I argue that when the constraint is respected, FNIJ cannot account for misleading NIJ. According to Fumerton, acquaintance with the truth-maker for a proposition is an essential component of any infallible justification for believing that proposition. Fumerton’s account of fallible justification, however, does not require acquaintance with the truth-maker; it allows acquaintance with facts that aren’t truth-makers to contribute to fallible NIJ. So far, so good. But Fumerton should not allow acquaintance with false-makers to contribute to one’s justification for believing a proposition. Suppose I’m acquainted with the fact that my sensation is a painless itch. This fact is a false-maker for my thought that the sensation is a pain; therefore, it cannot contribute to any justification I have for thinking that the sensation is a pain. Let us say that a contradicting fact for P is any fact that non-trivially entails ~P.5 Perhaps there are more contradicting facts for P than there are false-makers for P. I tentatively think the worry generalizes to any contradicting fact. In other words, I propose: Constraint: If S’s acquaintance with the fact that Q non-trivially entails the fact that ~P, then S’s acquaintance with the fact that Q can’t constitute, in whole or in part, any justification for P that S has. I expect Constraint to sound plausible to most epistemologists. Indeed, Huemer (2007: 35) suggests something stronger, namely that acquaintance with such a fact might prevent me from having justification to believe that the sensation is a pain. While I think the stronger claim is plausible, weaker is safer. Constraint leaves open that I can be justified in believing that the sensation is pain even when I’m acquainted with the fact that it’s a painless itch. What it denies is that my acquaintance with the “[O]ne fact entails another when the proposition the former makes true entails the proposition the latter makes true” (Fumerton 2010: 137, nt. 3). A proposition A non-trivially entails another proposition B iff A entails B and the modal profile or conceptual content of both A and B contribute to the entailment. If B is necessarily true, then its modal profile guarantees, by itself, that every proposition entails it. But only some propositions will non-trivially entail B, e.g., the conjoined premises of an elegant proof. 5

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND FALLIBLE NON - INFERENTIAL JUSTIFICATION



sensation’s being a painless itch can contribute to my justification for believing that it’s a pain. Even some acquaintance theorists find such a weak constraint plausible.6 It’s harder than you might expect to explain why Constraint is so plausible. Let us say that a justification for P is essentially errant or anti-infallible iff it guarantees (i.e. non-trivially entails) the falsehood of the proposition it justifies. It’s popular (if not mandatory) to allow that we can be non-inferentially justified in believing necessary falsehoods. But it doesn’t follow that a theory can properly allow X to provide justification for P even though X guarantees that P is false. If Fumerton were to reject Constraint, then he would allow that we can have anti-infallible justification. He would allow that our justification for P can guarantee that P is false. Is this a problem? It may seem obvious that anti-infallible justification is impossible, but we should tread carefully lest we rule out our own views. Suppose that a belief is justified if it is produced by a properly functioning faculty. Now suppose that, given an individual subject’s evidence, a person’s properly functioning inferential faculty produces the belief that there is no such thing as proper function. Our crude proper functionalism entails that the belief is justified. The existence of this justification guarantees that the belief is false, for the justification is provided by proper function and the belief denies that there is any such thing. Or suppose that process reliabilism is true, but that given the evidence at my disposal, my reliable inferential faculty produces the belief that the inferential processes human beings use aren’t sufficiently reliable to produce justified beliefs. Here again we have a case of anti-infallible justification: the reliability of my inferential faculty is at least part of what justifies my belief that no inferential process is reliable enough to produce justification. Here’s one final example which hits closer to home. Suppose Phenomenal Conservatism is true and that I’m justified in believing that there is no such thing as a seeming, because it seems to me that there is no such thing as a seeming. If we rule out any sort of anti-infallible justification as impossible, then proper functionalists, reliabilists, and phenomenal conservatives should give up and go home, for they make anti-infallible justification possible in some circumstances. If there is a problem with violating Constraint, it’s not merely that Fumerton would allow for the possibility of anti-infallible justification. It’s that he would allow for the possibility of a certain kind of anti-infallible justification. Let’s look more carefully at the type of anti-infallible justification allowed by proper functionalists, reliabilists, and phenomenal conservatives. In each case, the anti-infallible justification results from a contradicting fact that is external to the subject’s perspective. In each case, some unknown fact about the way the belief was produced (the belief was produced by proper function, the process was reliable enough, the belief was caused by a seeming) entails that the believed proposition is false.

6

Hasan vouches for its plausibility in section 6 of Hasan and Fumerton (2014).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



CHRIS TUCKER

The cases of anti-infallible justification that seem especially problematic are due to contradicting facts that are internal to the subject’s perspective. For example, suppose that Mary and Sherry are identical twins. Suppose I know them well enough to know that, on a given occasion, I’m talking to Mary. My knowledge that I’m talking to Mary can’t provide justification for thinking that I’m talking to Sherry no matter how similar they may be. My knowledge puts a certain fact in my perspective—this person is Mary—which guarantees that the person is not Sherry. Consequently, my knowledge that the person is Mary can’t justify me in believing that the person is Sherry. The problem with violating Constraint is that it makes internal anti-infallible justification possible. To violate Constraint is to allow facts in my perspective that contradict P to contribute to P’s justification. Suppose I’m acquainted with the fact that my sensation is a painless itch. If I’m acquainted with this fact, the fact is within my perspective. Indeed, you might think your being acquainted with a fact, having that fact directly before your consciousness, is the way to get a fact in your perspective par excellence. To allow acquaintance with this fact to justify my belief that the sensation is pain is to allow a fact within my perspective that contradicts P to justify P. I’ll bet that you didn’t find the kind of anti-infallible justification allowed by proper functionalism, reliabilism, and Phenomenal Conservatism all that worrisome. And I’ll bet that you find the sort of anti-infallible justification that would result from violating Constraint to be far more suspicious. My explanation is that when the contradicting fact is outside the subject’s perspective, anti-infallible justification is no problem. When the contradictory fact is within the subject’s perspective, as it is when one is acquainted with a contradicting fact, then it’s a problem. I’m not entirely happy with this way of putting things, because even when I find ‘subject’s perspective’ talk illuminating, I also find it a bit vague and metaphorical. But this somewhat unhappy way of putting things is, I think, enough for us to latch on to what seems problematic about violating Constraint and why this problem does not generalize to rival epistemological theories.7

5 Fallible NIJ without Contradicting Facts? If fallible non-inferential justification can’t consist in acquaintance with contradicting facts, in what can it consist? How we word our answer will depend on how we understand the nature of sensations. Fumerton (2005: 127) rejects the idea that sensory states are intentional (representational). Instead, he seems to think that 7

If Russellianism about proper names is true, then we may have to tolerate some cases of internal antiinfallible justification. For example, a justified belief that (MT) Mark Twain wrote Huckleberry Finn might be used in an argument to establish that (~SC) Samuel Clemens did not write Huckleberry Finn, even though MT and SC are the same proposition. As I mentioned in the previous section, however, Fumerton rejects Russellianism about proper names.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND FALLIBLE NON - INFERENTIAL JUSTIFICATION



they are either sense data or ways of appearing (2005: 125, 127). Fumerton (2005: 133–4) sometimes assumes for convenience that sensory states are sense data. I will do likewise, because he “suspect[s] one can find ways of translating the relevant points into the language favored by the appearing theorist” (2005: 134). To say that a sensory state is a sense datum is to say that it is a special kind of object. “When someone hallucinates something red and round, there may be no physical object present, but there is something present to consciousness and that thing really is phenomenally red and round” (2005: 125, emphasis original). If sensory states are sense data, we need to distinguish between de re acquaintance and de facto acquaintance. De re acquaintance is acquaintance with an object and de facto acquaintance is acquaintance with a fact. De re acquaintance comes cheaply. Consider an analogy with de re seeing. If I see any truth about the squirrel, then I count as seeing the squirrel. If I see that it is moving, but I have no idea what kind of thing is moving, then I count as seeing the squirrel. Likewise, I’m acquainted with a sensation if I’m acquainted with any fact about it. I’m acquainted with a 48-speckled sensation if I’m acquainted with the fact that the sensation has many speckles or the fact that it has at least one. De facto acquaintance does not come so cheaply. Just as I can see the squirrel without seeing that it’s a squirrel, I can be acquainted with a 48-speckled sensation without being acquainted with the fact that it has 48 speckles (cf. Fumerton, 2005: 133–5). I might be acquainted with a painless itch without being acquainted with the fact that it’s a painless itch. For Fumerton, it is acquaintance with facts, not objects, that contributes to noninferential justification. For fallible NIJ, Fumerton’s idea is that acquaintance with certain facts about painless itches can give rise to non-inferential justification that the sensation is a pain. Yet if I’m acquainted with some fact about the painless itch, then I’m acquainted with the painless itch and am, thereby, acquainted with something that guarantees that my thought is false. Is this a problem for Fumerton? I doubt it. If I merely see a squirrel (and do not see that it is a squirrel), this seeing does nothing to prevent me from being justified in believing that the object is a chipmunk. If I merely am acquainted with a painless itch (and am not acquainted with the fact that it is a painless itch), this acquaintance does nothing to prevent me from being justified in believing that the object is a pain. In the parlance from the previous section, your being acquainted with a painless itch isn’t enough to put in your perspective the fact that it is a painless itch. So which facts about the painless itch can contribute to non-inferential justification that the sensation is pain? According to Constraint, it can’t be the fact that it is a painless itch. Here are some possibilities:

• •

the sensation’s being pain-like, where being pain-like is compatible with being pain; the sensation’s having features that are (objectively) similar to those of borderline pains, where objective similarity is compatible with identity;

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



CHRIS TUCKER

• •

the sensation’s sharing features with pains; the sensation’s having features X, Y, and Z, where X, Y, and Z are features that pains also have.

In each of the first three possibilities, the relevant fact involves some comparison with pains. In the fourth possibility, the fact is just the sensation’s having certain features, where potentially unbeknownst to the subject those features are also had by pains. To have fallible NIJ for the thought that I’m in pain, Fumerton requires that I have acquaintance with a fact that is ‘very similar’ to the fact that I’m in pain. Each of these possibilities can make at least some sense of this idea. A sensation’s being pain-like (at least when the pain-likeness is very strong) is presumably very similar to a sensation’s being pain. A sensation’s having X, Y, Z when these features are prototypical of pain might count as being very similar to a sensation’s being pain. Hence, we have a way of interpreting Fumerton that respects both Constraint and what little Fumerton says about fallible NIJ. In what follows, I don’t distinguish between the four possibilities. They all suffer from the same basic problem. Fumerton’s account of fallible NIJ, as we are currently interpreting it, holds that awareness of a sensation’s similarity to pain gets us NIJ for thinking we are in pain, at least when we are also acquainted with the thought that the sensation is a pain and the near correspondence between the fact and thought. Yet awareness of X’s similarity to things of type T provides justification for thinking X is T only if one has some antecedent justification for believing that X’s having these similarities reliably indicates being of type T.8 Perhaps a sensation’s having certain similarities with pains guarantees that the sensation is a pain, and one can know that this guarantee obtains a priori. But we are considering whether Fumerton can account for misleading noninferential justification. For that purpose, we need to consider the kinds of similarities to pain that even painless itches can have. For such similarities, it is contingent whether having those similarities to pain reliably indicates being a pain. Hence, if my sensation’s similarity to pain provides me with justification to believe it’s pain, then I must have some antecedent empirical (e.g., perceptual or introspective9) justification for thinking that those similarities reliably indicate being pain. Consider some analogies. My dusty minivan shares many similarities with my friend’s minivan, but my minivan is a Toyota and his is a Honda. On the other hand, my minivan doesn’t have much in common with the futuristically sleek and sporty FT-1, but the latter is also a Toyota. Professional cricket matches share many similarities with what takes place in Yankee stadium, but cricket is a lame sport and

8 A coherentist or holist can sensibly reject my claim that one needs this justification antecedently. It’s a feature of their view that justification can emerge from certain collections of beliefs even if none of those beliefs are justified antecedently to any other. Fumerton, though, doesn’t seem to have much sympathy with such positions (e.g., see his 1995, ch. 5). 9 For simplicity, I’ll also treat memorial and testimonial justification as empirical when they don’t concern propositions for which we can have a priori justification.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND FALLIBLE NON - INFERENTIAL JUSTIFICATION



baseball isn’t. On the other hand, what the kids in the street are doing bears little resemblance to what the Yankees are doing, but both activities arguably count as baseball. It takes empirical investigation to determine which similarities to things of type T reliably indicate being of type T. Likewise, it takes empirical investigation to determine whether X’s sharing, say, certain phenomenal features with pain makes it likely that X is pain. And without some reason to think that these phenomenal features make it likely that X is pain, it’s hard to see why awareness of these features would provide us with justification for thinking that X is pain. I’ve argued in this section that, without some empirical justification for thinking that the similarities I’m aware of make it likely that the sensation is pain, then awareness of those similarities can’t justify me in thinking that the sensation is pain. But suppose that I am antecedently justified in believing that those similarities make it likely that the sensation is in pain. For example, consider the following case: E.

(i) A sensation’s having phenomenal feature F is similar to the sensation’s being pain, (ii) I am antecedently and empirically justified in believing that things which instantiate F tend to be pains, and (iii) I am acquainted with the fact that my sensation has F, the thought that my sensation is pain, and the very high degree of correspondence between the thought and fact.

Perhaps E does provide me with justification for thinking that the sensation is pain, even if it is a painless itch. This would amount to misleading justification, but it’s at least partly inferential. I never doubted that Fumerton could make sense of misleading justification when that justification is partly inferential. The issue in the chapter is whether Fumerton can account for misleading non-inferential justification. Consider E(ii). Given Fumerton’s view, how could I have such justification? Perhaps I’ve noticed over time that these similarities are more often exemplified by pains than non-pains. In other words, perhaps I have some memorial justification that these similarities reliably indicate pain. But, for Fumerton, memorial justification is a type of inferential justification. Any justification composed in part of inferential justification is at least partly inferential. Can I have non-inferential justification of the relevant regularity? Given that the regularity holds contingently, it’s doubtful that I can be acquainted with it. If I can’t be acquainted with that regularity, then Fumerton holds that I can’t have infallible NIJ for thinking it obtains. Independently of the concern I’m pressing in this section, it’s also doubtful that Fumerton’s account of fallible NIJ can account for this justification. To have fallible justification that this contingent regularity holds, I need to have acquaintance with a similar fact. But what fact is similar to this contingent regularity but nonetheless is something with which I can be acquainted? I haven’t the slightest idea. It seems, then, that any empirical justification I have for thinking that a sensation’s having F indicates being pain must be at least partly inferential. If so, then any justification E might provide for thinking that the sensation is pain is at least partly inferential. Acquaintance with similar facts might

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



CHRIS TUCKER

help account for some kind of misleading justification, but not misleading noninferential justification.

6 Hasan to the Rescue? Fumerton tries to make room for misleading non-inferential justification within his classical acquaintance theory, but his efforts seem to be in vain. Yet Fumerton isn’t the only classical acquaintance theorist who tries to make room for fallible noninferential justification. In this section, I consider the views of Ali Hasan. I’ll argue that his view cannot accommodate misleading non-inferential justification while remaining true to classical foundationalism.

6.1 Hasan’s Principle: the first horn Hasan tries to make room for fallible introspective justification by appealing to: Hasan’s Principle (HP): “[(i)] If one is directly aware of x’s seeming to be G,10 and [(ii)] if x’s seeming to be G renders it probable for one that it is G,11 then [(iii)] one thereby has at least some defeasible justification for believing that x is G” (Hasan 2013: 130; cf. Hasan and Fumerton 2014, sec 6). The first condition of HP requires acquaintance, but HP represents a big departure from Fumerton’s acquaintance theory. For Fumerton, there is supposed to be a deep parallel between the infallible and fallible cases of justification. In the infallible cases, we are acquainted with the fact that the sensation is a pain. In misleading fallible justification, acquaintance with some fact about the marginal itch provides us with justification for believing that the sensation is pain. In both cases, what provides us with justification is, essentially, acquaintance with facts about some sensation and some degree of correspondence between those facts and the thought that the sensation is a pain. According to HP what gives us fallible justification isn’t acquaintance with the fact that the sensation is pain (pain-like) and the (near) correspondence with the fact and thought: it’s acquaintance with the seeming that the sensation is a pain. And, in principle, it can seem to one that x is a pain even if x is a pleasurable feeling of deep fulfillment. I do think that HP can account for misleading justification. It holds, roughly, that one has justification that one is in pain when one is acquainted with a seeming and 10 I’ve omitted a parenthetical clause for simplicity’s sake, but I’m also a bit puzzled by the clause. The clause allows inclinations to believe, in addition to seemings, to contribute to fallible justification, provided that you are acquainted with the inclination. I doubt, however, that inclinations are the sort of thing to which one can be acquainted. One can be directly aware of a feeling that one is inclined, but not the inclination itself. If this is right, then HP is more committed on the ontology of seemings than Hasan (2013: 130, nt. 20) seems to think. 11 Hasan builds more into (ii) than its wording suggests. He requires that one must also “grasp” the relevant probability relation (2013: 129, 132). I don’t think my complaints about HP turn on whether we need to grasp the relevant probability relation, but I don’t have the space to defend this point.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND FALLIBLE NON - INFERENTIAL JUSTIFICATION



that seeming makes it probable for one that P. I might satisfy these conditions even if my seeming leads me astray. The question is whether HP can account for misleading non-inferential justification without abandoning classical foundationalism. HP faces a dilemma. It concerns condition (ii). Suppose that a given seeming makes it probable for one that x is a G. Does this probability hold contingently or necessarily between its relata? If it’s the former, then HP can’t account for misleading non-inferential justification. If it’s the latter, then HP abandons classical foundationalism. Suppose that the relevant kind of probability is contingent. Let ‘XG’ be the proposition that x is a G and ‘SXG’ be the seeming that x is a G. What could explain why SXG makes XG probable? Perhaps it’s that I have some memorial justification concerning the track record of my seemings: I’ve noticed that when I have SXG, XG is usually true. Yet classical acquaintance theorists tend to think that memorial justification is inferential justification. If memorial justification for some proposition—e.g., usually when I have SXG, XG is true—partly explains why the seeming makes XG probable, then any justification provided by HP is at least partly inferential. Hasan’s favored account of the relevant probabilistic connection apparently boils down to XG’s truth being the best explanation of why I have SXG: It is plausible that the shape’s seeming to me to be pink makes it probable for me that it is pink, for it is plausible that the best explanation available to me of the fact that some phenomenal property attended to seems to have some phenomenal character F is that it does have character F. (2013: 131)

When I discuss the second horn in the next section, we’ll consider what follows if it’s a necessary truth that XG is best explained by SXG. If it’s a contingent matter whether XG’s truth is the best explanation of SXG for one, then what makes it the best explanation of SXG? The traditional view, I take it, is that one’s background knowledge is what determines whether something is the best explanation of something else. If I’ve never even heard of quantum mechanics, then quantum states won’t be the best explanation of anything for me. If, on the other hand, I have justified beliefs about quantum mechanics, then maybe quantum states can be the best explanation of some fact for me. Given this traditional picture, then, HP can account for justification one has to infer the best explanation.12 Yet it can’t account for non-inferential justification.

12 Since part of the ‘inference base’ is a non-doxastic state, a seeming, I follow Hasan in using ‘infers’ and its cognates loosely (Hasan 2013: 132, nt. 27). The key point is that the justification that emerges isn’t purely non-inferential. It’s not purely non-inferential justification because, in addition to the seeming that X is a G, one needs antecedent justification for thinking that X’s being G is the best (and presumably also good) explanation of why it seems that X is G. This antecedent justification will presumably be empirical since, at this point, we are assuming that the relevant probability relation holds contingently.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



CHRIS TUCKER

6.2 Hasan’s Principle: the second horn Hasan’s Principle holds that (i) if you’re acquainted with SXG and (ii) SXG makes it probable for one that XG, then you have defeasible justification for XG. We saw that if the probability relation in (ii) is contingent, then HP cannot account for misleading non-inferential justification. We should consider, then, what happens when we think of the relevant probability as one that holds necessarily between SXG and XG. When we do, we’ll find that HP would abandon classical foundationalism. One very permissive alternative to classical foundationalism is: Phenomenal Conservatism (PC): if it seems to S that x is a G, then S thereby has at least some defeasible justification for believing that x is a G.13 You should endorse PC; it’s what all the cool kids are doing these days. But that’s a point for another day. For today, the point is this: while HP doesn’t collapse into PC, it’s too close to PC to be something that will be attractive from the point of view of classical foundationalism. A major objection to PC is that it allows seemings caused by wishful thinking to provide prima facie justification for their contents (see, e.g., Markie 2005: 356–7; Siegel 2012). HP will also be subject to such examples, because as Hasan notices, seemings can “be influenced by background cognitive states” (2013: 131). In my 2014, I argued that if this sort of objection to PC raises a genuine problem for PC, it raises an equal or bigger problem for all of its rivals . . . except for one (58–9). I conceded that a Fumerton-style acquaintance theory may avoid these problems, so I managed to say something nice about Fumerton’s view on that particular afternoon. But if Fumerton follows Hasan in endorsing PC, I’d have to take back that one nice thing I said about Fumerton’s view. PC and HP both allow wishfully produced seemings to justify their contents, but perhaps there are other differences between them that make HP more palatable to the classical foundationalist. One difference is that HP concerns only those seemings that make it probable for one that x is a G. When the relevant kind of probability is contingent, this is a significant constraint. Some seemings to which PC grants justificatory power will be denied such power by HP. We are considering the possibility, however, that the relevant sort of probability holds necessarily. If it does hold necessarily, then the mention of probability in HP may not reflect substantive disagreement with PC. The proponent of PC may think there must be some sense in which seemings necessarily make their contents probable, otherwise why think that seemings necessarily justify their contents? A second difference is that HP concerns only those seemings that one is acquainted with. If one can have seemings with which one is not acquainted, then

13 I articulate Phenomenal Conservatism this way to make it closer to the wording of HP. For defenses of Phenomenal Conservatism and more official articulations of the view, see Huemer (2007), Lycan (2013), and Tucker (2010, 2011).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND FALLIBLE NON - INFERENTIAL JUSTIFICATION



only PC allows that such seemings have justificatory power. I’m not sure that there are any such seemings, but suppose that I can have a seeming that I’m not aware of. Speaking as one who defends PC, the seemings that I’m most confident that provide justification are the ones that I’m acquainted with; I don’t have a settled opinion concerning the other ones. If it seems to me that P, but I’m not in any way aware of this seeming, then perhaps that seeming can’t justify. If so, then PC is false but might as well be true. Except for appeasing some controversial access requirement—e.g., E can justify P only if S is aware of E—there are no significant (dis)advantages to granting justificatory power to only those seemings with which one is acquainted. There is supposed to be a third difference between HP and PC, namely that HP grants only introspective seemings the power to justify whereas PC grants such power to all seemings. This difference, if genuine, would make HP considerably more attractive to classical foundationalists; however, there is nothing in the wording of HP that would justify such a restriction. So why does HP apply only to introspective seemings? Hasan explains: Can something similar to [HP] be used to defend the view that external world beliefs could be empirically foundational? Perhaps, but unless we can be directly aware of external world objects or their properties, something that classical foundationalists deny, the principle will have to specify some epistemically relevant relation between phenomenal or mental properties we can be directly aware of and non-mental properties of the external world, and this seems bound to introduce more complications and sources of error than in the case of introspection. (2013: 132)

For HP to grant perceptual seemings justificatory power, Hasan says we’d need direct awareness of either (a) some epistemically relevant relation between the mental seemings and the non-mental physical world or (b) physical objects and their properties. Since he’s not optimistic that we can be acquainted with either one, he concludes that HP doesn’t grant any perceptual seeming the power to justify. Consider (a) first. It’s simply false that we’d need to be acquainted with a connection between perceptual seemings and the physical world. At most what we’d need is acquaintance with a connection between perceptual seemings and their contents, contents which concern physical objects. Given our assumption in this sub-section that introspective seemings necessarily make their contents probable, why can’t perceptual seemings necessarily make their contents probable? And if they necessarily make their contents probable, we presumably can be acquainted with that connection. Now consider (b). Recall how HP applies to introspective justification. HP allows me to have justification that I’m in pain even if I’m acquainted with neither the fact that I’m in pain nor the fact that I’m in some other mental state that is a lot like pain. This feature of HP is what makes HP such a big departure from Fumerton’s work.14 What

14

Hasan apparently fails to appreciate this departure when he appeals to Fumerton’s account of fallible justification (2013: 132, nt. 29).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



CHRIS TUCKER

matters, for HP, is whether I’m acquainted with my seeming that I’m in pain. But I can be equally well acquainted with the way things seem, regardless of whether they concern what’s going on in my own mind or what’s going on around me. So why do I need to be acquainted with physical objects in order for HP to grant justificatory power to seemings? Hasan will argue that we have a better explanation of a seeming when the seeming’s content represents something we can be directly aware of. He contends that perceptual processing “seems bound to introduce more complications and sources of error than in the case of introspection” (2013: 132). Maybe, but it’s still contingent whether introspective seemings are more reliable than perceptual ones. Indeed, it seems more important for our survival that we know when there’s a tiger hunting us or when there’s food available than that we can reliably distinguish 47 speckles in our visual field from 48. We might expect, then, that some perceptual seemings are better explained by the truth of their contents than some introspective seemings. Despite these points, suppose that introspective seemings are always better explained by the truth of their contents than perceptual ones. If so, we’d have some explanation for why introspective seemings provide more justification than perceptual ones. It does not follow, however, that appeals to the best explanation are good enough to support justification in the introspective cases but not good enough in the perceptual cases. If Hasan justifies HP’s truth by appealing to best explanation considerations, it’s unclear how many seemings are being granted justificatory power. Nonetheless, it apparently grants justificatory power to many perceptual ones. My complaint about HP is not that it’s false: it’s that it can’t account for misleading NIJ from within classical foundationalism. If HP’s second condition requires the relevant probability relation to hold contingently, then HP can’t account for misleading non-inferential justification. If the second condition instead requires the relevant probability relation to hold necessarily, HP has too much in common with Phenomenal Conservatism to be attractive from the point of view of classical foundationalism. Like Phenomenal Conservatism, it allows seemings produced by wishful thinking to have justificatory power. And, while it may not grant justificatory power to the full range of seemings as PC does, it grants justificatory power to many perceptual ones, which is a substantial departure from classical doctrine.15

15 Fales (1996) also tries to make room for fallible non-inferential justification. The probabilistic nature of such justification, he says, “is itself an element of the experience that grounds the judgment” (175). One way of understanding this view treats Fales as relying on a view like Phenomenal Conservatism. When Huemer talks about how we have justification for believing we are in pain when we are in pain, he allows that the seeming that one is in pain may be token identical to the pain, i.e. the pain can be a seeming of itself that it is a pain (2007: 46). Perhaps Fales takes Huemer’s view one step further: painless itches can also be token identical to the seeming that one is in pain. Fales encourages this reading when talking about degrees of a priori justification. There he approvingly cites Bealer’s (1992) account of intuition (Fales 1996: 185, nt. 28; nt. referenced on 161). For Bealer (1992: 101–4), intuition is just an intellectual seeming to be understood analogously to perceptual seemings. On this PC-like reading of his view, Fales allows introspective and a priori seemings to justify their contents. But then, like Hasan, he must

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND FALLIBLE NON - INFERENTIAL JUSTIFICATION



7 Conclusion This chapter has concerned those versions of classical foundationalism that understand non-inferential justification at least partly in terms of acquaintance. Such views are well suited to account for infallible non-inferential justification. If my acquaintance with my pain provides me with justification that I’m in pain, then my justification can’t lead me astray. Yet there are a number of cases in which it is plausible that we can have non-inferential justification that does lead us astray. We considered two attempts to account for misleading non-inferential justification by appealing to acquaintance. We paid especially close attention to the views of Fumerton, but we also considered what Hasan had to say. Neither was able to account for misleading non-inferential justification while remaining a classical foundationalist.16

References Bealer, George. (1992). “The Incoherence of Empiricism,” Proceedings of the Aristotelian Society, 66: 99–138. Fales, Evan. (1996). A Defense of the Given. Lanham, MD: Rowman and Littlefield. Fumerton, Richard. (1985). Metaphysical and Epistemological Problems of Perception. Lincoln: University of Nebraska Press. Fumerton, Richard. (1995). Metaepistemology and Skepticism. Lanham, MD: Rowman and Littlefield. Fumerton, Richard. (2001). “Replies to Pollock and Plantinga,” in Michael DePaul (ed.), Resurrecting Old-Fashioned Foundationalism. Lanham, MD: Rowman and Littlefield, 69–78. Fumerton, Richard. (2005). “Speckled Hens and Objects of Acquaintance,” Philosophical Perspectives, 19: 121–38. Fumerton, Richard. (2010). “Poston on Similarity and Acquaintance,” Philosophical Studies, 147: 379–86. Fumerton, Richard. (2013). Knowledge, Thought, and the Case for Dualism. New York: Cambridge University Press. Hasan, Ali. (2013). “Phenomenal Conservatism, Classical Foundationalism, and Internalist Justification,” Philosophical Studies, 162: 119–41. Hasan, Ali and Richard Fumerton. (2014). “Knowledge by Acquaintance vs. Knowledge by Description,” Stanford Encyclopedia of Philosophy. Stable URL: . Huemer, Michael. (2007). “Compassionate Phenomenal Conservatism,” Philosophy and Phenomenological Research, 74: 30–55.

explain why perceptual seemings can’t justify their contents, if his view is to be attractive from the point of view of classical foundationalism. I suspect that he doesn’t intend his view to be interpreted in a PC-like way. Yet if not, it’s unclear how probabilistic justification that one is in pain can itself be ‘an element of ’ a painless itch. 16

Thanks to Michael Bergmann, Richard Fumerton, Ali Hasan, Ted Poston, and the audience of the 2014 Orange Beach Epistemology Workshop for helpful comments.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



CHRIS TUCKER

Lycan, William. (2013). “Phenomenal Conservatism and the Principle of Credulity,” in Chris Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, 293–305. Markie, Peter. (2005). “The Mystery of Perceptual Justification,” Philosophical Studies, 126: 347–73. Poston, Ted. (2010). “Similarity and Acquaintance: A Dilemma,” Philosophical Studies, 147: 369–78. Siegel, Susanna. (2012). “Cognitive Penetrability and Perceptual Justification,” Nous, 46: 201–22. Tooley, Michael. (2013). “Michael Huemer and the Principle of Phenomenal Conservatism,” in Chris Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, 306–27. Tucker, Chris. (2010). “Why Open-Minded People Should Endorse Dogmatism,” Philosophical Perspectives, 24: 529–45. Tucker, Chris. (2011). “Phenomenal Conservatism and Evidentialism in Religious Epistemology,” in Raymond Van Arragon and Kelly James Clark (eds), Evidence and Religious Belief. Oxford: Oxford University Press, 52–73. Tucker, Chris. (2014). “If Dogmatists Have a Problem with Cognitive Penetration, You Do Too,” Dialectica, 68: 35–62.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

3 Foundational Justification, Meta-Justification, and Fumertonian Acquaintance Matthias Steup

According to foundationalism, the following two theses are true: first, a subject’s belief system is divided into a superstructure, comprised of inferentially justified beliefs, and a foundation, comprised of non-inferentially justified beliefs. Second, all beliefs in the superstructure ultimately receive their justification from beliefs in the foundation. Competing versions of foundationalism differ from each other in the way they answer the following three questions (see Feldman 2003: 52): (1) How do foundational beliefs justify beliefs in the superstructure? (2) What makes a belief foundational? (3) Which beliefs are in the foundation? Cartesian foundationalism answers these questions thus:

• • •

Foundational beliefs justify beliefs in the superstructure via deduction. What makes a belief foundational is infallible justification: the kind of justification that rules out being mistaken. The foundation consists exclusively of beliefs about the content of one’s own mental states. Beliefs about the external world are invariably located in the superstructure.

It would be difficult to be more ambitious than that. Today, most epistemologists reject Cartesian foundationalism because the view makes knowledge of the external world impossible. There are two main reasons why the Cartesian approach guarantees this skeptical outcome. First, beliefs about the external world cannot be deduced from beliefs about the content of one’s own mental states. Therefore, if foundational beliefs are expected to justify beliefs in the superstructure through deduction, the superstructure beliefs remain unjustified. Second, if the beliefs in the foundation are exclusively about one’s own mental states, as they must be if foundational justification is infallible, then normal people in normal situations do not have very

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



MATTHIAS STEUP

many foundational beliefs (see Pollock 2001: 43, and 1986: 62). As a result, their nonfoundational beliefs about the external world are unjustified. No one pursues any longer the goal of finding a deductive link between foundational and non-foundational beliefs. However, some epistemologists still hold in high regard the Cartesian conception of foundational justification, most notably Laurence BonJour and Richard Fumerton, both of whom are known for their advocacy of what Fumerton calls classical foundationalism (BonJour 2001, Fumerton 1995 and 2001). Neither of these philosophers makes infallibility an outright necessary condition for membership in the club of foundationally justified beliefs. However, both of them hold that foundational justification needs to be identified with a kind of justification that at the very least approaches infallibility. Fumerton is of course known for his claim that what provides near infallible justification and thus opens the door to admission into the foundation is the relation of acquaintance. In this chapter, I will make an attempt to get clear on what Fumertonian acquaintance-based justification is all about, particularly given Fumerton’s commitment to internalism. I will argue that, while acquaintance-type justification is indeed a fine thing inasmuch as it is sufficient for foundational justification, it is not necessary for it.

1 Internalism vs. Externalism and the Expectation of Non-Accidentality What is the distinctive mark of epistemic justification, as opposed to its moral and prudential variants? According to the dominant approach on this matter, it is its connection to truth and knowledge. Laurence BonJour has articulated the connection with admirable clarity. He considers epistemic justification to be of instrumental value in the pursuit of truth. If justification were not of such instrumental value, then, BonJour says, “epistemic justification would be irrelevant to our main cognitive goal and of dubious worth. It is only if we have some reason for thinking that epistemic justification constitutes a path to truth that we as cognitive beings have any motive for preferring epistemically justified beliefs to epistemically unjustified ones” (BonJour 1985: 8). Since justification is supposed to be a path to truth, a justified belief must be such that, if true, its truth is non-accidental. Hence, given that nonaccidental truth is the distinctive mark of knowledge, one’s having a justified belief is, at least to some extent, being on one’s way towards having knowledge.1 The kind of non-accidentality that justification is expected to secure can be judged in two ways. It can be judged from (a) the subject’s perspective, and (b) the perspective of an independent observer who knows things the subject might not be in a position to know. 1

Justified true beliefs are typically instances of knowledge, but, as Gettier cases show, not always.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

FOUNDATIONAL JUSTIFICATION , META - JUSTIFICATION



According to internalism, whether a belief satisfies the demand for nonaccidental truth is to be judged from (a); according to externalism, it is to be judged from (b). Internalists and externalists agree that the point of justification is to secure non-accidental truth. They disagree about what determines nonaccidental truth. The internalist understanding of justification as a means to non-accidental truth is prominently at work in Laurence BonJour’s explanation of why Norman, the unwittingly reliable clairvoyant in his famous counterexample to reliabilism, fails to be justified in believing what his (unrecognized) clairvoyance has revealed to him, namely that the President is in New York. Here is what BonJour says. Since Norman’s belief is reliably produced, the truth of his belief would not be an accident from the standpoint of [a] hypothetical external observer who knows all the relevant facts and laws. But how is this supposed to justify Norman’s belief? From his subjective perspective it is an accident that the belief is true. And the suggestion here is that the rationality or justifiability of Norman’s belief should be judged from Norman’s own perspective rather than from one which is unavailable to him. (BonJour 1985: 43f)

Such is the internalist understanding of justification. Its key component is the expectation that a belief, if both justified and true, is not accidentally true when judged from within the subject’s perspective. In what follows, I will refer to this feature of internalist justification as the internalist demand for non-accidental truth.2

2 BonJour’s Anti-Foundationalist Argument The internalist demand for non-accidental truth is also at work in BonJour’s famous argument against foundationalism (BonJour 1985: 30–3). Since Fumerton is no less committed to internalism than BonJour, an examination of Fumertonian acquaintance-based justification will benefit from considering how such justification relates to the conception of internalism at work in BonJour’s argument, which is divided into two parts. The first gives an account of what it is to meet a demand for meta-justification. Suppose you challenge me to justify a certain belief of mine. In response, I cite what I take to be a reason for believing in support of that belief. Let us assume the reason I cite is a certain property of my belief, F. Now you might ‘go meta’ and ask me: “Why do you think that your belief ’s having property F justifies 2 Michael Bergmann takes the internalist demand for non-accidental truth, as advocated by BonJour, to be the standard against which internalist justification must be measured. In his dilemma argument against internalism, he argues that, while strong internalism (internalism with higher-level requirements) generates a vicious infinite regress, weak internalism (internalism without higher-level requirements) produces a notion of justification that fails to measure up to internalism’s own standard for non-accidental truth (see Bergmann 2006: ch. 1). More on this in note 11.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



MATTHIAS STEUP

your belief?” In response to this meta-challenge, I could reply: “Beliefs having property F are likely to be true.” The kind of meta-justification this reply is meant to provide is encapsulated in the following truth-indicative-feature (TIF) argument.3

Part I: The TIF argument (1) My belief that p has property F. (2) Beliefs with property F are likely to be true. (3) My belief that p is likely to be true. Giving a TIF argument is what’s needed to meet a justificatory meta-challenge. Its point is to provide second-order justification: justification for thinking that a certain property of a belief makes it likely that the belief is true. An example of such a property could be being grounded in a clear and distinct visual experience. If you challenged me to justify my belief that there is a tree before me, I might cite this property. And if you then asked me why I think that this property renders my belief justified, I could say that beliefs with that property are likely to be true. By saying that my belief is grounded in a clear and distinct visual experience, I provide you with a justification for my belief. By saying that beliefs based on a clear and distinct visual experience are likely to be true, I provide you with a meta-justification for my belief. The second part of BonJour’s argument rests on two further premises. The first lays down a general condition of justified belief: a belief is justified only if one has a meta-justification for it in the form of a TIF argument. What motivates this premise is precisely the internalist demand for non-accidental truth, introduced in the previous section. BonJour’s thought is that, if one is not in possession of meta-justification, having a reason R for a belief B just isn’t enough to satisfy the internalist demand for non-accidental truth. To meet this demand, what’s required is having an argument justifying the assumption that R makes the truth of B probable. The second premise of the second part of BonJour’s argument asserts that, if justification requires meta-justification, basic or non-inferentially justified beliefs are not possible.4 This premise rests on two sub-premises. Sub-Premise 1: A belief is basic iff (i) it is justified and (ii) it does not receive any of its justification from further beliefs. Sub-Premise 2: To have a TIF argument is to believe each of its premises, and to do so with justification. Consider now the second part of BonJour’s argument. 3 4

The term ‘TIF argument’ is due to Richard Feldman (see his 2003: 75ff). I will use the terms ‘basic,’ ‘non-inferential,’ and ‘foundational’ interchangeably.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

FOUNDATIONAL JUSTIFICATION , META - JUSTIFICATION



Part II: The need for meta-justification (4) Having an empirically justified belief that p requires having a TIF argument for believing that p.5 (5) If (4) is true, then no empirical belief is basic. (6) No empirical belief is basic. If all of the premises and sub-premises are true, then non-inferential justification is impossible. However, the argument has two weak spots: line (4) and sub-premise 2. In line (4), BonJour asserts a highly controversial principle, namely: The Principle of Meta-Justification: All empirical justification requires metajustification.6 Externalists reject this principle, and so do many internalists.7 The other weak spot of BonJour’s argument is sub-premise 2. Epistemologists who accept the principle of meta-justification could still be foundationalists, given the understanding of basicality sub-premise 1 articulates. To avoid BonJour’s anti-foundationalist conclusion, all they need to do is insist on an alternative understanding of what it is to have a TIF argument. They might, for example, replace sub-premise 2 with: Sub-Premise 2*: To have a TIF argument is to be in possession of evidence for each of its premises. It is possible to be in possession of evidence for each of the premises of a TIF argument, A, without having reflected on A and without believing any of A’s premises. If sub-premise 2* replaces sub-premise 2, then the claim made on line (5) is no longer true. It is then perfectly possible for a belief of mine, B, to be such that (i) B is justified without receiving its justification from any other beliefs, and (ii) I have a TIF argument for B because I have evidence for the premises of such an argument. So, if BonJour’s doxastic construal of what it is to have an argument is replaced with an evidential construal, internalists who wish to accept the principle of meta-justification can avoid BonJour’s anti-foundationalist conclusion by rejecting line (5). Next, I will introduce two internalist theories that illustrate the two responses to BonJour’s argument that we just considered.

5 The insertion of ‘empirical’ is necessary because BonJour’s argument is directed only against foundationalism about empirical justification. As far as a priori justification is concerned, BonJour advocated foundationalism even in his (1985). 6 Again, note the restriction to empirical beliefs. 7 Van Cleve conceives of the rejection of the principle meta-justification as an externalist move. See his (2003). This seems odd, given that there are internalists who reject it. Huemer’s conservatism and Pryor’s dogmatism, both of which are internalist theories, are incompatible with this principle.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



MATTHIAS STEUP

3 Two Foundationalist Responses to BonJour’s Argument Theories that are both foundationalist and internalist differ with regard to whether they accept or reject the principle of meta-justification. Michael Huemer’s phenomenal conservatism (PC) is a version of internalist foundationalism that rejects the principle of meta-justification. According to PC, seemings—mental states that represent their content as true—are necessarily a source of justification:8 Phenomenal Conservatism: Sp ! (J for Bp)9 According to PC, a subject’s having a seeming that p is—by itself, without metajustification playing any role at all—sufficient for giving the subject justification for believing that p. PC is a version of foundationalism because if a belief is based on an undefeated seeming, the belief is basic: justified without receiving any of its justification from another belief. Here are two examples. First, consider a perceptual seeming. If an object looks red to me, then I have, according to PC, justification for believing that the object is red. I may in fact have evidence in support of taking this

8 For Huemer’s characterization of seemings, see his (2013a). Regarding the label ‘phenomenal conservatism,’ it would actually be better to refer to the view as ‘conservative phenomenalism.’ This would allow one to distinguish between two views, one conservative, the other non-conservative, both of them sharing phenomenalism as their common core. ‘Phenomenalism,’ in this usage, denotes the view that at least some seemings justify, and that at least some justification we have for our beliefs is due to seemings. With phenomenalism thus defined, we may distinguish between four views: (i) conservative phenomenalism: all seemings are a source of justification; (ii) non-conservative phenomenalism: only properly credentialed seemings are a source of justification; (iii) universal phenomenalism: all justification is due to seemings; (iv) non-universal phenomenalism: some justification is due to seemings. Huemer, it would seem, advocates both conservative and universal phenomenalism (see 2013a: secs. 1c and 4d, and 2013b: 338–41). 9 Two important points should be noted. In this articulation of PC, the principle is about propositional, not doxastic, justification. That is, the principle asserts that having a seeming that p is sufficient for having justification for believing that p (which does not entail actual belief); it does not assert that having a seeming that p is sufficient for the belief that p to be justified (see Huemer 2013a: 2). Second, the reader might wonder why the consequent does not include a qualification to the effect that the justification provided by the seeming is merely prima facie or defeasible. The answer is that any such qualification is unnecessary at best and outright mistaken at worst. In his (2001: 101), Huemer asserts that a seeming that p is sufficient for having at least prima facie (defeasible) justification for believing that p. By inserting ‘at least,’ he means to emphasize that the justification the seeming that p provides might in fact be indefeasible. Completely spelled out, Huemer’s (2001) articulation of PC says the following: If one has a seeming that p, then one has, if not indefeasible, at least defeasible justification for believing that p. This is of course equivalent to: if one has a seeming that p, then one has either defeasible or indefeasible justification for believing that p. Since the distinction between defeasible and indefeasible justification is exhaustive, the “defeasible or indefeasible” qualification is redundant, which is to say that Huemer’s (2001) articulation of PC boils down to: if one has a seeming that p, then one has justification for believing that p. If, as is prevalent in the literature, the articulation is embellished with a prima facie qualification—as in: if one has a seeming that p, then one has defeasible justification for believing that p—then the result is actually subject to counterexample. Consider, as Huemer suggests, “I have a splitting headache” as a replacement for p: if it seems to me that I have a splitting headache, then I have defeasible justification for believing that I have a splitting headache. This conditional is false if the seeming in question is, as it could very plausibly be claimed, indefeasible (see Huemer 2001: 101f ).

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

FOUNDATIONAL JUSTIFICATION , META - JUSTIFICATION



seeming to be an instance of a reliable type of seeming, but such evidence is, according to PC, not needed. I may in fact have the belief that beliefs based on such seemings tend to be true. But, according to PC, no such meta-justificatory belief is necessary for having justification for believing that p. That the object looks red to me is, by itself, sufficient for generating justification for me to believe that the object is red. If I have no defeaters and my belief that the object is red is based solely on its looking red to me, then my belief is non-inferentially justified, that is, basic. Second, consider a memorial seeming. It seems to me that I had cereal for breakfast. This seeming, all by itself, provides me with justification for believing that I had cereal for breakfast. For the seeming to accomplish this, it is not necessary that I have evidence in support of memorial reliability, or that I am justified in believing that beliefs based on clear and distinct memorial seemings tend to be true. If my belief that I had cereal for breakfast is based solely on the memorial seeming, and if that seeming is undefeated, then my belief about what I had for breakfast is basic: justified without any help from further beliefs. Theories similar to Huemer’s PC are James Pryor’s dogmatism and Earl Conee and Richard Feldman’s evidentialism (see Pryor 2000, Conee and Feldman 2004, and Feldman 2003: ch. 3). They all assert that, for a perceptual or a memorial experience E to be a source of justification, it is not necessary that the subject be in possession of a justification for believing that E is a source of justification, that is, be in possession of a meta-justification.10 So, according to each of these theories, the mistake of BonJour’s anti-foundationalist argument lies in line (4), which asserts the principle of meta-justification. Compare these theories with another version of foundationalism: Internalist Reliabilism (IR).11 The basic idea of IR is that a belief source is a source of justification for a subject if, and only if, the subject has evidence of its reliability. So, according to IR, a seeming that p, by itself, is not sufficient for having justification for believing that p. Rather, what is sufficient for having justification for believing that p is the conjunction of a seeming that p and evidence for ascribing reliability to that seeming. Symbolically: Internalist Reliabilism: {Sp & E[R(Sp  p)]} ! (J for Bp)12

10 Pryor’s dogmatism, in fact, asserts something much stronger, namely that foundational justification for believing p (immediate justification, in his terminology) is a kind of justification that is not also justification for believing propositions in addition to p. 11 See Steup (2004 and 2013). Cohen (2002) advocates a similar view. In my (2013), I argue that internalist reliabilism is not vulnerable to Bergmann’s dilemma argument against internalism. Since the higher-level requirements internalist reliabilism imposes are evidential, not doxastic, the regress problems of strong internalism are avoided. And, precisely because of internalist reliabilism’s evidence-of-reliability requirement, the view blocks the objection that internalist justification is too weak to measure up to the internalist demand for non-accidental truth. For Bergmann’s response, see Bergmann (2013). 12 I have not presented IR as stating a seeming plus reliability is necessary for having justification for believing that p, since I want to leave it open for the IR advocate to deny that all justification comes from seemings. However, a complete statement of the view would have to include the claim that justification for

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



MATTHIAS STEUP

The antecedent’s second conjunct states a requirement for meta-justification: a seeming that p gives the subject justification for believing that p only if the subject has evidence for a reliable correlation between the seeming that p and the truth of p. However, the requirement as articulated by IR does not demand, as BonJour does, meta-justificatory beliefs. It merely requires the subject to have the kind of evidence that meta-justificatory beliefs, if they were formed, would need to be justified. IR qualifies as a version of foundationalism precisely because it does not require meta-justificatory beliefs. Suppose again an object x looks red to me. Suppose further the conditions of observation are normal. Since I have plenty of evidence for the reliability of my color vision under normal conditions of observation, the conditions the antecedent of IR states are satisfied. Finally, suppose I have no defeaters and base my belief about x’s color solely on my visual experience. In that case, my belief is basic: justified without receiving any of its justification from further beliefs. According to IR, the mistake in BonJour’s anti-foundationalist argument is not the principle of meta-justification asserted on line (4), but rather sub-premise 2. Advocates of IR would insist that sub-premise 2 be replaced with sub-premise 2*. As pointed out above, this move has the effect of blocking BonJour’s anti-foundationalist conclusion. IR, then, illustrates the compatibility of foundationalism with the principle of meta-justification. Let us call theories that reject the principle of meta-justification monistic theories. They are monistic because they assert that one single bit of evidence—for example, a perceptual experience or seeming with content p—is sufficient for providing a subject with evidence in support of believing that p. In contrast, theories on the other side of the divide—theories committed to the principle of meta-justification—may be called holistic, for such theories deny that a seeming that p qualifies by itself as a source of justification for believing that p.13 According to holistic theories, a seeming, S, can play that role only if it is embedded within a complex evidential support structure that provides justification for attributing reliability to the type of seeming of which S is an instance (see Cohen 2002). Holistic theories are versions of coherentism if they require that the metajustificatory support structure take the form of beliefs. As internalist reliabilism illustrates, holistic theories can also be versions of foundationalism, at least as long as foundational justification is equated with justification that is not conferred by any beliefs. However, as we will see next, monistic theories allow for a kind of foundational justification that, according to holistic theories, is not possible.

believing that p requires reliability evidence, and that, consequently, a seeming that p all by itself is not sufficient for justification for believing that p. I’m following Cohen, who, in his (2002), uses the label ‘holism’ to refer to theories according to which it’s not possible to acquire knowledge through a source S without knowing that S is reliable. 13

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

FOUNDATIONAL JUSTIFICATION , META - JUSTIFICATION



4 Two Conceptions of Foundational Justification One way for justification to be foundational is to exclude any beliefs from what confers it. According to an alternative conception, what makes justification foundational is the exclusion of any meta-justification. The two kinds of basicality can be defined as follows: Doxastic Basicality S has basic justification for believing that p $ S’s justification for believing that p neither includes nor comes from any further beliefs.14 Doxastic basicality must be distinguished from: Evidential Basicality S has immediate justification for believing that p $ S’s justification for believing that p comes not even partially from S’s having a meta-justification for believing that p.15 Suppose again I believe that object x is red because it looks to me that way. According to a monistic theory like PC, to have justification for believing that x is red, the only evidence I need is: J1:

x’s looking red to me.

According to a holistic theory like IR, J1 by itself does not give me justification for believing that x is red. In addition to J1, I also need: J2:

evidence for believing that J1 is a reliable indicator of truth.

Suppose I have in fact both J1 and J2 in the way of support for believing that x is red, without, however, holding any meta-justificatory beliefs on the basis of J2. According to IR, my justification for believing that x is red consists of the conjunction of J1 and J2. Nevertheless, despite being holistic, IR allows for my justification to be basic in the doxastic sense. What makes my justification, the conjunction of J1 and J2, basic in the doxastic sense is the exclusion of any beliefs from it. Since the conjunction of J1 and J2 comprises all of the justification I have for believing that x is red, and since that conjunction does not include nor come from any beliefs, IR classifies my justification for believing that x is red as basic in the doxastic sense. But IR does

14 I am here articulating doxastic basicality as a feature of propositional justification. Compare Feldman’s corresponding account of doxastic basicality as a feature of doxastic justification: “B is a justified basic belief = df B is justified, but is not justified on the basis of any other beliefs” (Feldman 2003: 50). 15 Neither doxastic nor evidential basicality capture what Pryor means by immediate justification. According to Pryor, what makes one’s justification for believing that p immediate is the exclusion of any justification for rejecting defeaters and incompatible skeptical hypotheses. See the conception of immediate justification he proposes in his (2000) and (2013a).

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



MATTHIAS STEUP

not classify it as basic in the evidential sense. For the conjunction of J1 and J2 is justification not only for believing that x is red, but also justification for believing the further proposition that x’s looking red is a reliable indicator of truth. So the conjunction of J1 and J2 provides both justification and meta-justification. This is precisely what blocks basicality in the evidential sense.16 Advocates of PC might agree that J2 gives me additional justification for my belief. They might agree that, if I have both J1 and J2, I have a stronger justification for believing that x is red than I would have if I had only J1. Nevertheless, unlike internalist reliabilists, advocates of PC can insist that I have basic justification, in the evidential sense, for believing that x is red, namely J1 by itself. To sum up: monistic theories, rejecting the principle of meta-justification, allow for justification to be basic in both the doxastic and the evidential sense. Holistic theories, precisely because they accept the principle of meta-justification, allow for justification to be basic only in the doxastic sense. Basic justification in the evidential sense is, according to holistic theories, impossible. Let us recapitulate. We have seen that there are two kinds of internalism: monistic and holistic. Monistic internalists reject the principle of meta-justification; holistic internalists accept it. What motivates the principle of meta-justification is the internalist demand for non-accidental truth. This demand has two components. The first, non-accidentality, flows from the role justification is supposed to play: that of making true belief eligible to be an instance of knowledge; to move, as it were, true belief a good distance towards the goal of knowledge. As far as the importance of non-accidentality is concerned, internalists and externalists are in agreement. They disagree about where the factors responsible for non-accidental truth are located. According to internalists, of course there are external factors of non-accidentality, fixing the objective probability of a belief ’s being true. However, in addition to those, internalists recognize factors of non-accidentality that are internal to the subject— located within the subject’s perspective. The second component of the internalist demand for non-accidental truth, then, is the specifically internalist way of conceiving of non-accidentality. Now, when considered from within a subject’s perspective, what is necessary for making a belief ’s truth non-accidental? Holistic internalists hold that what’s necessary is evidence that the belief has its origin in a reliable or truth-conducive source. For externalists, what’s required for non-accidental truth is de facto reliability. In their attitude towards monistic internalism, externalists and holistic internalists are partial allies. Both of them think that, for rendering a belief ’s truth non-accidental, reliability is key—evidence of reliability for holistic internalists, de facto reliability for externalists. Monistic internalists deny that either one of these is necessary for making a belief ’s truth non-accidental. They think that an experience

16

IR is a non-conservative theory in Huemer’s terminology, and a non-dogmatic theory in Pryor’s terminology.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

FOUNDATIONAL JUSTIFICATION , META - JUSTIFICATION



as of p, or a seeming that p, is by itself sufficient for making a true belief that p, from the subject’s own perspective, non-accidentally true.17

5 Fumerton’s Principle of Inferential Justification Let us now turn to Fumerton. Obviously, he is an internalist.18 But is he a monistic or a holistic internalist? He appears to be both. He accepts and also rejects the principle of meta-justification. Consider Fumerton’s Principle of Inferential Internalism (Fumerton 1995: 67f, and 2001: 5): To be justified in believing one proposition P on the basis of another E one must be (1) justified in believing E, and (2) justified in believing that E makes probable P. Condition (2) expresses a demand for meta-justification. Its name suggests the following two points: first, Fumerton thinks that meta-justification is necessary for inferential justification; second, he thinks that the need for meta-justification is a consequence of an internalist understanding of justification. However, Fumerton asserts no corresponding principle for foundational justification. He seems to think that, whereas inferential justification requires meta-justification, foundational justification does not. Two questions arise. First, why does Fumerton exempt foundational justification from the demand for meta-justification? Second, is such an exemption defensible? To turn to the first of these questions, I assume Fumerton’s motivation for the exemption is the thought that foundational justification and meta-justification are incompatible. Above, I have argued that there is an understanding of foundational justification according to which this thought is false. Although holistic internalism is incompatible with evidential basicality, it is compatible with doxastic basicality as long as the required meta-justification need not take the form of beliefs. As for the second question, it would appear that Fumerton accepts the internalist demand for nonaccidental truth. Clause (2) of Fumerton’s principle nicely takes care of this demand. However, the demand is completely general. It applies to all justification. Surely, if we think that a justified belief, if true, must be such that its truth is non-accidental from within the subject’s perspective, this thought applies to both inferential and noninferential justification. Therefore, to the extent Fumerton’s foundationalism is a version of internalism, it manifests a certain amount of tension and instability resulting from his split attitude towards the need for meta-justification: accepting it for inferential beliefs and rejecting it for non-inferential beliefs. 17

I take monism to be a view about what’s sufficient for justification. Monists may, but need not, go further and also hold that a belief that p cannot be internalistically justified without the subject having a seeming that p. 18 See Fumerton’s Metaepistemology and Skepticism, in which he introduced an early version of an important and influential line of argument against externalism, namely that reliabilism makes the acquisition of reliability knowledge too easy (see Fumerton 1995: 175ff, cf. Cohen 2002, and Vogel 2000).

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



MATTHIAS STEUP

Monistic internalism rejects the need for meta-justification for both noninferential and inferential justification. It thus avoids the bifurcated and inconsistent approach that causes trouble for Fumerton’s internalism. However, monistic internalism is vulnerable to the charge that it does not take the demand for nonaccidentality seriously enough. To satisfy this demand, reliability must somehow be brought into the picture. Externalists do this by demanding de facto reliability. Internalists can do it by requiring evidence of reliability. As an internalist, Fumerton requires reliability evidence for inferential justification but demurs when it comes to foundational justification. As indicated above, it seems to me his demurral rests on a mistake, namely the thought that foundational justification and meta-justification are incompatible. Next, I will argue that Fumerton’s demurral is actually a mere surface feature of his theory. Surprisingly, it turns out that, hidden within the complexities of Fumerton’s acquaintance account of foundational justification, we find a rather hefty demand for meta-justification. Hence, whereas initially it looks like Fumerton’s approach to meta-justification is bifurcated, it is, at least by implication, entirely consistent: Fumerton requires meta-justification explicitly for inferential justification and implicitly for non-inferential justification as well.

6 Fumertonian Non-Inferential Justification Let us turn to the core-component of Fumerton’s classical foundationalism: his theory of justification by acquaintance. Fumerton’s account of non-inferential justification goes as follows: S has non-inferential J for Bp iff S is acquainted with (i) the fact that p (truth-maker), (ii) the thought that p (truth-bearer), (iii) the relation of correspondence holding between the thought that p and the fact that p. Some people might be bothered by the old-fashioned metaphysics providing the framework for this approach. I am not. But even those who are sympathetic to Fumerton’s deployment of propositions, facts, and the correspondence theory of truth might still wonder what justification via acquaintance comes to. To illuminate what he means by the concept, Fumerton offers several metaphors: acquaintance is a direct confrontation with truth, a direct awareness of the item with which one is acquainted, and it is such that nothing stands between the subject’s mind and the item with which one is acquainted (see Fumerton 2001: 14). These metaphors are helpful, but, as Fumerton himself points out, they go only so far in helping us understand what justification by acquaintance comes to (see 2001). A crucial clue comes to light when we consider the question: why is foundational justification as conceived by Fumerton a version of classical foundationalism?

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

FOUNDATIONAL JUSTIFICATION , META - JUSTIFICATION



In other words, why are only beliefs about one’s own mental states eligible for acquaintance foundation? Here is why: Fumerton insists that one cannot be acquainted with external physical objects.19 He says: If one concedes, for example, that the justification one has now for believing that there is a table before one is perfectly compatible with the table’s not being there, then one has just conceded that one is not directly acquainted with the table. (2001: 15f)

And: It seems to me that traditional arguments from the possibility of hallucination do establish that one cannot be directly acquainted with any facts about the external world. (2001: 16)

Clearly, Fumerton thinks that a belief is justified by virtue of acquaintance if, and only if, the justification provided does not, in some way or another, allow for the possibility of mistake. However, Fumerton does not hold that acquaintance justification is exactly the same as infallible justification; that is, he would reject the following: AQ1: One’s justification J for Bp is acquaintance justification iff [~♢ (one has J for Bp & ~p)]. Fumerton would reject AQ1 for two reasons. First, he thinks that acquaintance justification for Bp can also justify one in believing p*, where p* is false but similar to p. So acquaintance justification is not, strictly speaking, infallible.20 Second, as a card-carrying internalist, he would not want to say that the fact of infallibility does anything in the way of justification (see 2001: 10). When one believes a proposition that is necessarily true, one cannot be mistaken. Clearly, though, it is possible to believe a necessary truth without having justification for believing it. From the internalist point of view, therefore, infallibility can justify only if it is somehow transformed into seeming infallibility: the appearance of infallibility from within the subject’s perspective. That, given his internalist credentials, is the only way in which Fumerton can link acquaintance with foundational justification. One way to internalize infallibility is to demand justified beliefs about one’s infallibility. This is going to get an infinite regress started—a regress of meta-beliefs— and thus is an obvious recipe for disaster. A better way to internalize infallibility is to demand not beliefs about, but instead evidence for, infallibility. So, given Fumerton’s

19 The primary contrast here is that between propositions about the physical world and propositions about how the physical world appears to one. According to classical foundationalism, when it comes to that contrast, one can have foundational justification only for the latter. Propositions about abstract objects are a different matter. Realists consider abstract objects to be mind-independent, external objects. It does not seem to me it’s an essential part of classical foundationalism that one cannot have foundational justification for propositions about external objects thus conceived. 20 Fumerton (2001: 15). We must, therefore, distinguish between the relation of acquaintance and acquaintance justification. While the former is factive, the latter is not.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



MATTHIAS STEUP

commitment to internalism, what is needed to correct AQ1 is an ‘E’ (for having evidence for) at the right spot, as in: AQ2:

One’s justification J for Bp is acquaintance justification iff E[~♢ (one has J for Bp & ~p)]

AQ2, in my mind, is the way to make sense of the two passages I quoted above. Now, let us assume Fumerton is right when he proposes that acquaintance justification comes about when there are, clearly and distinctly presented before one, with nothing intervening: (i) one’s thought that p, (ii) the fact that p, and (iii) the relation of correspondence holding between them. The question I am interested in is: if, by virtue of standing in these three acquaintance relations, one enjoys acquaintance justification for believing that p, is one’s justification monistic or holistic? Does it include or exclude meta-justification? Is one’s justification foundational in the evidential sense or merely in the doxastic sense? It seems to me the answer is as follows: to have acquaintance justification is to have the best kind of metajustification one could possibly have, namely evidence for thinking that what one believes originates in an infallible source.21 So, acquaintance justification is holistic, it unavoidably comes with meta-justification, and it can be foundational only in the doxastic sense. The point I am making here should not be misunderstood. If we think of metajustification as always taking the form of meta-justificatory beliefs, then I would be mistaken in claiming that Fumertonian acquaintance justification necessarily involves meta-justification. However, it’s possible to have a meta-justification without forming any meta-justificatory beliefs. Suppose you have a reason R for believing that p. If you also have a reason R* for believing that R is a reason for believing that p, then you have a meta-justification for believing that p whether or not you actually form, based on R*, the belief that R is a reason for believing that p. Now, suppose R is such that you can’t have R without also having R*. In such a case, you can’t have the first-order justification R gives you without also having a meta-justification for believing that p. In the case of Fumertonian acquaintance justification, something or other—an introspective experience, the mental state with which you are acquainted, or the total body of all three acquaintance relations that make up Fumerton’s account—is the source from which acquaintance justification flows. Let R refer to that source. According to Fumertonian internalism, R gives you an assurance that you are not being misled by R: that it’s not possible for you to 21 While acquaintance justification involves the best meta-justification one could have, it still allows for differences in the degree of justification. It is possible that two subjects, S1 and S2, both have acquaintance justification for believing that p, but S1 is more strongly justified in believing that p than S2 is. Suppose, for example, S1 believes she has a headache, and S2 believes she has a headache. However, S2’s headache is much stronger than S1’s. Arguably, S2 has a higher degree of acquaintance justification for her belief than S1 even though both have meta-justification for believing that what justifies their beliefs—the experience of pain—rules out the possibility of being mistaken.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

FOUNDATIONAL JUSTIFICATION , META - JUSTIFICATION



have reason R for believing that p while p is false. So, if R gives you acquaintancetype justification, then you cannot have R without also having a reason R* for thinking that R is an infallible source of justification. Acquaintance justification, therefore, is a type of justification that involves both first-order justification and meta-justification. It turns out, then, that Fumerton’s classical foundationalism is in fact a version of holistic internalism. Even though his principle of inferential internalism requires meta-justification only for inferential justification, Fumerton’s account of noninferential justification as acquaintance justification, involving meta-justification, does not in this respect differ in principle from his account of inferential justification.

7 Three Versions of Foundationalism The idea of a seeming, Sp, providing foundational justification for an actual or possible belief, Bp, involves two components. The first is that Sp provides justification for Bp. The second component is a structural constraint on the first. For the justification that Sp provides to be foundational, it must exhibit a certain kind of structural independence. A specification of what that justification must be independent of results in a particular conception of foundational justification. Candidates for the target of the desired independence are justification for further beliefs, justification for taking Sp to be reliable, justification for rejecting defeaters for Sp, and justification for rejecting skeptical alternatives that are incompatible with p. These two components—the justificational support relation and the structural independence requirement—are in tension with each other. Strengthening the independence relation weakens the justificational support relation; strengthening the justificational support relation weakens the degree to which the justification Sp provides is independent from additional layers of justification. Consider phenomenal conservatism: PC:

Sp ! (J for Bp)

PC, declaring a seeming that p to be sufficient for having justification for believing that p, puts on offer a kind of justification that involves maximum-strength foundational independence. The cost is a rather weak—indeed, arguably unacceptably weak—justification relation. To say that Sp is sufficient for providing justification for believing that p is to say that the following is possible: Sp—all by itself, without any additional layers of justification involved—provides justification for believing that p. When this is the case—all justification for Bp coming solely from Sp—the justification for Bp does not involve any of the following: (i) beliefs in any further propositions; (ii) justification for taking Sp to be a reliable; (iii) justification for rejecting defeaters for Sp as a source of justification for believing that p;

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



MATTHIAS STEUP

(iv) justification for rejecting skeptical hypotheses that are incompatible with the truth of p.22 At a minimum, foundational justification requires the exclusion of a justificatory contribution from any beliefs. After all, only justified beliefs can confer justification. Hence, as soon as a first belief ’s justification is even in part conferred on it by a second belief, the first belief ’s justification is no longer non-inferential, that is, foundational. It is not so clear, however, that the exclusion of (ii)–(iv) is necessary if Sp is to provide foundational justification for believing that p. Here are four reasons for thinking that, if we want to avoid a destructive weakening of the justificational component from our account of foundational justification, Sp provides justification only if it is embedded in a total body of evidence that includes (ii)–(iv): (1) Justification is supposed to be a path towards non-accidental truth. By itself, a seeming that p is not a reason to think that p is probably true, and therefore is not a source of justification. That is, a seeming that p makes the truth of p nonaccidental from within the subject’s perspective only if it is accompanied by evidence of reliability. Such a body of evidence would, at least to some degree, involve justification for rejecting defeaters and justification for rejecting skeptical hypotheses that are incompatible with p. (2) We do not trust instruments, gauges, and witnesses unless we have reason to consider them reliable.23 If you didn’t have any reason at all to consider your gas gauge reliable, the fact that it reads ‘Full’ would give you no justification at all for believing that there is gas in the tank. Likewise, if a detective investigating a crime interviews a witness and the witness says p, this does not give the detective any reason whatever to believe that p is true unless the detective has evidence that certifies the witness’s reliability. I see no reason to think that we should hold our faculties to a different standard. It seems arbitrary to say that, whereas instruments, gauges, and witnesses are a source of justification only if we have evidence of their reliability, the seemings that our faculties produce are a source of justification even in the absence of reliability evidence. (3) If seemings are a source of justification by virtue of being embedded in a body of evidence that supports attributing reliability to them, then we can explain the justificatory power of seemings. The explanation is: when seemings justify, they do so because we have reason to consider them a reliable source of information. If PC were true, no such explanation would be available. The justificatory power of seemings would be a brute, unexplainable fact (see Steup 2004). 22

Cf. the conception of immediate justification Pryor proposes in his (2000) and (2013a). It is of course true that little children trust what their parents tell them, long before they have evidence of their parent’s trustworthiness. And it is true that, on many occasions, people trust instruments and witnesses without possessing evidence of their reliability. My response is that, in such cases, the beliefs in question are not justified. 23

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

FOUNDATIONAL JUSTIFICATION , META - JUSTIFICATION



(4) If PC were true, there would arguably be abominable conjunctions. Suppose, looking at a table that appears red to me, I say the following: NR: I have no reason whatever to believe that the table’s looking red to me is a reliable indicator of truth, but I have at least some justification to believe that this table is red.24 To my ears at least, there is considerable tension between the two conjuncts. This tension, it seems to me, is due to the fact that justification is supposed to be a path towards non-accidental truth. Therefore, if I have no reason at all to think that my visual experience of the table’s color is a reliable path to truth, then it follows that this experience fails to emit any justificatory juice. Due to the lack of reliability evidence, it is not a source of justification. More reasons for rejecting PC can be given, but I will stop here to consider internalist reliabilism:25 IR:

{Sp & E[R(Sp  p)]} ! (J for Bp)

According to IR, a seeming that p together with evidence for taking that seeming to be a reliable belief source is sufficient for having justification for believing that p. IR qualifies as a version of foundationalism because it says that a seeming that p, if accompanied by evidence of its reliability, is sufficient for justification. By virtue of making this sufficiency claim, IR says that, for a seeming that p to provide justification, no beliefs are needed. However, that is the only way in which IR allows for foundational justification. Consider defeaters. Arguably, many of them undermine the reliability of our belief sources. Hence, according to IR, if Sp provides justification for believing p for me, I must have justification for rejecting all defeaters that undermine the attribution of reliability to Sp. The same applies to skeptical hypotheses. Many of them pose a threat to the reliability of my belief sources. Hence, if IR is true, Sp is a source of justification for me only if I have justification for rejecting all skeptical hypotheses that impugn Sp’s reliability. PC and IR handle the tension between the justificational and the structural component of foundational justification in opposite ways. PC delivers a maximum strength conception of the foundational components. As a result, PC weakens the 24 In normal situations, we do in fact have evidence that our color experiences are reliable. Thus, under normal circumstances, it would be odd for someone to assert NR. But evidence for the reliability of one’s current color experiences is easily eliminated. Imagine I am seeing a play in a theater, and there is a redlooking table on the stage. The lights that are illuminating the stage are hidden from view. The location of the present scene is a nightclub, so I have no inductive evidence for ruling out red light illumination. In this situation, I have evidence neither for nor against the reliability of my color vision. 25 Some have proposed counterexamples to PC (see Peter Markie’s gold digger and Susanna Siegel’s angry Jack examples; Markie 2013 and Siegel 2012). For Huemer’s responses to these examples, see his (2013b). See also Audi (2013) and Brogaard (2013), both of whom reject conservatism on the basis of counterexamples. Others have objected that conservatism/dogmatism create problems of easy knowledge. See Cohen (2002) and Comesana (2013). Finally, Roger White (2006) has argued that dogmatism conflicts with Bayesian epistemology. Pryor responds to these objections in his (2013b).

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



MATTHIAS STEUP

justificational component. According to the critics of PC, the maximum strength basicality that PC ascribes to seemings in fact destroys the justificational component. IR, on the other hand, strengthens the justificational component at the expense of the structural component. Allowing for foundational justification via the exclusion of beliefs, IR delivers merely a weak conception of basicality. On the other hand, since IR requires reliability evidence, it delivers a strong justificatory support relation between Sp and Bp, and therefore has no trouble when it comes to meeting the expectation that justification must be a path to non-accidental truth when judged from within the subject’s perspective. For if we aim at non-accidental truth, we can do hardly better than to base our beliefs on sources whose reliability is certified in the light of our total evidence. Finally, let us consider the acquaintance theory: AQ:

{Sp & E[~♢(Sp & ~p)]} ! (J for Bp)26

According to AQ, a seeming that p and having evidence for taking that seeming to an infallible belief source is sufficient for having justification for believing that p. Regarding the tension between the justificational support relation and the structural basicality requirement, AQ puts, even more so than IR does, a premium on the justificational component of foundational justification. Only belief sources that we have reason to consider infallible can provide foundational justification. As a result, AQ is structurally analogous to IR. Obviously, defeaters and skeptical hypotheses undermine whatever evidence one might have for thinking that one has infallible justification for believing that p. So, if one has evidence indicating that Sp is an infallible belief source, one must have justification for rejecting defeaters or skeptical hypotheses that are incompatible with viewing Sp as infallible.27 Nevertheless, by allowing Sp to provide justification without the involvement of any beliefs, AQ secures at least one kind of structural independence for Sp and thus qualifies, as does IR, as a version of foundationalism.

26 Fumerton does not explicate acquaintance justification as a constraint on what kind of seemings are a source of justification. In fact, Fumerton is a skeptic about the existence of seemings, taking them to be either beliefs or mere inclinations to believe (as he mentioned in conversation). However, acquaintance justification can easily and sensibly be construed in this way, and doing so facilitates easy comparison. Moreover, it is not clear to me it is possible to make sense of acquaintance-type justification without thinking of acquaintance as a special epistemic status resulting from exercising such faculties as introspection or a priori intuition under ideal conditions. It strikes me as plausible, therefore, to construe acquaintance-type justification as a kind of justification we enjoy when our introspective experiences or seemings satisfy the conditions that are necessary and sufficient for acquaintance. Finally, my main point— acquaintance justification involves a kind of meta-justification—holds whether or not the acquaintance theory is construed as a special version of a seemings theory. 27 Whether AQ requires that one have justification for rejecting all defeaters and skeptical hypotheses depends on whether there are defeaters and skeptical hypotheses that do not undermine one’s evidence for taking Sp to be infallible. It seems unlikely to me that there are such defeaters and skeptical hypotheses. I suspect, therefore, that evidence of infallibility requires having evidence for rejecting all of them.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

FOUNDATIONAL JUSTIFICATION , META - JUSTIFICATION



8 Does Foundational Justification Require Acquaintance? Since acquaintance-type justification does not require a justificatory contribution from any beliefs, there is at least one way in which it qualifies as foundational justification. Alas, it doesn’t seem to me Fumerton wants to say that acquaintance justification is just one way among others in which we can have foundational justification for beliefs. Rather, I think his view is that acquaintance justification is the only way in which we can have foundational justification. His reason is, I assume, that justification consistent with the possibility of hallucination or evil demon deception cannot be foundational. But this seems wrong to me. Let us consider two motivations for advocating foundationalism. The first motivation is supplied by the need to terminate a chain of inferential justification with a belief that is non-inferentially justified. To meet the desideratum of regress termination, what’s needed is a kind of justification that excludes a justificatory contribution from any beliefs. AQ states a condition that is sufficient for justification of this kind and therefore delivers what foundationalists are looking for: a way to terminate the regress. But so do PC and IR. They do so because what’s essential to regress termination is not the high-voltage strength of acquaintance-type justification but rather the exclusion of a justificatory contribution coming from beliefs. As long as experiences or seemings can provide at least as much justification as is needed for knowledge, it is difficult to see why only acquaintance-type foundationalism succeeds with the task of terminating the regress. The second motivation for advocating foundationalism, or perhaps better a particular kind of foundationalism, has been proposed by James Pryor. He distinguishes between two arguments for skepticism about justification for our beliefs about the external world. The first appeals to the possibility of deception and goes like this: (1) If I have justification for my external world beliefs, then I have justification for rejecting the hypothesis that I am deceived by an evil demon. (2) I have no justification for rejecting the hypothesis that I am deceived by an evil demon. Therefore: (3) I have no justification for my external world beliefs. Pryor thinks that this argument does not represent the strongest skeptical challenge to the assumption that our external world beliefs are justified (see Pryor 2000: 522f ). According to him, the stronger challenge goes as follows: (1) Justification for external world beliefs is epistemically prior to justification for rejecting the evil demon hypothesis.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



MATTHIAS STEUP

(2) Justification for rejecting the evil demon hypothesis is epistemically prior to justification for external world beliefs. (3) If (1) and (2) are true, I have no justification for my external world beliefs. Therefore: (4) I have no justification for my external world beliefs.28 Pryor targets the second premise for rejection (see 2000: 532). According to him, we have a kind of justification for our external world beliefs—namely, justification coming from or perceptual experiences—that is independent of any epistemically prior justification we might have for rejecting the evil demon hypothesis or similar skeptical scenarios. So, for Pryor, the exclusion of beliefs is not enough for giving us foundational justification. Rather, we also need the exclusion of any evidence against skeptical hypotheses. Pryor thinks that perceptual experiences do the trick. For them to give us justification for our external world beliefs, we do not need prior justification for thinking that skeptical scenarios do not obtain. Now, is acquaintance-type foundational justification of any help to securing the kind of justification Pryor is looking for? What Pryor wants is justification that is foundational not only by virtue of excluding a justificatory contribution from any beliefs, but also by virtue of excluding any justification for rejecting skeptical scenarios—and indeed by virtue of excluding any additional justification whatsoever. As I have argued above, the answer to that question, it seems to me, is “no.” As Fumerton makes abundantly clear, acquaintance-type justification for p is not compatible with not being in an epistemic position to reject skeptical hypotheses that are incompatible with p (see 2001: 15f ). To the contrary, to have acquaintance-type justification for believing p is to have justification for believing that one’s being in such a skeptical scenario is not a possibility. Fumertonian acquaintance-type justification is not, therefore, of any help when it comes to Pryor’s project of rejecting the second premise of the skeptical epistemic-priority argument displayed above. I conclude: first, acquaintance-type justification is not the ticket if we are looking for the high-octane conception of foundational justification that conservatives and dogmatists advocate. Second, acquaintance-type justification is sufficient for the less ambitious kind of foundational justification the mark of which is the exclusion of a justificatory contribution from any beliefs. Third, acquaintance-type justification is not necessary for foundational justification of the kind that excludes a justificatory contribution from beliefs.29

28

See Pryor (2000: sec. 2). See also Jessica Brown’s helpful discussion in her (2013). An earlier and abridged version of this chapter was presented at the 6th Annual Orange Beach Epistemology Workshop, sponsored by the University of South Alabama, May 19–20. I wish to thank the workshop participants for helpful discussion, particularly Michael Huemer, Richard Fumerton, Susanna Schellenberg, and Chris Tucker, and Mike Bergmann for his comments on the penultimate draft. 29

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

FOUNDATIONAL JUSTIFICATION , META - JUSTIFICATION



References Audi, Robert. (2013). “Epistemic Innocence: Phenomenal Conservatism and Grounds of Justification,” in Chris Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, 154–78. Bergmann, Michael. (2006). Justification without Awareness. New York: Oxford University Press. Bergmann, Michael. (2013). “Phenomenal Conservatism and the Dilemma for Internalism,” in Chris Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, 154–78. BonJour, Laurence. (1985). The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press. BonJour, Laurence. (2001). “Toward a Defense of Empirical Foundationalism,” in Michael DePaul (ed.), Resurrecting Old-Fashioned Foundationalism. Lanham, MD: Rowman and Littlefield, 21–38. Brogaard, Berit. (2013). “Phenomenal Seemings and Sensible Dogmatism,” in Chris Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, 270–89. Brown, Jessica. (2013). “Immediate Justification, Perception, and Intuition,” in Chris Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, 71–88. Cohen, Stewart. (2002). “Basic Knowledge and the Problem of Easy Knowledge,” Philosophy and Phenomenological Research, 65: 309–29. Comesana, Juan. (2013). “There Is No Immediate Justification,” in Matthias Steup, John Turri, and Ernest Sosa (eds), Contemporary Debates in Epistemology (Second edition). Oxford: Wiley Blackwell, 222–35. Conee, Earl and Richard Feldman. (2004). Evidentialism: Essays in Epistemology. New York: Oxford University Press. Feldman, Richard. (2003). Epistemology. Upper Saddle River, NJ: Pearson. Fumerton, Richard. (1995). Metaepistemology and Skepticism. Lanham, MD: Rowman and Littlefield. Fumerton, Richard. (2001). “Classical Foundationalism,” in Michael DePaul (ed.), Resurrecting Old-Fashioned Foundationalism. Lanham, MD: Rowman and Littlefield, 3–20. Huemer, Michael. (2001). Skepticism and the Veil of Perception. Lanham, MD: Rowman and Littlefield. Huemer, Michael. (2013a). “Phenomenal Conservatism,” Internet Encyclopedia of Philosophy: 1–10. Huemer, Michael. (2013b). “Phenomenal Conservatism Über Alles,” in Chris Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, 328–50. Markie, Peter. (2013). “Searching for True Dogmatism,” in Chris Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, 248–69. Pollock, John L. (1986). Contemporary Theories of Knowledge. Totowa, NJ: Rowman and Littlefield.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



MATTHIAS STEUP

Pollock, John L. (2001). “Nondoxastic Foundationalism,” in Michael DePaul (ed.), Resurrecting Old-Fashioned Foundationalism. Lanham, MD: Rowman and Littlefield, 41–57. Pryor, James. (2000). “The Skeptic and the Dogmatist,” Nous, 34: 517–49. Pryor, James. (2013a). “There Is Immediate Justification,” in Matthias Steup, John Turri, and Ernest Sosa (eds), Contemporary Debates in Epistemology (Second edition). Oxford: Wiley Blackwell, 202–22. Pryor, James. (2013b). “Problems for Credulism,” in Chris Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, 89–131. Siegel, Susanna. (2012). “Cognitive Penetrability and Perceptual Justification,” Nous, 46: 201–22. Steup, Matthias. (2004). “Internalist Reliabilism,” Philosophical Issues, 14: 403–25. Steup, Matthias. (2013). “Does Phenomenal Conservatism Solve Internalism’s Dilemma?” in Chris Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, 135–53. Van Cleve, James. (2003). “Is Knowledge Easy—or Impossible? Externalism as the Only Alternative to Skepticism,” in Steven Luper (ed.), The Skeptics: Contemporary Essays. Farnham: Ashgate Publishing, 45–59. Vogel, Jonathan. (2000). “Reliabilism Leveled,” Journal of Philosophy, 97: 602–23. White, Roger. (2006). “Problems for Dogmatism,” Philosophical Studies, 131: 525–57.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

Perceptual Belief

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

4 Staying Indoors How Phenomenal Dogmatism Solves the Skeptical Problem without Going Externalist Berit Brogaard

Many naive realists and theorists who hold related positions (e.g., disjunctivism) are primarily motivated by epistemological considerations. They argue that unless we accept naive realism or related forms of direct realism, we cannot overcome the problem of skepticism. In this chapter I look closer at this concern and attempt to figure out why naive realism is sometimes thought to be the only reasonable account of perception if one wants to avoid the skeptical problem. I then argue that the naive realist’s response to the skeptical challenge is ultimately unsatisfactory and that phenomenal dogmatism, which is reconcilable with a representational view of perception, provides an adequate response to the skeptical problem.

1 The Threat of Skepticism The threat of skepticism is one of the main motivations for naive realism and related forms of direct realism (Fumerton 1995, Alston 2002, McDowell 1982, Fish 2009a and 2009b). Skepticism comes in many flavors: strong and weak, global and local (Fumerton 1995). Weak skepticism is the view that we cannot have knowledge of a certain subject matter or of all subject matters, whereas strong skepticism is the view that we cannot have justified beliefs about a certain subject matter or about all subject matters. Global skepticism is skepticism about all knowledge or justified belief, whereas local skepticism concerns knowledge or justified belief about a specific subject matter, for example, skepticism about the external world, induction, the future, or theoretical physics. Although weak skepticism is considered worrisome by some, one could argue that it is not too troublesome if we cannot have knowledge as long as we can have justified beliefs. Presumably, justified beliefs can play most, if not all, the functional roles that knowledge is thought to play. So, the really worrisome position is strong skepticism.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



BERIT BROGAARD

But not all forms of strong skepticism are coherent. Strong, global skepticism is widely considered incoherent, as it implies denying that we can have any justified beliefs. But that makes it irrational to defend the very position in the first place. So, the main serious skeptical challenges are forms of strong local skepticism. Strong local skepticism about the external world is perhaps the most commonly discussed form of skepticism and also the most worrisome. The standard skeptical argument proceeds by showing that the justification for some claim p is equally good justification for some alternative skeptical hypothesis q. For example, your perceptual evidence for thinking that the universe started billions of years ago is equally good evidence for the hypothesis that the universe started five minutes ago. If a deity had created the universe five minutes ago with its appearances of age and human beings rife with all their memories, things would phenomenally seem exactly as they actually do. Likewise, your perceptual evidence for thinking that there is an external world is equally good evidence for the hypothesis that we are all brains in vats. It is, of course, implausible to think that we somehow have direct conscious access to mind-independent physical objects in the non-veridical skeptical scenario. As we have the same justification in the actual world and the skeptical scenario, it is also implausible to think we have direct conscious access to mindindependent physical objects in the actual scenario, and hence it is unreasonable to think that we have non-inferential justification in the actual scenario. It follows that we don’t have any non-inferentially justified beliefs about the external world. So, if we have justification for our beliefs about the external world, that justification is inferential. This is the first part of the standard skeptical argument. The next part consists in arguing against the possibility of inferentially justified beliefs about the external world. In Metaepistemology and Skepticism (1995) Richard Fumerton argues that the skeptic’s argument against the possibility of inferentially justified beliefs rests on the following principle. Principle of Inferential Justification (PIJ) To be justified in believing one proposition p on the basis of another proposition e, one must be (1) justified in believing e and (2) justified in believing that e makes p probable.

PIJ has a high degree of prima facie plausibility. Consider the beliefs ‘water is pouring down outside the window’ and ‘it’s raining.’ In order for the former belief to justify the second, you must be justified in holding the former, and be justified in thinking that the former makes the latter probable. If you experience water pouring down outside the window, then both the first and the second clause seem to be satisfied beyond doubt. However, Fumerton (1995) argues, the skeptic will normally invoke the second clause first and then when the non-skeptic replies to the first challenge, she will invoke the first clause. The skeptic about the external world will first invite the nonskeptic to explain how our experiences about the external world make it probable

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PHENOMENAL DOGMATISM WITHOUT GOING EXTERNALIST



that there is an external world. The non-skeptic might reply by referring to past correlations between our experiences and the existence of external, mindindependent objects. In the past we have observed that our experiences of external objects were correlated with the existence of external objects. The skeptic will then invoke the first clause to shed doubt on the soundness of this sort of inductive argument. She will point out that in order to be justified in believing that there are such correlations, we would need to have direct conscious access to physical objects, a kind of access that is independent of experience. But we have no such experienceindependent access to the physical world, so we have no justification for our belief that there is a constant conjunction of external objects and sensory experiences about those objects.

2 Externalism Externalists have traditionally avoided the skeptical challenge by rejecting the second clause of PIJ. This follows from the externalist’s general claim that inferentially derived beliefs are justified only if the inferential process is a reliable one. As an example, consider Goldman’s reliabilism (Goldman 1979, also defended by Nozick 1981: 264 for justified belief). On this view, beliefs are non-inferentially justified when they are produced via a reliable belief-forming process. If I form the belief that it’s raining on the basis of the belief that water is pouring down outside the window, then the former belief is justified only if the process of forming the first belief on the basis of the second is a reliable process, regardless of whether or not I believe it. That is, I need not believe that water pouring down makes rain probable. In response to the reliabilist, the skeptic cannot appeal to us having the same justification in veridical and non-veridical scenarios in order to establish that we cannot have non-inferential justification for our beliefs because if reliabilism is true, then whether a belief is justified turns on the process by which it is formed and not on the phenomenology of sensory experience. Reliabilism thus denies the skeptic’s implicit assumption that non-inferential justification consists of a kind of direct conscious access to external objects. To have a justified belief about the external world is to have a belief formed by a reliable belief-independent process. Whether the belief stems from direct conscious access to the external world or not is irrelevant to the question of whether the belief is justified. Externalism thus provides an answer to the skeptical challenge. However, there are two serious problems with the externalist approach. The first problem is that it is not clear that traditional externalists have a way of avoiding what naive realists would find a counterintuitive result, viz., the result that beliefs based on veridical experience and beliefs based on a hallucination (e.g., after taking a drug) are justified in the same way as long as both beliefs are formed via belief-independent reliable processes. It is this thought that seems to be the heart of the “new evil demon problem” for reliabilism.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



BERIT BROGAARD

The second problem with traditional externalist frameworks is that it is not clear that their analysis of justification as a matter of being the result of a reliable process is an analysis of a philosophically interesting concept of justification. As Fumerton (1995) argues, if standard externalist accounts of justification are correct, then philosophers are not particularly competent in terms of answering normative questions in epistemology, so it would be wise to dismiss all types of skeptical inquiry as irrelevant to philosophy, which is an undesirable result.1 Fumerton’s position implies that if epistemology is to be of any interest to philosophers, then there are normative epistemic properties that are not reducible to naturalistic properties. Reliabilism is an attempt to reduce the notion of justification to a naturalistic property. Fumerton’s response to externalism is reminiscent of Moore’s open question argument, which aims at showing that moral facts cannot be natural facts. On the assumption that the natural properties of a given action are completely pinned down, it remains an open question whether that action is morally good or bad (Moore 1903). Moore concluded that moral facts are non-natural facts. This entails a denial of strong naturalism, the view that all moral facts can be deduced, at least in principle, from facts about physical entities on the lowest level of organization, for example, truths about the elementary particles and forces. A similar argument could be advanced for the view that facts about justification are non-natural facts. Completely specifying the natural properties of a belief-forming process, Fumerton would say, leaves open the question of whether the belief is epistemically justified. Against this one might argue that even if strong naturalism is false, at least weak naturalism is true. Weak naturalism is the view that the only entities that can be causally efficacious are physical entities. This view is also sometimes captured as the idea that the physical realm is causally closed, that is, all physical effects can be explained by microphysical phenomena and fundamental microphysical laws. If, however, weak naturalism is true and moral facts are non-natural, then it follows that moral facts cannot be causally efficacious (Harman 1986). It might seem to follow from this that human beings with physical brains cannot have knowledge of moral facts, which is an undesirable consequence. I think, however, that this last part of the argument is unsound. It would follow only if knowledge of moral facts were to require direct cognitive access to moral facts. As it turns out, Fumerton does hold a view that implies that knowledge requires direct cognitive access to moral facts. But this view is by no means mandatory. The version of dogmatism I defend below does not require this. So, on this view, one can have knowledge of moral facts even if moral facts are non-natural properties. By the same line of reasoning, one can have knowledge of epistemic facts even if epistemic facts are non-natural properties. So, it is indeed open to us to agree with Fumerton 1

Bergmann (2000) identifies some problems with this argument. Here, however, I will assume that Fumerton’s points about externalism are correct.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PHENOMENAL DOGMATISM WITHOUT GOING EXTERNALIST



that while reliabilism does succeed in avoiding the skeptical challenge, it is an unsatisfactory position because it reduces normative epistemic properties to naturalistic properties. It doesn’t follow from the view I defend that we cannot have knowledge of epistemic properties.

3 Naive Realism and Disjunctivism Naive realism and related disjunctive conceptions of experience avoid both of the problems that threaten to undermine externalist views like reliabilism. The naive realist’s notion of non-inferential justification is a variant on the one defended by Fumerton (see e.g. McDowell 1982). According to Fumerton, one has a noninferentially justified belief that p 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. This notion of non-inferential justification is well suited for the purposes of the naive realist. The notion of acquaintance, the naive realist may say, is simply the perceptual relation that obtains between a subject and a mind-independent physical object and its perceptible property instances. In the case of hallucinations the perceptual relation does not obtain. So, a belief based on a hallucination is not justified. Although John McDowell is not strictly speaking a naive realist, as he seems to allow that experiences have content, his position serves as a good example of a disjunctive theory that is committed to the view that beliefs based on hallucinations are not justified. Naive realism and disjunctivism are both versions of direct realism. According to McDowell, a perceptual relation obtains between the perceiver and a mind-independent physical object when a fact is made manifest to the perceiver. Within McDowell’s framework, a fact p is made manifest in an experience E just when E is a mental state that has p as one of its relata. Or as McDowell puts it, “[when a fact is made manifest] the fact itself is [the] object [of the experience]; so that its obtaining is not, after all, blankly external” (1982: 25). A further alleged virtue of direct realism is that unlike the reliabilist’s notion of non-inferential justification, the direct realist’s notion of non-inferential justification is arguably one that is philosophically interesting. This is because direct realism supposedly transcends traditional forms of externalism. As McDowell puts it in “Criteria, Defeasibility, and Knowledge”: The root idea is that one’s epistemic standing on some question cannot intelligibly be constituted, even in part, by matters blankly external to how it is with one subjectively. For how could matters beyond one’s ken make any difference to one’s epistemic standing? . . . But the disjunctive conception of appearances shows a way to detach this ‘internalist’ intuition from the requirement of non-question-begging demonstration. When someone has a fact made manifest to him, the obtaining of the fact contributes to his epistemic standing on the question. But

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



BERIT BROGAARD

the obtaining of the fact is precisely not blankly external to his subjectivity, as it would be if the truth about that were exhausted by the highest common factor. (1982: 26–7, italic added)

In the passage just cited McDowell responds to the common internalist intuition that the good epistemic standing of good perceptual cases cannot be grounded in something blankly external to the agent’s subjectivity. McDowell grants this intuition but then replies that (i) external facts can contribute to the perceiver’s subjectivity by being constituents (or relata) of a mental state, and (ii) if the difference in epistemic standing between veridical and non-veridical cases were not grounded in a difference between the states, then it would be grounded in something purely external to the states themselves. In arguing against his opponents McDowell appeals to the intuitive pull of a principle which he takes to be endorsed also by state internalists. The principle suggested by McDowell is the following. Principle of Internality The good epistemic standing of veridical states is not purely external to the state itself.

Given this principle and the assumption that there is a difference in epistemic standing between veridical and non-veridical cases, McDowell’s internality argument in favor of naive realism can be articulated as follows: Internality Argument 1. If the difference in the epistemic standing between good and bad cases is not grounded in a difference between the states themselves, then what grounds the difference in epistemic standing between veridical and non-veridical cases is purely external to the states themselves. 2. The difference in epistemic standing between veridical and non-veridical cases is not purely external to the states themselves. Conclusion: the difference in epistemic standing between veridical and non-veridical cases is grounded in a difference between the states themselves.

To get to a disjunctive conception of experience from this conclusion, we would need a further premise that externalism about justification is not a suitable option, and the assumption that the difference between good and bad cases is of fundamental importance. The option that remains, then, seems to be naive realism because only naive realism takes the different epistemic standings of good and bad cases to be grounded in a fundamental difference between good and bad cases, viz. a difference in terms of whether the perceiver stands in a perceptual relation to a mindindependent physical object. Although naive realism and the disjunctive conception of experience at first glance appear to avoid the problems threatening reliabilism, the claim that there is a difference in epistemic standing between good and bad cases runs into difficulties. As McDowellstyle disjunctivism is most commonly cashed out, the position is committed to the claim that all veridical (non-illusory, non-hallucinatory) perceptions have a better epistemic standing than the corresponding hallucinations (see e.g. Byrne and Logue

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PHENOMENAL DOGMATISM WITHOUT GOING EXTERNALIST



2008; Fish 2009b). It is further thought that this claim motivates a treatment of good and bad cases as fundamentally different. To take this to be the case, however, would be too hasty. There is evidence that McDowell does not in fact hold this view. In a footnote in “Criteria, Defeasibility, and Knowledge” he says: An experience subjectively indistinguishable from that of being confronted with a tomato, even if it results from confrontation with a tomato, need not count as experiencing the presence of a tomato. Another case in which it may not count as that is one in which there are a lot of tomato facades about, indistinguishable from tomatoes when viewed from the front. (1982: 26 nt. 39)

The sentiment McDowell here expresses (albeit in passing) is that not every veridical (non-illusory, non-hallucinatory) visual perception of an object counts as seeing the object and hence that not every veridical visual perception counts as a good case (in the visual case: a seeing). Presumably, McDowell would say that in such cases veridical perception does not have a better epistemic standing than the corresponding bad cases. Nonetheless, McDowell-style disjunctivism is typically construed as drawing a sharp distinction between veridical perception and illusions/hallucinations. Following Byrne and Logue (2008), let us call this common construal of McDowell-style disjunctivism ‘V v I/H disjunctivism.’2 V v I/H disjunctivism and V/I v H disjunctivism, I will argue, turn out to be epistemically unmotivated. In the next section I will show that naive realism in general fails to block the skeptical challenge. The objection to V v I/H disjunctivism can be articulated as follows: V v I/H Objection (1) Some veridical perceptions have no better epistemic standing than the corresponding hallucinations/illusions. (2) If (1), then a difference in kind between veridical perception and hallucination/illusion does not adequately account for their relative epistemic standings. Conclusion: A difference in kind between veridical perception and hallucination/ illusion does not adequately account for their relative epistemic standings. The argument is obviously valid. Moreover, given V v I/H disjunctivism, the second premise is very compelling. Part of McDowell’s motivation for disjunctivism is that when a fact is made manifest in experience, then the subject is in a position to know, whereas when a fact is not made manifest in experience, the subject is not in a position to know. As McDowell’s position is most commonly cashed out, a fact is made manifest in experience in all cases of veridical perception. For example, if I stand in the right sort of relation to a red chair, then a red-chair-fact is made manifest in my experience. So, if there are cases in which a veridical perception has

2 This is the claim that there’s a fundamental distinction between veridical perception and illusions/ hallucinations. As formulated, it is not a claim to the effect that the former has a better epistemic standing than the latter. This latter claim about epistemic standing would be something that the naive realist would need to argue for.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



BERIT BROGAARD

no better epistemic standing than the corresponding hallucination/illusion, then it is plausible that the fact-made-manifest thesis cannot account for the difference in epistemic standing between veridical perceptions and hallucinations/illusions. So, the second premise is compelling. The controversial premise is the first. I will offer two related arguments in favor of the first premise. You are looking at a red wooden chair. Unbeknownst to you, someone slowly removes pieces of wood from the chair while simultaneously feeding you drugs which ‘reconstruct’ the missing parts in your visual field. So, just as someone removes a piece of wood from the back of the chair, a drug is simultaneously administered, and it thus seems to you as if the chair is still intact. Of course, as we imagine the case, you are not aware of any of these changes. You are unaware that someone is taking apart the chair in front of you, and you are unaware that drugs are administered that ‘reconstruct’ the missing parts of the chair in your visual field. Now, at the beginning of this scenario your experience is a veridical (non-hallucinatory and non-illusory) perception, it then becomes an illusion, and finally, when all the pieces of the chair have been removed, your experience is a (falsidical) hallucination. On V v I/H disjunctivism, as illusive experiences (normally) are falsidical, no fact is made manifest in such cases. But consider now an arbitrary instant just before the chair is dismantled and drugs are administered to you to ‘reconstruct’ the missing parts. Since you believe that there is a red wooden intact chair in front of you throughout the dismantling process, you believe that there is a red wooden chair in front of you at this instant. Moreover, your belief is true. There is a red wooden chair in front of you. Your experience is a (non-hallucinatory and non-illusive) veridical perception. Nonetheless, your belief that there is a red wooden intact chair in front of you is not safe (see Williamson 2000). For in some of the closest situations in which you have the same belief, there is not a red wooden intact chair in front of you. Since your belief is not safe, it does not count as knowledge. In fact, you are not in a position to know that there is a red wooden intact chair in front of you. We thus have an example of veridical perception which does not put you in a position to know and which therefore does not differ in epistemic standing from the corresponding illusion. Note that it’s not because you cannot subjectively discern the difference between your veridical experience and your illusion that you are not in a position to know in the good case. We can grant that knowledge states are external states. It still follows that your belief does not (and could not) amount to knowledge because it fails to be safe. Of course, the disjunctivist could argue that safety is not required for knowledge. But this move would be idle. For, surely knowledge states must satisfy some anti-luck constraint. But no such constraint is satisfied in the envisaged case. If safety is not the right sort of anti-luck constraint, perhaps sensitivity is. S’s belief that p is sensitive just in case in all of the closest situations in which p is false, s does not believe that p. But your belief that there is a red wooden intact chair in front of you is not sensitive.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PHENOMENAL DOGMATISM WITHOUT GOING EXTERNALIST



Some of the closest situations in which there is no red wooden intact chair in front of you are the future situations in which, unbeknownst to you, small pieces of wood have been removed from the chair. But in these situations you believe that there is a red wooden intact chair in front of you. So, in some of the closest situations in which it is not the case that there is a red wooden intact chair in front of you, you believe that there is one. Hence, your belief is not sensitive. It may perhaps be said that your veridical perception and your illusions have different epistemic standings because, even though your veridical perception does not put you in a position to know that there is a red wooden intact chair in front of you, it gives you justification for believing that there is one. But this cannot be right. For, there is no difference in your internal justification for your beliefs that there is a red wooden intact chair in front of you throughout the dismantling process. And if justification is construed as safety (a move made by Williamson 2000), then there is no difference in your justification before and after the onset of the dismantling process. So, there is no difference in epistemic standing between your veridical perception before the onset of the dismantling process and your illusions after the chair is partly dismantled. In some discussions of disjunctivism illusions are treated as good cases, like veridical perceptions (see e.g. Langsam 1997 and Snowdon 1980). Following Byrne and Logue (2008), call this “V/I v H disjunctivism.” The main motivation for this treatment of illusions is that in the case of illusions, the subject is perceptually related to an object. Moreover, just like veridical perceptions illusions could reasonably be thought to give rise to knowledge. For example, if I am having an experience that attributes redness to a white chair I cannot come to know that there is a red chair in front of me but I can come to know that there is an object shaped in a certain way in front of me. We can still maintain a version of the fact-made-manifest thesis in this case. If I have an experience that attributes redness to a white chair, there is a fact made manifest to me, for example, the fact that there is an object shaped like a chair in front of me. The chair example just considered undermines the main motivation for V v I/H disjunctivism. But V/I v H disjunctivism, too, falls prey to a version of the chair problem. Consider one of the last instants at which you are having an illusion as of there being a red wooden intact chair in front of you. At one of the very next instants your experience will be a hallucination. On the view under consideration, illusions are supposed to put us in a position to know, whereas hallucinations are not. But in the envisaged scenario this is hardly so. Firstly, at one of the last instants during the dismantling process the cause of what you appear to be seeing is the drug. The drug makes the wooden stub that remains look like a much larger, differently shaped and differently colored object (viz, the chair). Secondly, in the closest situations in which you believe that there is an object in front of you, your belief is false. So, even your belief that there is an object in front of you fails to be safe. We thus have an example of an illusion which does not put you in a position to

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



BERIT BROGAARD

know anything and which therefore does not differ in epistemic standing from the corresponding hallucination. The chair example admittedly does nothing to undermine versions of disjunctivism that classify illusions sometimes as good cases and sometimes as bad cases (see Fish 2009a). Disjunctivists who hold this view could simply say that when only a few pieces of wood have been removed from the chair, the perceptual case is still a good one, and that this explains why the veridical hallucination and the illusion have the same epistemic standing. When, on the other hand, all but a stub of the chair has been removed the perceptual case is to be treated as a bad one, and this explains why the illusion and the hallucination have the same epistemic standing. The objection against standard versions of disjunctivism does not threaten to undermine all forms of naive realism either. For example, it doesn’t undermine the form defended by Brewer (2011). The reason for this is that Brewer would deny the standard distinction between illusions and hallucinations. On his view, if I perceive a square as a circle, this is a hallucination and not an illusion. However, as I will now argue, naive realism and disjunctivism do not have the ability to respond adequately to the skeptical challenge.

4 Why Naive Realism Does Not Solve the Skeptical Problem While naive realism and the disjunctive conception of experience enable us to have non-inferentially justified beliefs about external objects, the two views don’t obviously do anything to prevent the skeptical challenge with respect to inferential beliefs. The naive realist can say that we are non-inferentially justified in believing that, say, there are dark clouds in the sky, if we are seeing that that is the case, because in that case we are directly acquainted with the fact that there are dark clouds in the sky. If we are also directly acquainted with the belief that there are dark clouds in the sky and the correspondence relation obtaining between the thought and the fact, then we are non-inferentially justified in believing that it will rain. However, naive realism cannot provide the same account of how we can be justified in believing that it will rain on the basis of the belief that there are dark clouds in the sky. As the former belief is inferred from the belief that there are dark clouds in the sky, we are not directly related to the fact that it will rain. There are a couple of different ways that naive realists can explain how we might be justified in holding beliefs that are inferred from other beliefs. They might accept a form of externalism, such as reliabilism, which rejects PIJ. This will get them out of the skeptical challenge. However, if they make this move, then the main epistemic motivation for naive realism is undermined, as reliabilism avoids skepticism regardless of whether we accept naive realism. Furthermore, the concern that the reliabilist notion of justification isn’t an interesting philosophical concept lingers.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PHENOMENAL DOGMATISM WITHOUT GOING EXTERNALIST



Alternatively, naive realists can accept PIJ but say that we are justified in believing that dark clouds make rain likely, because the constant conjunction of dark clouds and rain in the past has made itself manifest to us. They might add that we have had direct conscious access to the dark clouds and the rain in the past. The problem, however, is that while it may be true that we have had direct conscious access to dark clouds and rain, it is less likely that we have had direct conscious access to the constant conjunction of clouds and rain, let alone the probability relation between clouds and rain. If, however, the probability relation between clouds and rain is inferred from the conjunctive instances, then the question arises how we come to have inferential justification for the probability relation between the conjunctive instances and the initial probability claim (viz. that dark clouds make rain likely). In other words, we would need to account for the following instance of the second clause of PIJ: A constant conjunction of dark clouds and rain makes it likely that (dark clouds make rain likely) It seems that we are stuck in a vicious circle or perhaps faced with a regress problem. So, it seems that the naive realist in the end is no better off than the externalist. As Fumerton points out (1995: ch. 7), one way to avoid these types of regresses is to treat probability relations obtaining between two kinds of occurrences as kinds of a priori necessary truths. Probability, on this Keynesian view, is an internal relation between kinds of occurrences. It’s similar in this respect to the relation of being darker than that obtains as a matter of necessity between black and white or the relation of being a lower note than that obtains between middle C and middle E on the piano. If there are such relations of probability, then we can be directly acquainted with them and avoid the vicious regress. Contrary to Fumerton’s conviction, however, the Keynesian view does not allow for inferentially justified beliefs about the external world. The reason is that probability claims about kinds of occurrences are not a priori necessities. For something to be a priori it must be conceptually necessary. But we cannot rule out on a priori grounds that the rest of history will not be the way it was in the past. The rest of history could conceivably be one in which dark clouds no longer make rain likely. But if the rest of history were that way, then it would be false in general that dark clouds make rain likely. So, we cannot know on a priori grounds that dark clouds make rain likely. Fumerton argues that Bertrand Russell (1948) entertained something like the Keynesian view of probability as a response to the problem of induction. According to Russell, no evidence can support the principle of induction. Since no evidence is available, we must regard them as first truths upon which our reasoning rests. This view, however, is very different from the Keynesian view of probability. Russell’s position is that we must treat probability claims about the past as epistemically primitive facts. But this is not the same as saying that we can know these facts on a priori grounds. The view Russell entertained is simply making an assumption about

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



BERIT BROGAARD

the past that can be challenged by the skeptic. This is the main reason that Russell abandoned the view.

5 Phenomenal Dogmatism A form of foundationalism that is reconcilable with a representational view of perception and that fares better than naive realism in terms of avoiding the skeptical problem is phenomenal dogmatism (Brogaard 2013a). This position can be articulated as follows: Phenomenal Dogmatism If it phenomenally seems to S as if p, then, in the absence of defeaters, S thereby has at least some degree of justification for believing that p. Phenomenal dogmatism holds that phenomenal seemings, or appearances of the perceptual, memory-based, intellectual, or introspective kind, confer prima facie justification on beliefs. Beliefs that are prima facie justified in this way are foundational, that is, they do not depend on any other beliefs for their justification. If S possesses a defeater, the defeater prevents the prima facie justification from constituting ultima facie justification. Defeaters should here be understood as reasons to which one has potential conscious access. Causal influences on beliefs do not count as defeaters. For example, suppose a judge makes harsher judgments if he hasn’t eaten than if he is given a sandwich. In this case the sandwich is not a defeater of the judge’s belief. On Pollock’s well-known distinction, an undercutting defeater is a reason for supposing that one’s ground for believing p is not indicative of the truth of the belief that p (Pollock 1984: 38). A rebutting defeater is a reason for holding the negation of p or for holding some proposition, q, incompatible with p. If it seems to me that the table is red, I have prima facie justification for believing that the table is red, according to phenomenal dogmatism. But if I find out that I have an eye problem that may or may not cause me to misperceive colors, I have an undercutting defeater of my belief. The red appearance of the table no longer supports my belief that the table is red, so I no longer have justification for believing that the table is red. Likewise, if several individuals in a police line-up seem to an eyewitness as if they committed the crime, each appearance is then an undercutting defeater of the justification for her belief that any one of the individuals in the line-up is the criminal. So the eyewitness does not have justification for believing that any particular individual in the line-up is the criminal, though she may still have justification for believing the existential claim that at least one of the identified individuals is the criminal. If it seems to me that you are sad, I have prima facie justification for believing that you are sad, according to phenomenal dogmatism. But if you tell me that you are not sad, then I have a rebutting defeater of my belief. The appearance that you are sad is no longer indicative of the truth of the belief that you are sad, so I no longer have justification for believing that you are sad.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PHENOMENAL DOGMATISM WITHOUT GOING EXTERNALIST



Let us look closer at how phenomenal dogmatism avoids skepticism about the external world. The first step in the skeptic’s argument against the internalist is to show that we could have exactly the same perceptual justification for the belief that there are external objects in a skeptical scenario and in the actual world. At first glance, phenomenal dogmatism seems susceptible to this first and devastating skeptical move. After all, the phenomenal seeming that there is a tomato in front of me confers the same prima facie evidence on my corresponding belief in the real world and in the skeptical scenario. But if this is so, then the skeptic can say that the justification in question cannot be non-inferential. However, the dogmatist can respond to the skeptic in much the same way that the reliabilist does. She can reject the skeptic’s assumption that non-inferential justification requires direct conscious access to the external world. Phenomenal dogmatism holds that prima facie non-inferential justification for a belief that p requires merely a phenomenal seeming that p rather than direct conscious access to external objects. This reply to the skeptic has the consequence that one can be equally noninferentially justified in both a skeptical scenario and a normal world. This consequence goes against McDowell’s intuition that there is a difference in epistemic status between good and bad cases of perception but we have already seen that that position potentially runs into trouble. One might also wonder how solid McDowell’s intuition in fact is. Dogmatists do not share it. Moreover, as Fumerton points out, a well-known objection to reliabilism, as originally formulated, relies on the intuition that one can be equally justified in a skeptical and a non-skeptical scenario (Fumerton 1995: 113). The objection runs as follows. Consider human beings in a demon world who have exactly the same hardware and software as we do. Moreover, the input and outputs of the software are exactly the same as ours. The input is data from the sensory systems and the outputs are beliefs about the physical objects in the external world. In this scenario, reliabilism would have the consequence that we are justified in holding the beliefs we do, whereas the victims of the demon are not. There is a strong intuition to the effect that this is the wrong outcome. The victims may well lack warrant (in Plantinga’s 1993a, 1993b sense) and definitely lack knowledge, but many people feel that they do not lack justification. And Goldman (1988) did indeed revise his original formulation of reliabilism to accommodate this intuition. In Epistemology and Cognition he argues that reliability is to be understood with reference to normal worlds. This modification allows that the victims in the demon world can have beliefs that are justified to the same extent as perceivers in normal worlds. The above reply to the skeptic, however, does not fully resolve the skeptical challenge. For, unlike the reliabilist, the phenomenal dogmatist will admit that there is inferential justification in addition to non-inferential justification. Moreover, she accepts the PIJ. But consider now a case in which it phenomenally seems to me that there are dark clouds in the sky. This seeming confers immediate justification on my belief that there are dark clouds in the sky. I infer on the basis of this belief that it

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



BERIT BROGAARD

is going to rain. If I am justified in believing that it is going to rain, then I must also be justified in believing that the dark clouds in the sky make rain likely to occur. But what is the nature of my justification for believing that dark clouds in the sky make rain likely? It may seem that the dogmatist could say that her justification for the probability claim is itself inferentially based on conjoint occurrences of dark clouds and rain in the past. But saying this causes trouble. For, according to PIJ, in order to account for how we can be justified in believing the probability claim (viz., that dark clouds make rain likely), we would need to provide an account of our justification for believing that multiple instances of conjunction makes the probability claim likely. That is, we would need to account for the following instance of the second clause of PIJ: We are justified in believing that a constant conjunction of dark clouds and rain makes it probable that (dark clouds make rain probable). It seems that we quickly face either a vicious circle or vicious regress. However, there is hope for the dogmatist. It phenomenally seems to me that in the past there has been a constant conjunction of dark clouds and rain. This seeming is a memorybased seeming that confers prima facie justification on the belief that dark clouds make rain likely. It may seem that I could not have a memory-based seeming of this sort without having performed an inference from individual instances of dark clouds conjoined with rain. This, however, is not the case. The first time I observe dark clouds and rain, the two experiences are stored separately in memory. When these observations are repeated, however, I form a memory association between dark clouds and rain. This is quite similar to the memory association you may have formed between a certain scent and a person or event in your past, or the association that most of us have formed between blue and sadness. These memory associations are formed when the brain generates new synaptic connections between two types of stored information. If information about dark clouds and information about rain are neurally connected, then there is no more of an inference involved in associating dark clouds and rain than there is in associating blue and sadness. When you retrieve a memory of a constant conjunction of dark clouds and rain, this is accompanied by the memory-based seeming that dark clouds make rain likely, as likelihood and constant conjunction are conceived of in the same way by ordinary folks. The failure of ‘it seems’ to agglomerate with conjunction (failure of closure) may seem to present a problem for this proposal. Failure of ‘it seems’ to agglomerate with conjunction is evident in cases like the following. It may seem likely to me that ticket 1 wins, and it may seem likely to me that ticket 2 wins. But it may not seem likely to me that tickets 1 and 2 win. So, if I have a memory seeming of dark clouds and I have a memory seeming of rain, it doesn’t follow that I have a memory seeming of dark clouds conjoined with rain. However, the memory-based seeming that dark clouds make rain likely is not a case of agglomeration with conjunction. It’s a formation of a

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PHENOMENAL DOGMATISM WITHOUT GOING EXTERNALIST



memory association on the basis of repeated exposure to a constant conjunction of two types of events. It’s based on an ancient skill that benefited our ancestors who needed to form memory associations between, say, danger and dangerous things in order to survive. It seems, then, that the dogmatist can meet the skeptical challenge with respect to prima facie justification. But what about ultima facie justification? Can the possibility of skeptical scenarios serve as defeaters of prima facie justification? If so, then we are saddled with the odd view that those who believe in the possibility of skeptical scenarios cannot be justified in believing anything about the external world, whereas those who don’t believe in the possibility of skeptical scenarios perhaps because they have never considered these issues can be prima facie and ultima facie justified in their beliefs about the external world. As it turns out, however, skeptical scenarios are not defeaters of prima facie justified beliefs. Consider my prima facie justified belief that there is a computer in front of me. The possibility of a skeptical scenario, for example, the possibility that I am a brain in a vat, is not a rebutting defeater of this belief, as ‘there is a computer in front of me’ and ‘it is possible that I am a brain in a vat’ are perfectly compatible. So, if the possibility of skeptical scenarios is a defeater, it is an undercutting defeater. The ground of my belief is the phenomenal seeming that there is a computer in front of me. Does the possibility that I am a brain in a vat give me a reason to think that the phenomenal seeming that there is a computer in front of me is not indicative of the truth of the belief that there is a computer in front of me? If dogmatism is true, then the answer is “no.” In order for the skeptical scenario to defeat the ground of my belief, the skeptic would need to show that because there is no direct conscious access to objects in the skeptical scenario, we cannot have prima facie justified beliefs in the skeptical scenario. But this argument only works against theories that take direct conscious access to external objects to be required for beliefs to be justified. Since this standard skeptical move does not get off the ground, the possibility of the skeptical scenario cannot serve as an undercutting defeater of my justification for my belief that there is a computer in front of me. Dogmatism thus provides a far more plausible response to the skeptical challenge than naive realism.

6 Bootstrapping Phenomenal dogmatism, however, has received its fair share of criticism. One problem that has been raised for dogmatism is that it is doubtful that all seemings are equally epistemically significant. Peter Markie provides two illustrations of this point. The first runs as follows. Suppose that we are prospecting for gold. You have learned to identify a gold nugget on sight but I have no such knowledge. As the water washes out of my pan, we both look at a pebble,

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



BERIT BROGAARD

which is in fact a gold nugget. My desire to discover gold makes it seem to me as if the pebble is gold; your learned identification skills make it seem that way to you. According to (PC), the belief that it is gold has prima facie justification for both of us. Yet, certainly, my wishful thinking should not gain my perceptual belief the same positive epistemic status of defeasible justification as your learned identification skills. (2005: 356–7)

The problem here is that because it seems to both the expert and the novice that the pebble is gold, phenomenal dogmatism entails that they both have prima facie justification for believing that it is gold. Since the novice does not possess a defeater that he has access to, and the novice’s belief is based on his seeming, Markie argues, the expert and the novice are equally justified in believing that the pebble is gold. This is unintuitive, as the novice’s belief is based on wishful thinking. Intuitively, the novice does not have any justification for believing that the pebble is gold. I am somewhat skeptical about this case. I am tempted to think that most of us would be sufficiently acquainted with gold to be prima facie justified in believing that a gold nugget in a pan is gold in the envisaged circumstances. But nothing hinges on this concern, as we can change the example to involve an item that is harder for most people to spot, such as valuable antique furniture. In that scenario it does seem prima facie implausible that the antique furniture expert and the novice should be equally justified in believing that a piece of furniture is valuable. Markie’s second objection to phenomenal dogmatism runs as follows: Suppose that I perceive the walnut tree in my yard, and, having learned to identify walnut trees visually, it seems to me that it is a walnut tree. The same phenomenological experience that makes it seem to me that the tree is a walnut also makes it seem to me that it was planted on April 24, 1914. Nothing in the phenomenological experience or my identification skills supports things seeming this way to me. There is no date-of-planting sign on the tree, for example. (2005: 357)

Markie here allows for the possibility that my phenomenal seeming can confer prima facie justification on my belief that the tree is a walnut. It could do so because it is plausible that the tree being a walnut is presented in my experience. But it is not plausible that the property of having been planted on April 24, 1914 is presented in my experience. The property of having been planted on April 24, 1914 is not a visually detectable property. We cannot detect it through vision alone. But if it really does seem to me that the tree is a walnut and was planted on April 24, 1914, then phenomenal dogmatism would seem to entail that my belief that the tree is a walnut and my belief that it was planted on April 24, 1914 have the same degree of prima facie justification. On the assumption that I don’t have access to any defeaters, I have justification for my belief that the tree was planted on April 24, 1914. But if the only ground for my belief that the tree was planted on April 24, 1914 is an appearance that this is so, then clearly I do not have any justification for my belief. Although these problem cases may be threatening to some forms of dogmatism, they do not present a problem for the version of the theory that I am defending here.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PHENOMENAL DOGMATISM WITHOUT GOING EXTERNALIST



This is because the only seemings that can confer justification on belief are phenomenal seemings. Epistemic seemings do not confer justification on belief. Because epistemic seemings disappear in the presence of a defeater when the agent is rational, they are not firmly grounded in the content of perceptual, introspective, or memoryrelated experiences (Brogaard 2013b). Suppose that I hear on the radio that a tsunami is going to cause flooding in my area and I say to my housemates “It seems that we ought to evacuate.” If the radio host later announces that the earlier announcement was a hoax, “a tsunami is going to cause flooding in my area” is still the content of what I heard but it no longer seems to me to be in my best interest to evacuate. So my seeming that it’s probably in my best interest to evacuate is not grounded in the content “a tsunami is going to cause flooding in my area.” Of course, it might seem to me to be in my best interest to evacuate on the basis of a coarse-grained memory to the effect that I probably should evacuate. But if all I remember and continue to remember is that I probably should evacuate, then the seeming doesn’t go away as the result of a radio announcement. So in this case the seeming is not epistemic. As epistemic seemings do not confer justification on belief, Markie’s two examples do not cause trouble for the version of dogmatism I have set forth above. In the first of the two cases the novice’s belief that a particular piece of furniture is valuable is not phenomenal, as it goes away in the presence of a defeater if the agent is rational. If the expert tells the novice that the furniture is not valuable, then the novice will suspend his belief. The same goes for the seeming that the tree was planted on April 24, 1914 in the second case. This seeming could not possibly be a phenomenal seeming. The human visual system does not generate phenomenal seemings of this sort. In a recent paper Susanna Siegel (2012) has provided two related counterexamples to phenomenal dogmatism (or close variants). For argument’s sake, she grants that beliefs, attitudes, expectations, and so forth, can penetrate and change the content of visual experience. For example, if you expect that someone is sad, you may, partially in virtue of that expectation, have a visual experience that represents them as sad. Siegel’s first counterexample proceeds as follows. Case 1: Angry-looking Jack Jill believes, without justification, that Jack is angry at her. The epistemically appropriate attitude for Jill to take toward the proposition that Jack is angry at her is suspension of belief. But when she sees Jack, her belief makes him look angry to her. If she didn’t believe this, her experience wouldn’t represent him as angry.

On Siegel’s view, because Jill believes that Jack is angry at her, Jill has a visual experience that represents Jack as angry. Had she not had the belief that he was angry at her, she wouldn’t have had a visual experience of him being angry at her. According to phenomenal dogmatism, if it seems to Jill that Jack is angry at her, then absent defeaters, she has justification for believing that he is angry at her. The

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



BERIT BROGAARD

problem this case poses for phenomenal dogmatism is this. While Jill’s initial belief had no justification, once her belief is firmly grounded in her visual experience, she suddenly does have justification for her belief. This is strange, as it somehow was her unjustified belief that generated the justification for that very belief. The second case is parallel to the first: Case 2: Preformationism Many of the first users of microscopes favored preformationism about mammalian reproduction. Some of them claimed to see embryos in sperm cells that they examined using a microscope. Prior to looking at sperm cells under the microscope, the preformationist favored the hypothesis that there were embryos in healthy sperm cells. At the time no theory of mammalian reproduction was well-confirmed, and the epistemically appropriate attitude to take toward preformationism was suspension of belief. But the preformationists did not suspend belief. When they looked under the microscope, they saw embryos in the sperm cells.

The preformationists’ visual experiences represented sperm cells as containing embryos. What caused them to see sperm cells in this light, however, was their prior unjustified belief that sperm cells contain embryos. According to phenomenal dogmatism, appearances can confer justification on belief in the absence of defeaters. As the preformationists did not have any defeaters of their fundamental beliefs, their visual experience of sperm cells containing embryos conferred justification on their initial beliefs. So, it seems that an unjustified belief can somehow generate justification for itself through a process of cognitive penetration. This just seems plainly wrong. While Siegel’s two cases present a serious threat to some forms of dogmatism, they do not threaten to defeat the version I defend. This is because the appearances, or seemings, she invokes in her counterexamples are epistemic. Recall that an epistemic seeming is one that disappears in the presence of a defeater in a rational agent, whereas a non-epistemic seeming is one that lingers, at least to some extent, even in the presence of a defeater. If Jack convinces Jill (or gives her some reason to believe) that he is not angry at her (for example, over the phone) before she sees him, she will—if she is rational—no longer believe that Jack is angry at her. So if she is rational, it will no longer seem to her as if he is angry at her when she sees him. As the appearance of him being angry at her does not linger in the presence of a defeater if she is rational, the seeming is epistemic and so it does not confer prima facie justification on her belief. The same applies in the case of preformationism. If someone convinces the preformationist that sperm cells do not contain embryos, or at least gives him some reason to believe that they do not, the preformationist, if rational, will no longer believe that sperm cells contain embryos. So he will no longer have a visual experience of sperm cells containing embryos when he looks in the microscope. Since his seeming does not withstand the presence of a defeater assuming rationality, it is epistemic. So phenomenal dogmatism does not entail that his visual seeming confers prima facie justification on his belief.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PHENOMENAL DOGMATISM WITHOUT GOING EXTERNALIST



7 Conclusion Naive realists often appeal to the threat of skepticism as a reason to accept the position. Naive realism, they say, is the only reasonable way of avoiding skepticism about the external world. The idea is that when a perceptual relation obtains between the subject and the external world, then the subject has non-inferential justification for the corresponding belief. The reason that this is said to avoid skepticism is that the skeptic cannot provide a skeptical scenario in which the justification is the same as, or just as good as, the justification the subject actually has. It turns out, however, that naive realism cannot provide an adequate account of inferential justification. I have argued that a different position, phenomenal dogmatism, which is reconcilable with a representational view of perception, avoids the threat of skepticism both in the case of non-inferential justification and inferential justification. So, the naive realist’s epistemological reason for accepting the view doesn’t hold up and in fact turns out to indirectly support a representational view of perception.3

References Alston, William P. (2002). “Sellars and the Myth of the Given,” Philosophy and Phenomenological Research, 65: 69–86. Bergmann, Michael. (2000). “Externalism and Skepticism,” Philosophical Review, 109: 159–94. Brewer, Bill. (2011). Perception and Its Objects. Oxford: Oxford University Press. Brogaard, Berit. (2013a). “Phenomenal Seemings and Sensible Dogmatism,” in C. Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. New York: Oxford University Press, 270–92. Brogaard, Berit. (2013b). “It’s Not What It Seems. A Semantic Account of ‘Seems’ and seemings,” Inquiry, 56: 210–39. Byrne, Alex and Heather Logue. (2008). “Either/Or,” in Adrian Haddock and Fiona Macpherson (eds), Disjunctivism: Perception, Action, Knowledge. Oxford: Oxford University Press. Fish, William. (2009a). “Disjunctivism,” Internet Encyclopedia of Philosophy [Online Journal], http://www.iep.utm.edu/disjunct/#SH4a, August 2010. Fish, William. (2009b). Perception, Hallucination, and Illusion. New York: Oxford University Press. Fumerton, Richard. (1995). Metaepistemology and Skepticism. Lanham, MD: Rowman and Littlefield. Goldman, Alvin. (1979). “What Is Justified Belief?,” in George Pappas (ed.), Justification and Knowledge. Dordrecht: Reidel, 1–23. Goldman, Alvin. (1988). Epistemology and Cognition. Cambridge, MA: Harvard University Press. Harman, Gilbert. (1986). “Moral Explanations of Natural Facts,” Southern Journal of Philosophy, 24: 69–78. 3

I am grateful to Michael Bergmann for comments on an earlier version of this chapter and John Doris for help with the title.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



BERIT BROGAARD

Langsam, Harold. (1997). “The Theory of Appearing Defended,” Philosophical Studies, 87: 33–59. Markie, Peter. (2005). “The Mystery of Direct Perceptual Justification,” Philosophical Studies, 126: 347–73. McDowell, John. (1982). “Criteria, Defeasibility, and Knowledge,” Proceedings of the British Academy, 68: 455–79. Cited as reprinted in Thomas Baldwin and Timothy Smiley (eds), Studies in the Philosophy of Logic and Knowledge. Oxford: Oxford University Press, 2005: 7–29. Moore, G. E. (1903). Principia Ethica. Cambridge: Cambridge University Press. Nozick, Robert. (1981). Philosophical Explorations. Cambridge, MA: Harvard University Press. Plantinga, Alvin. (1993a). Warrant and Proper Function. Oxford: Oxford University Press. Plantinga, Alvin. (1993b). Warrant: The Current Debate. Oxford: Oxford University Press. Pollock, John. (1984). “Reliability and Justified Belief,” Canadian Journal of Philosophy, 14: 103–14. Russell, Bertrand. (1948). Human Knowledge: Its Scope and Limits. New York: Simon and Schuster. Siegel, Susanna. (2012). “Cognitive Penetrability and Perceptual Justification,” Nous, 46: 201–22. Snowdon, Paul. (1980). “Perception, Vision and Causation,” Proceedings of the Aristotelian Society, 81: 175–92. Williamson, Timothy. (2000). Knowledge and Its Limits. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

5 Experience and Evidence Abridged Susanna Schellenberg

Prologue to ‘Experience and Evidence Abridged’ In my paper ‘Experience and Evidence’, I argue for an externalist view of perceptual evidence that makes room for a phenomenal conception of evidence.1 More specifically, I argue that perceptual experience provides us with both phenomenal and factive evidence in the case of an accurate perception, and with phenomenal evidence in cases where things go wrong, such as hallucination and illusion. To a first approximation, we can understand phenomenal evidence as determined by how our environment sensorily seems to us when we are experiencing. To a first approximation, we can understand factive evidence as necessarily determined by the environment to which we are perceptually related such that the evidence is guaranteed to be an accurate guide to the environment. I argue that the rational source of both phenomenal and factive evidence lies in employing perceptual capacities that we have in virtue of being perceivers. In showing that both kinds of evidence have the same rational source, I provide a unified account of perceptual evidence and its rational source in perceptual experience. What follows is a reprint of the first two sections of that paper. In section 1, I distinguish perceptual evidence from introspective evidence. In section 2, I develop an externalist conception of phenomenal evidence.

1 Perceptual Evidence and Introspective Evidence If we have evidence, it is rational to heed this evidence. Perceptual evidence is evidence provided by perceptual experience. In so far as perceptual experience is directed at our environment, the evidence that perceptual experience provides us with is of (or as of ) our environment. The idea that perceptual evidence is of (or as of ) our environment is neutral on a whole range of vexed questions. It is neutral on whether perceptual This chapter is a reprint of the first two sections of Schellenberg (2013) with a new prologue by permission of Oxford University Press on behalf of the Mind Association. 1

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



SUSANNA SCHELLENBERG

evidence has content. It is neutral on what the nature of its content is—assuming there is evidential content. Moreover, it is neutral on what the relation is between the content of perceptual evidence (if any) and the perceptual experience that provides us with perceptual evidence. Finally, it is neutral on whether all aspects of our perceptual evidence are accessed or even accessible.2 I take a stance on all these choice points in section 3 of Schellenberg (2013). But for the most part, the arguments in this chapter can be accepted irrespective of what stance one takes on these issues. Regardless of how perceptual evidence is understood, it must be distinguished from introspective evidence. Introspective and perceptual evidence differ in what they are of: while perceptual evidence is of (or as of) one’s environment, introspective evidence is of (or as of) one’s experience or some other mental state. They differ in their source: while perceptual evidence stems from perception, introspective evidence stems from introspection. They differ in what one attends to: while one gains perceptual evidence in virtue of attending to one’s environment, one gains introspective evidence in virtue of attending to one’s experience or some other mental state (which may be of one’s environment). When I speak of experience as providing us with evidence directly, I mean that we need not attend to our experience to have the evidence. So we need not introspect our experience to gain evidence: we have evidence simply in virtue of experiencing.3 I am not denying that when we experience we can introspect our experience and thereby gain introspective evidence. However, as I will argue in the next section, experience yields evidence without us having to introspect our experience. Indeed, I will argue—contra Williamson—that even when we are hallucinating our experience yields at least some evidence without having to resort to an appearance proposition. On Williamson’s view, the evidence one has when one hallucinates is an appearance proposition of the form ‘it seems to me that p’ and so is provided by attending to the fact that one’s environment seems a certain way to one. Appearance propositions involve appearance concepts—for example ‘it seems’ or ‘it appears’— and entertaining such a proposition requires the ability to refer to oneself. Animals that do not possess appearance concepts and that are not capable of self-reference 2 For the view that all evidence is propositional, see Williamson (2000); for the view that evidence can be non-propositional, see Plantinga (1993). For the view that evidence is necessarily accessible, see Chisholm (1977); for the view that evidence is not necessarily accessible, see again Williamson (2000). 3 This constraint is neutral on a whole range of ways of thinking of direct and indirect perception. For a discussion of the notions of ‘direct’ and ‘indirect’ perception, see Jackson (1977) and Snowdon (1992). It should be noted that on a radical view of the transparency of experience we are never aware of properties of our experience but only ever of what our experience is about. There are both empirical and philosophical reasons to deny that experience is radically transparent in this way. To name just one reason: when our epistemic access to our environment changes—for instance, because we take off our glasses—our experience will be different. The difference is due to how we experience our environment. While we are not necessarily aware of the fact that the difference in experience is due to a change in the experience rather than the environment, we can be. The fact that we can be aware of this is reason alone to reject the thesis that experience is radically transparent. For a discussion of this set of issues, see Smith (2002) and Martin (2002).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

EXPERIENCE AND EVIDENCE ABRIDGED



can hallucinate. They gain evidence in virtue of hallucinating even though they are not capable of entertaining appearance propositions. After all, they act on their hallucination.4 While it is a fact that the environment seems a certain way to us when we experience, we should distinguish between this fact and the sensory state we are in when such a fact holds. If we gain evidence in hallucination only by attending to the fact that it sensorily seems to us as if our environment is a certain way and so only by attending to our experience (rather than by attending to our environment, albeit failing to perceive), then the evidence we gain in hallucination is not provided directly through experience. I will present a view of perceptual evidence on which evidence need not be understood as propositionally or conceptually structured and on which phenomenal evidence need not involve appearance concepts.

2 The Phenomenal Evidence Argument The basic argument for the thesis that perceptual experience provides us with phenomenal evidence goes as follows: 1. If a subject S is perceptually directed at her environment (while not suffering from blindsight or any other form of unconscious perception), then it sensorily seems to S as if her environment is a certain way (regardless of how it in fact is).

premiss

2. If it sensorily seems to S as if her environment is a certain way (regardless of how it in fact is), then S is in a sensory state that provides phenomenal evidence.

premiss

3. If S is perceptually directed at her environment (while not suffering from blindsight or any other form of unconscious perception), then S is in a sensory state that provides phenomenal evidence.

1, 2

The first premiss makes a claim about what is the case when we are perceptually directed at our environment. We can be perceptually directed at our environment without being perceptually related with our environment: when suffering a hallucination that is subjectively indistinguishable from a perception, we are perceptually directed at our environment, but fail to be perceptually related with our environment. The premiss states that it sensorily seems to us as if our environment is a certain way, if we are perceptually directed at our environment.5 It is neutral on whether our environment could sensorily seem the very same to us regardless of whether we are 4 Williamson (2000: 199) denies that such animals gain evidence through their hallucination. Such a view requires an independent explanation of why animals act on their hallucinations. 5 One might object that the notion of being perceptually directed to one’s environment is equivalent to the notion of the environment sensorily seeming a certain way to one. In response, we can say that one could have a notion of being perceptually directed at one’s environment while being eliminativist about sensory seemings. This alone shows that the notion of being perceptually directed is distinct from the notion of sensory seemings.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



SUSANNA SCHELLENBERG

perceiving, hallucinating, or suffering an illusion. So it is compatible with a whole range of views about the nature of sensory seemings. Moreover, the premiss is neutral on whether experience has content. So it is compatible with a whole range of views about the nature of experience. Since the relevant sensory seemings are restricted to those in which our environment seems a certain way to us, the scope of the premiss does not extend to the ways things seem to us when we imagine. After all, when we visually imagine an object in our environment, it is not our environment that sensorily seems a certain way to us. It is rather what we imagine (e.g. our mental imagery) that sensorily seems a certain way to us.6

2.1 Premiss 2: sensory states and phenomenal evidence The second premiss of the argument is more controversial. It states that sensory states provide us with phenomenal evidence. Accounts on which evidence is necessarily factive (Williamson 2000) and disjunctivist accounts (Snowdon 1981, McDowell 1982) will reject this premiss. In order to give support to this premiss, we need to address the question of what the relationship is between sensory states and phenomenal evidence.7 Since a sensory state is a kind of mental state, the thesis that sensory states provide phenomenal evidence entails—together with the theses that only mental states provide phenomenal evidence and that phenomenal evidence exists only if it is provided by something—the widely accepted thesis that our phenomenal evidence supervenes on our mental states.8 One might argue that there is a much stronger relation between phenomenal evidence and mental states, namely, identity. But for the sake of the phenomenal evidence argument, the relation of supervenience is all that is needed. A different way of understanding the question of what relationship there is between sensory states and phenomenal evidence is as a question about their epistemic relation. The key epistemological questions are: what is the epistemic bridge that gets us from being in a sensory state to having phenomenal evidence? More generally, why is it rational to heed the testimony of our senses—especially if unwittingly we happen to be hallucinating? In different ways, these questions ask for the motivation behind premiss 2.9 I will give support to premiss 2 by arguing that It is controversial whether blindsighters are perceptually directed at their environment. One could argue that they do not perceive, but merely detect or register particulars in their environment. Dretske (2006) argues that there is no such thing as unconscious perception and so would deny that blindsighters perceive. If there is no such thing as unconscious perception, then the qualifying clause in premiss 1 can be dropped. 6

The second premiss does not overgeneralize to imagination for the same reason. For the purposes of this chapter, I am following Chisholm (1966) and Jackson (1977) in taking the relation between sensory seemings and sensory states to be a simple one. For dissenting views, see Sosa (2007) and Bengson (forthcoming). 8 See Feldman and Conee (1985) and Pryor (2000) for versions of this view and Gupta (2006), White (2006), Wright (2007), DeRose (2011), and McGrath (2013) for critical discussions. 9 Here and throughout, I understand ‘rational’ in an epistemic sense. I am not here concerned with practical rationality. 7

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

EXPERIENCE AND EVIDENCE ABRIDGED



sensory states provide us with phenomenal evidence, since sensory states are systematically linked (in ways to be explained) to the particulars that they single out in the case of an accurate perception. Due to the existence of this systematic link it is rational to heed the testimony of our senses. What is the notion of rationality in play? For present purposes, it will suffice to work with the following understanding: if it is rational to heed the testimony of the senses, then a person who does not heed the testimony of her senses is blameworthy—provided she does not have defeaters. She is, for example, subject to the criticism that she is ignoring relevant information that is available to her. In order to get a better grip on the question of why it is rational to heed the testimony of our senses, it will be helpful to consider the shortcomings of internalist conceptions of evidence. This conception of evidence goes back to at least Russell (1984 [1913]) and arguably to Descartes (1641, especially Meditation II). Russell understood evidence in terms of sense data, that is, strange particulars that are directly present to the mind. Neo-Russellians and more generally evidential internalists understand perceptual evidence in terms of conscious mental states that can be the very same regardless of the environment of the experiencing subject (e.g. Pollock 1974, Feldman and Conee 1985, Pryor 2000, and Tucker 2010). If our conscious mental states can be the very same regardless of our environment and if these conscious mental states determine our perceptual evidence, then our evidence will be the very same in the good and the bad case—that is, our evidence will be the very same regardless of whether we are accurately perceiving or suffering a hallucination.10 But if perceptual evidence is the very same in the good and the bad case, then it is mysterious why it would be rational to heed the testimony of our senses (see Goldman 1999 for this line of criticism). It is plausible that the reason for why it is valuable to take how our environment seems to us at face value is because doing so constitutes a useful way of pursuing an accurate view of the world. Evidence can play that role, however, only if there is a systematic link between our sensory seemings and the way our environment actually is. In so far as evidential internalists do not account for such a link, they fail to account for the role of evidence as being a guide to how the world is. In fairness, it must be noted that at least some evidential internalists take phenomenal evidence as determined simply by how the world sensorily seems to us, where that seeming need not be a guide to how the world actually is.11 So they are 10 Illusions can be understood as a version of the good or the bad case. For discussion, see Antony (2011). For present purposes, we can remain neutral on how best to classify them. So as to avoid unnecessary complications, I will focus on the uncontroversial good and bad cases: accurate perception and hallucination. In section 2.3, I will show how the suggested view applies to illusions. 11 See for example Pollock and Cruz (2004); though note that they talk of justification, rather than evidence. They argue that justification bears no deep connection to truth, but is rather to be understood in internalist procedural terms.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



SUSANNA SCHELLENBERG

unlikely to be moved by the above line of argument. But the aim was not to argue against evidential internalists. The aim was to motivate the claim that an account of perceptual evidence ought to explain why it is rational to heed the testimony of our senses. The thesis that evidence is a guide to how the world is puts into focus what phenomenal evidence is evidence for. Evidence is always evidence for something. Phenomenal perceptual evidence is evidence for what our experience is of—or would be of, were we perceiving. In order to make this explicit in the phenomenal evidence argument, we need to reformulate premiss 2 as follows: Premiss 2*: If it sensorily seems to S as if her environment is a certain way (regardless of how it in fact is), then S is in a sensory state that provides phenomenal evidence for the presence of particulars of the type that the sensory state is of in the good case. How should we understand this? Consider Percy who perceives a white cup on the desk in front of him and Hallie who suffers a subjectively indistinguishable hallucination as of a white cup on the desk in front of her. Percy’s sensory state is of her environment and provides phenomenal evidence that there is a white cup on the desk. Similarly, Hallie’s sensory state provides phenomenal evidence that there is a white cup on the desk. So Percy and Hallie both have phenomenal evidence in virtue of their environment seeming a certain way to them. The argument in support of premiss 2* goes as follows: 2a. If it sensorily seems to a subject S as if her environment is a certain way (regardless of how it in fact is), then S is in a sensory state that is systematically linked to external, mind-independent particulars of the type that the sensory state is of in the good case.

premiss

2b. If S is in a sensory state that is systematically linked to external, mindindependent particulars of the type that the sensory state is of in the good case, then S is in a sensory state that provides phenomenal evidence for the presence of particulars of the type that the sensory state is of in the good case.

premiss

2.* If it sensorily seems to S as if her environment is a certain way (regardless of how it in fact is), then S is in a sensory state that provides phenomenal evidence for the presence of particulars of the type that the sensory state is of in the good case.

2a, 2b

The conclusion is the reformulation of the second premiss of the basic phenomenal evidence argument, that is, the premiss for which we needed further support. If we conjoin this argument for why sensory states provide phenomenal evidence with the basic phenomenal evidence argument, we get the following comprehensive phenomenal evidence argument:

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

EXPERIENCE AND EVIDENCE ABRIDGED

1.

If a subject S is perceptually directed at her environment (while not suffering from blindsight or any other form of unconscious perception), then it sensorily seems to S as if her environment is a certain way (regardless of how it in fact is).



premiss

2a. If it sensorily seems to S as if her environment is a certain way (regardless of how it premiss in fact is), then S is in a sensory state that is systematically linked to external, mindindependent particulars of the type that the sensory state is of in the good case. 2b. If S is in a sensory state that is systematically linked to external, mindindependent particulars of the type that the sensory state is of in the good case, then S is in a sensory state that provides phenomenal evidence for the presence of particulars of the type that the sensory state is of in the good case.

premiss

2.* If it sensorily seems to S as if her environment is a certain way (regardless of how it in fact is), then S is in a sensory state that provides phenomenal evidence for the presence of particulars of the type that the sensory state is of in the good case.

2a, 2b

3.* If S is perceptually directed at her environment (while not suffering from blindsight or any other form of unconscious perception), then S is in a sensory state that provides phenomenal evidence for the presence of particulars of the type that the sensory state is of in the good case.

1, 2*

We already discussed premiss 1. In the rest of this section, I will give support to premisses 2a and 2b.

2.2 Premiss 2a: sensory states and perceptual capacities In order to give support to premiss 2a, it will be necessary to show that sensory states are systematically linked to external, mind-independent particulars of the type that the sensory state is of in the good case and to specify how that systematic linkage is to be understood. Doing so will require presenting a modest externalist view of sensory states. The basic idea of this view is that when we perceive, we employ perceptual capacities by means of which we differentiate and single out particulars in our environment. The relevant particulars are external and mind-independent objects, events, property-instances, and instances of relations. Sensory states are understood as determined by employing perceptual capacities in a sensory mode, that is, modes such as seeing, hearing, touching, smelling, or tasting.12 I will argue that if a subject S’s environment sensorily seems to contain F particulars to her (regardless of how it in fact is), then S is in a sensory state that is determined by employing perceptual capacities that function to single out F particulars. Consider Percy who perceives a white cup on a desk. He employs his capacity to discriminate white from other colours and to single out white in his environment. Similarly, he employs his capacity to differentiate and single out cup shapes from, say, Here and throughout, ‘determined’ is understood in the sense of ‘at least partially determined’. This leaves open whether there might be other determinants. 12

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



SUSANNA SCHELLENBERG

computer shapes and lamp shapes. He may also employ the capacity to differentiate and single out cups from, say, computers and lamps. The important point is that in virtue of employing such capacities, he is in a sensory state that is of a white cup. What happens in hallucination? When we hallucinate, we employ the very same capacities that in a subjectively indistinguishable perception are employed while being perceptually related to external, mind-independent particulars. Since in hallucination, we are not perceptually related to a particular, we fail to single out a particular in our environment. We merely purport to single out a particular. As a consequence, at least some of the capacities employed are baseless. They are baseless in the sense that the targets of discrimination and selection—external, mindindependent particulars—are absent. Analogously, if we employ concepts, but fail to refer, the concepts employed remain empty. Consider Hallie who suffers a hallucination as of a white cup on a desk. Like Percy, she employs the capacity to discriminate and single out white from other colours and she employs the capacity to differentiate and single out cup shapes from, say, computer shapes and lamp shapes. Since she is hallucinating rather than perceiving and so not perceptually related to a white cup, the capacities she employs are baseless. Yet even though she fails to single out any white cup, she is in a sensory state that is as of a white cup in virtue of employing the capacity to discriminate and single out white from other colours and cup shapes from other shapes. How should we understand the perceptual capacities in play? They can be understood to be discriminatory, selective capacities, concepts, or some kind of functional property. There is good scientific evidence that discriminatory, selective capacities are the cognitively most low-level mental capacities employed in perception, so I will focus on this specific kind of perceptual capacity.13 A discriminatory, selective capacity is a low-level mental capacity that functions to differentiate, single out, and in some cases type the kind of particulars that the capacity is of. For example, if we possess the discriminatory, selective capacity that functions to differentiate and single out red, we are in a position to differentiate instances of red from other colours in our environment and to single out instances of red. More generally, to possess a discriminatory, selective capacity is to be in a position to differentiate and single out a particular of the type that the capacity is of, were one related to such a particular.14 So if we possess such a capacity, then—assuming no finking, masking, or other exotic case obtains (see 13

For discussions of the role of basic visual capacities and pre-attentive discrimination in early vision, see Julesz (1981), Watson and Robson (1981), Sagi and Julesz (1985), Malik and Perona (1990), Krummenacher et al. (2010), and To and Gilchrist (2011). 14 The notion of capacity in play can but need not be understood in a teleological, phylogenetic, virtue epistemological, or ontogenetic manner. For such accounts, see Millikan (1989), Neander (1996), Sosa (1991, 2007), Zagzebski (1996), Greco (2001, 2010), and Burge (2003, 2010). As I will argue in the rest of this section, a sensory state provides phenomenal evidence in so far as it is determined by capacities that are metaphysically and explanatorily dependent on the good case. As I will show, accepting this idea is compatible with accepting that such capacities may more often than not be used in a way that fails to produce accurate representations of the environment.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

EXPERIENCE AND EVIDENCE ABRIDGED



Lewis 1997)—the following counterfactual should hold: if we were perceptually related to a particular that the capacity functions to single out, then we would be in a position to single out such a particular.15 Singling out a particular is a proto-conceptual analogue of referring to a particular. Non-rational animals and infants as young as four months old can perceptually single out objects and property-instances in their environment, yet they do not have the capacity to refer. While referring requires conceptual capacities, singling out particulars requires no such capacities. There are further analogies between discriminatory, selective capacities and concepts. Like concepts, the same discriminatory, selective capacity can be employed in different environments and in this sense such capacities are repeatable. It is worth noting that discriminating between two particulars does not require attending to both particulars. It requires only registering their differences—however much in the background of one’s sensory state this registering may occur. It is unclear what it would be to single out, say, the shade of a leaf without registering how it differs in at least one respect from its surround. More generally, it is unclear what it would be to single out a particular without registering how it differs in at least one respect from other particulars. How does appealing to such capacities help understand sensory states in a way that supports premiss 2a of the phenomenal evidence argument? The suggestion is that sensory states are determined by employing perceptual capacities in a sensory mode. Any two experiences in which all the same capacities are employed in the same sensory mode will have the same sensory character if all else is equal. Although such capacities are necessarily determined by functional connections between perceivers and their environment, arguably they can be employed even if one is misperceiving or hallucinating. After all, the capacities are determined by general, functional relations between the organism and its environment—for instance, global patterns of the organism’s response to its environment—and not by individual token responses. Yet, one could be prompted to employ such capacities due to nonstandard circumstances: unusual brain stimulations or misleading distal inputs. If this is right, then we can employ a discriminatory, selective capacity even if a relevant particular is not present—where a relevant particular is a token of the type that the capacity functions to single out. The capacities employed account for the fact that in hallucinations we can purport to single out particulars: from a first-person perspective it can seem as if we were perceptually related to particulars in our environment. Since sensory states are understood as determined by employing perceptual capacities rather than the capacities themselves, it is not revealed in our sensory character whether the capacities are baseless. So it is not revealed in our sensory 15 The inference from a claim about perceptual capacities to a counterfactual fails in finking, masking, and similarly exotic cases. However, all the standard ways of fixing the disposition-to-counterfactual inference can be exploited for the capacity-to-counterfactual inference. See in particular Lewis (1997). Finding a formulation of the capacity-to-counterfactual inference that is indefeasible in light of all possible finking, masking, and similarly exotic cases would be a project of its own. Therefore, I will here work on the assumption that no such exotic cases obtain. This assumption is independently plausible.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



SUSANNA SCHELLENBERG

character whether we succeed in differentiating and singling out particulars, and so whether we are perceiving or hallucinating. An example will help illustrate the point. We possess the capacity to perceive red. Sometimes we employ this capacity successfully to single out something red, and sometimes we employ this capacity but fail to single out anything red. In the latter case, we suffer a hallucination as of something red, or an illusion that an object we perceive is red when in fact it is not red. The important point is that one can distinguish the employment of the capacity—what perceptions, hallucinations, and illusions have in common—from discriminating and singling out a particular—the matter on which perceptions, hallucinations, and illusions differ. It is the employment of the capacity that determines the sensory state. Whether or not a particular is singled out does not affect the sensory state. If it is right that two experiences in which ceteris paribus all the same perceptual capacities are employed in the same sensory mode have the same sensory character, then subjectively indistinguishable perceptions, hallucinations, and illusions will share a metaphysically substantial common factor. The common factor is determined by the perceptual capacities that the subject employs in a sensory mode. But as I will show shortly, the fact that there is such a common factor does not imply that we are aware of a common factor, nor does it imply that the good case is analysed as a conjunction of a common factor and some additional element, such as a causal perceptual relation. Before I show how this way of understanding sensory states supports premiss 2a, I will address four potential misconceptions. First, the idea that sensory states are determined by employing discriminatory, selective capacities in a sensory mode is compatible with there being additional aspects that determine sensory states. Such aspects may be sensations, appearance properties, sense data, qualia, intentional objects, phenomenal properties, awareness relations to property-clusters, or (uninstantiated) universals—to name just a few options.16 For present purposes, we can remain neutral both on whether there are any such additional aspects and—if there are any—on what their nature is. We can remain neutral on this, since the aim here is not to give a full account of sensory states, but rather to analyse what it is about sensory states that makes them rational to heed. Second, we possess and make use of many discriminatory, selective capacities that are not phenomenally relevant—even when we perceive. I have not argued that

16 For sensations, see Peacocke (1983); for appearance properties, see Shoemaker (2007); for sense-data, see Robinson (1994); for qualia, see Levine (1983), Chalmers (1996), Block (2003), McLaughlin (2007); for intentional objects, see Harman (1990), Lycan (1996), Crane (1998); for phenomenal properties, see Chalmers (2006), Block (2007); for (uninstantiated) universals, see Dretske (1995), Byrne (2001), Tye (2002); for property-clusters, see Johnston (2004). For an excellent recent account of consciousness and overview of the current debate, see Hill (2009). While the proposal that sensory states are determined by employing perceptual capacities is compatible with there being such additional aspects that determine sensory states, the suggestion allows for a way to analyse sensory states without appealing to phenomenologically or metaphysically problematic entities, such as sense-data, qualia, intentional objects, or sensory awareness relations to (uninstantiated) universals, property-clusters, or other abstract entities. For a discussion of the problems of such views, see my (2011).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

EXPERIENCE AND EVIDENCE ABRIDGED



whenever we use such a capacity, we are in a sensory state. I have argued only that sensory states should be understood in terms of employing perceptual capacities in a sensory mode. We can accept this thesis while acknowledging that there are many capacities—including discriminatory, selective capacities—the employment of which has no repercussions for our conscious mental lives. Third, it is crucial that employing discriminatory, selective capacities is not just a matter of differentiating particulars, but also of singling out particulars. Due to this the sensory character of perceiving an instance of red is distinct from the sensory character of perceiving an instance of blue. Both cases may include differentiating red and blue, but in the former case, an instance of red is singled out, while in the latter case, an instance of blue is singled out. So the capacities employed are distinct and the sensory states differ. Finally, all sorts of things can be understood to discriminate, including thermometers and sunflowers. When I speak of discriminatory, selective capacities, I mean always a kind of low-level mental capacity. Since I am not trying to analyse what makes a capacity mental, I will help myself to the notion of a mental capacity. The notion of capacities in play does not apply to thermometers and sunflowers, since the relevant capacities are a kind of mental capacity and thermometers and sunflowers do not have mental capacities. Now, how does analysing sensory states in terms of employing perceptual capacities help explain why it is rational to heed the testimony of our senses? The aim was to develop a way of thinking about sensory states on which they are systematically linked to what they are of in the good case, and so, a way of thinking about sensory states that supports premiss 2a of the phenomenal evidence argument. How does appealing to perceptual capacities help develop such an account? As I will argue, sensory states are systematically linked to what they are of in the good case in the sense that the perceptual capacities employed in the bad case are explanatorily and metaphysically parasitic on their employment in the good case. There is an explanatory primacy of the good over the bad case since one can give an analysis of the perceptual capacities employed in the bad case only by appealing to their role in the good case. Consider again Hallie who suffers a hallucination as of a white cup on a desk. Even though she fails to single out anything white, she is in a sensory state that is as of an instance of white in virtue of employing the capacity to discriminate and single out white from other colours. She would single out an instance of white, were she in the good case—assuming again that no finking, masking, or other exotic case obtains. After all, she is employing a discriminatory, selective capacity the very function of which is to differentiate white from other colours and to single out white in her environment. In this sense, we need to refer to what Hallie would discriminate between and what she would single out in the good case in order to explain the role of the capacities she employs in the bad case. Underlying this explanatory primacy there is a metaphysical primacy of the good over the bad case. More specifically, the explanatory primacy is licensed by a more

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



SUSANNA SCHELLENBERG

basic metaphysical primacy. There is a metaphysical primacy of the good over the bad case in so far as one can possess the discriminatory, selective capacities employed in the bad case only in virtue of being the kind of being that could employ those very capacities in the good case. Call this the metaphysical primacy thesis. Why should we accept this thesis? The function of discriminatory, selective capacities is to differentiate and single out particulars of the type that the capacity is of. It would be unclear what it would mean to possess a discriminatory, selective capacity, the very function of which is to single out a kind of particular, without being in a position to single out such a particular when perceptually related to one. So the ‘could’ in the metaphysical primacy thesis should be understood to indicate a metaphysical rather than an epistemic possibility.17 An example will help illustrate the point. If we possess the capacity to discriminate and single out red from other colours, we can use this capacity to single out red in our environment. Were we not in a position to use our capacity in this way, when perceptually related to an instance of red, we would not count as possessing the capacity. In short, while discriminatory, selective capacities can be employed in hallucination, they are necessarily determined by relations between perceivers and their environment in so far as the function of the capacity is to differentiate and single out, say, instances of red in perception. In this sense, there is a metaphysical priority of the good over the bad case. The metaphysical priority thesis entails the counterfactual that if we possess a discriminatory, selective capacity, then—assuming that no finking, masking, or other exotic case obtains—we would be in a position to single out a relevant particular, were we related to such a particular. However, it also entails the counterfactual that if we possess such a capacity, we would fail to single out a relevant particular, were we not related to such a particular. Similarly, the explanatory priority thesis entails symmetric counterfactuals. So why should we accept that there is an asymmetry between the good and the bad case? Why not say that the bad case is no less fundamental than the good case? After all, perceptual capacities are characterized both by how they behave in the good and the bad case. In responding to this challenge, I will focus on the metaphysical priority thesis, since it licenses the explanatory priority thesis. My explanation for why the metaphysical priority thesis holds carries over to an explanation of why the explanatory priority thesis holds. While the metaphysical priority thesis entails symmetric counterfactuals, the thesis is not to be identified with them. The asymmetry buttressing the thesis is an asymmetry of function. Perceptual capacities function to single out particulars. They do not function to fail to single out particulars. It is compatible with this that they may be employed in hallucination thereby failing to single out particulars. In order to support this, it will be necessary to take a closer look at the notion of function in play. The heart has the function to pump blood. It does not have the More specifically, the ‘could’ should be understood to indicate a restricted metaphysical possibility. A plausible restriction is to scenarios in which the subject’s mental constitution is not radically altered. 17

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

EXPERIENCE AND EVIDENCE ABRIDGED



function to fail to pump blood—though in the bad case it will fail. One possible way to understand this asymmetry is in terms of evolution: the function of the heart is what it was selected for (Millikan 1984). However, it need not be understood in an evolutionary way. Any plausible account of natural function will support the idea that the heart has the function to pump blood rather than the function to fail to pump blood. Likewise, perceptual capacities have the function to single out particulars in the environment. They do not have the function to fail to single out particulars. An evolutionary account of function would posit that perceptual capacities evolved for the purpose of singling out particulars rather than for the purpose of failing to single out particulars: they were selected to single out particulars. However again, there is no need to explain the asymmetry in evolutionary terms. On any plausible account of natural function, we can say that perceptual capacities function to single out particulars rather than function to fail to single out particulars. Accepting the metaphysical priority thesis is compatible with acknowledging that one could possess a perceptual capacity that one has never actually used successfully in perception. Moreover, the perceptual capacities employed in hallucinations need not have been acquired through perceptions. They might be innate, they might have been acquired through testimony, or they might have been arrived at through imagination. So the metaphysical priority thesis does not imply that we must have successfully used a perceptual capacity in the past to count as possessing such a capacity. It implies only that we must be in a position to use the capacity successfully when perceptually related to a relevant particular, thereby singling out that very particular in our environment. It is worth highlighting that my argument does not depend on the notion of discriminatory, selective capacities. It depends only on the idea that sensory states are systematically linked to what they are of in the good case in the sense that the perceptual capacities employed in the bad case are explanatorily and metaphysically parasitic on their employment in the good case. The perceptual capacities in play can be understood to be discriminatory, selective capacities, but alternatives are to understand the capacities to be concepts or some kind of functional property. I focus on discriminatory, selective capacities only since they are arguably the cognitively most low-level mental capacities employed in perception. One can accept my argument while appealing to perceptual capacities that are not discriminatory, selective capacities. Now, does the existence of a perceptual capacity require the existence of at least one successful employment of that capacity? While it is possible to possess such a capacity without having been perceptually related to any particulars of the type that the capacity singles out in the good case, it is plausible that any such perceptual capacity is grounded in perception in so far as the existence of the capacity depends on perceptions of the particulars that the capacity singles out.18 If this is right, then it 18

This is not implied by the argument of the chapter. The phenomenal evidence argument requires only a weaker claim, namely, that any perceptual capacity is grounded in how things would come out in the

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



SUSANNA SCHELLENBERG

follows that there cannot exist any such perceptual capacity that is not grounded in perception. However, it does not follow from this that an individual subject must have had perceptions of the particulars that the capacity singles out to possess the relevant capacity. It follows only that there can exist a perceptual capacity that functions to single out a kind of particular, only if a particular of that kind has been perceived by someone, somewhere. The argument for the metaphysical priority of the good over the bad case does not depend on resolving the question of whether the existence of a perceptual capacity requires the existence of at least one successful application by someone, somewhere. However, depending on what stance one takes on this issue one must either reject or accept the metaphysical possibility of the scenario of a world of brains in a vat that can hallucinate. Regardless of what stance one takes on this issue, the suggested capacity view allows that a brain in a vat in our world could have hallucinations and so phenomenal evidence. Can Swampman possess perceptual capacities? Swampman is a being that came into existence through a bolt of lightning and so has no causal history (Davidson 1970). If perceptual capacities are understood in an evolutionary way, then Swampman could not possess the capacities in play. However, if they are understood in a non-evolutionary way, then Swampman could possess the relevant capacities. After all, no past experiences are necessary to possess such capacities. The condition for their possession is understood counterfactually: if one possesses the capacity to single out red, then one would be able to single out an instance of red, were one related to such an instance. For present purposes, we can remain neutral on whether capacities are understood in an evolutionary or a non-evolutionary sense. This opens the question of whether perceptual capacities are dependent on the particulars they function to single out.19 There are at least three different ways of understanding this question and my response is different depending on which way the question is understood. One way of understanding it is as a question about possessing capacities: could a subject possess a perceptual capacity, even though she has never been perceptually related to a particular of the kind that the capacity functions to single out? In response: yes. After all, the capacities could be innate, acquired through testimony, or acquired through imagination and she may have been unlucky and never been perceptually related to a relevant particular. Another way of understanding it is as an existence question: could a perceptual capacity exist that functions to single out a kind of particular that does not exist in our world, such as supersaturated red? In response: given what I argue in the chapter, that is possible. However, for empiricist reasons that go beyond the scope of this chapter, one might think that perceptual capacities must be grounded in perception in the sense that any

good cases. However, for empiricist reasons independent of the argument of this chapter, it is plausible to assume that such capacities are grounded in actual perceptions and not just possible perceptions. For a discussion of such reasons, see Goodman (1955). 19

Thanks to Matt McGrath for raising this question.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

EXPERIENCE AND EVIDENCE ABRIDGED



given perceptual capacity must have been used by someone, somewhere. On such a view, hallucinations of supersaturated red need to be analysed in terms of employing both the capacity to single out instances of red and the capacity to single out instances of supersaturatedness. A third way of understanding the question is as a question about employing capacities: could a perceptual capacity be employed even if the relevant particular is not present? In response: yes. After all, the very same perceptual capacity can be employed in hallucination and in perception. I have argued that sensory states are systematically linked to particulars of the type that the sensory state is of in the good case in the sense that the perceptual capacities employed in the bad case are explanatorily and metaphysically parasitic on their employment in the good case. The idea that sensory states are determined by employing such perceptual capacities is what supports premiss 2a of the phenomenal evidence argument.

2.3 Premiss 2b: phenomenal evidence and systematic linkage Recall that premiss 2b of the phenomenal evidence argument has it that if a subject is in a sensory state that is systematically linked to external and mind-independent particulars of the type that the sensory state is of in the good case, then she is in a sensory state that provides phenomenal evidence for the presence of particulars of the type that the sensory state is of in the good case. This premiss supports the crucial transition from the metaphysical fact that a sensory state is systematically linked to the external and mind-independent F particulars it is of in the good case to the epistemic fact that such a sensory state provides evidence for the presence of F particulars. The truth of premiss 2b depends on two principles. The first principle is that if sensory states are systematically linked to what they are of in the good case in the sense specified, then it is epistemically rational to heed the testimony of these sensory states. The second principle is that if it is epistemically rational to heed the testimony of sensory states, then they provide evidence. I will give support to each principle in turn. I argued that sensory states are systematically linked to particulars of the type that the sensory state is of in the good case in the sense that the perceptual capacities employed in the bad case are explanatorily and metaphysically parasitic on their employment in the good case. Sensory states are systematically linked to particulars in this way in so far as it is the function of the perceptual capacities that determine the sensory state to single out the relevant particulars. In speaking of it being the function of perceptual capacities to single out the relevant particulars, I do not mean to speak of their actual reliability but rather of how they are to be understood metaphysically. It is the function of a perceptual capacity to single out, say, instances of red. This is so regardless of how often the capacity is employed successfully to single out an instance of red. So this way of understanding why it is rational to heed the testimony of our senses has the advantage of not relying on any form of reliabilism. Our senses frequently lead us astray. Nevertheless, they provide us with evidence. On the

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



SUSANNA SCHELLENBERG

suggested capacity view, it is rational to heed the testimony of our senses since sensory states are systematically linked to the particulars that they are of in the good case. The notion of systematic linkage in play is understood in terms of a metaphysical and explanatory primacy notion, which is not a reliabilist notion. If perceptual capacities are employed in perception, then they happen to be reliable. However, even in this case it is the primacy of the good over the bad case that gives experience its epistemic force. On the account presented, the epistemic force of perceptual experience does not depend on whatever reliability (if any) perceptual experience might have. The second principle states a sufficient condition for something to count as evidence. It follows from a substantive but largely uncontroversial view about evidence, namely, that it is a crucial property of evidence that if it is epistemically rational to heed x in the absence of defeaters, then x provides evidence.20 Now, one might object that beliefs are linked to what they are of in the good case, but it is not rational to treat beliefs as evidence. So why is it rational to treat sensory states as evidence but not beliefs?21 In response, we can concede that many things are in some way linked to what they are of in the good case. It is not rational to treat all those things as evidence. However, I argued that sensory states are systematically linked to particulars of the type that they are of in the good case in the sense that the perceptual capacities employed in the bad case are explanatorily and metaphysically parasitic on their employment in the good case. So the systematic linkage between sensory states and what they are of in the good case was understood in a specific way. The capacities employed in perception link perceptual states with particulars in the environment. Indeed, perception is our primordial connection to particulars in our environment. For present purposes, the crucial difference between perception and belief is that perceptual capacities function to single out particulars, while the capacities employed in belief do not necessarily have this function. Any belief that is about particulars is arguably parasitic on perception. The capacities that determine beliefs are not systematically linked to what they are of in the good case in the sense that there is an explanatory and metaphysical primacy of their employment in the good case. Therefore, the argument provided for why it is rational to heed the testimony of our senses does not overgeneralize to beliefs. Now, what if we assume for the sake of argument both that beliefs are a kind of sensory state and that the capacities that determine beliefs are explanatorily and 20 For discussions of this property of evidence, see Ayer (1972); Kelly (2003, 2007); Neta (2003, 2008); Weatherson (2005); and Pryor (2012). An interesting question is what the connection is between the strength of the evidence we have for a proposition and our confidence in that proposition. For discussion of the relation between having evidence for p and having confidence in p, see Neta (2003, 2008) and Silins (2005). Since our concern here is restricted to the questions of what evidence perceptual experience provides us with and why it is rational to heed it, we can bracket this issue for the purposes of this chapter. I reserve a detailed discussion of how the account developed here connects to questions about confidence for a future paper. 21 Thanks to Alex Byrne and David Chalmers for pressing me on this point.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

EXPERIENCE AND EVIDENCE ABRIDGED



metaphysically parasitic on their employment in the good case? On these two controversial assumptions, it is plausible that beliefs provide us with evidence.22 So on these assumptions, the argument provided for why it is rational to heed the testimony of our senses generalizes to beliefs. It does not however overgeneralize and so would not be a problem for the developed capacity view, since beliefs are now understood to have many of the fundamental properties of perceptual states.

2.4 Coda I have argued that our phenomenal evidence in the bad case is brought about by employing the very same perceptual capacities that in the good case allow us to perceptually navigate our environment. While these capacities are determined by functional relations to the particulars they single out in perception, we can employ the same capacities while failing to single out a relevant particular. So having phenomenal evidence is compatible with our perceptual capacities being employed baselessly. As a consequence, hallucinations provide us with tangible, though misleading phenomenal evidence. So while the developed notion of phenomenal evidence is externalist in that phenomenal evidence is determined by employing perceptual capacities and the capacities employed in the bad case are both metaphysically and explanatorily parasitic on their employment in the good case, we can have phenomenal evidence even when we are in the bad case. The developed notion of phenomenal evidence is internalist only in so far as the phenomenal evidence of two experiencers in different environments can be the very same. It is not internalist regarding the accessibility of the evidence. More importantly, it is not internalist in so far as our phenomenal evidence is understood in terms of an asymmetric dependence of the bad on the good case.23 This externalist notion of phenomenal evidence makes room for the idea that having evidence is a matter of being in an epistemic position that is a guide to how the world is, while allowing that we can have evidence even if we happen to have been led astray and so are in a state that is not accurate with regard to our environment. As a consequence, the suggested capacity view shows how experience provides us with phenomenal evidence even in the bad case without retreating to introspective evidence.24 22

Indeed, Harman’s (1973) coherentist view of justification suggests—albeit for different reasons—that beliefs provide us with justification. 23 See Pryor (2001: 105–8) and Wedgwood (2002) for useful distinctions between ways of understanding the access requirement on our evidence and more generally different forms of epistemic internalism. 24 Thanks to John Bengson, Rachael Briggs, David Chalmers, Katie Elliott, Reinaldo Elugardo, James Genone, Alvin Goldman, John Greco, Leon Leontyev, John Maier, Angela Mendelovici, John Morrison, Ram Neta, Casey O’Callaghan, Jonathan Schaffer, Nico Silins, Ernest Sosa, Kurt Sylvan, and Joshua Spencer for detailed comments and to Ned Block, Marian David, Branden Fitelson, Mikkel Gerken, Anna-Sara Malmgren, Chris Peacocke, David Rosenthal, and Jason Stanley for helpful discussions and email exchanges. Particular thanks are due to Benj Hellie, Adam Pautz, and Matt McGrath, who presented comments and raised excellent questions on versions of this chapter at the Carolina Metaphysics Workshop, the APA Eastern Division Meeting 2012, and NeuPhi, respectively. I am grateful also to members of the audiences at those occasions for their probing questions and helpful suggestions. I am indebted to members

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



SUSANNA SCHELLENBERG

References Antony, Louise. (2011). ‘The Openness of Illusions’, Philosophical Issues, 21: 25–44. Ayer, A. J. (1972). Probability and Evidence. New York: Columbia University Press. Bengson, John. (Forthcoming). ‘The Intellectual Given’, Mind. Block, Ned. (2003). ‘Mental Paint’, in Martin Hahn and Bjorn Ramberg (eds), Reflections and Replies: Essays on the Philosophy of Tyler Burge. Cambridge, MA: MIT Press, 165–200. Block, Ned. (2007). ‘Max Black’s Objection to the Mind-Body Identity’, in Torin Alter and Sven Walter (eds), Phenomenal Concepts and Phenomenal Knowledge. Oxford: Oxford University Press, 249–306. Burge, Tyler. (2003). ‘Perceptual Entitlement’, Philosophy and Phenomenological Research, 67: 503–48. Burge, Tyler. (2010). Origins of Objectivity. Oxford: Oxford University Press. Byrne, Alex. (2001). ‘Intentionalism Defended’, Philosophical Review, 110: 199–240. Chalmers, David. (1996). The Conscious Mind. New York: Oxford University Press. Chalmers, David. (2006). ‘Perception and the Fall from Eden’, in Tamar Gendler and John Hawthorne (eds), Perceptual Experience. Oxford: Oxford University Press, 49–125. Chisholm, Roderick. (1966). A Theory of Knowledge. Upper Saddle River, NJ: Prentice Hall. Chisholm, Roderick. (1977). Theory of Knowledge. Englewood Cliffs, NJ: Prentice Hall. Crane, Tim. (1998). ‘Intentionality as the Mark of the Mental’, in Anthony O’Hear (ed.), Contemporary Issues in the Philosophy of Mind. Cambridge: Cambridge University Press. Davidson, Donald. (1970). ‘Mental Events’, in Donald Davidson, Essays on Actions and Events. Oxford: Clarendon Press, 207–25. Originally published in Lawrence Foster and Joe William Swanson (eds), Experience and Theory. Boston, MA: University of Massachusetts Press. DeRose, Keith. (2011). ‘Questioning Evidentialism’, in Trent Dougherty (ed.), Evidentialism and Its Discontents. Oxford: Oxford University Press, 137–46. Descartes, René. (1641). ‘Meditations on First Philosophy’, in John Cottingham, Robert Stoothoff, and Dugald Murdoch (eds), The Philosophical Writings of Descartes (Vol. 2). Cambridge: Cambridge University Press, 1–62. Dretske, Fred. (1995). Naturalizing the Mind. Cambridge, MA: MIT Press. Dretske, Fred. (2006). ‘Perception without Awareness’, in Tamar Gendler and John Hawthorne (eds), Perceptual Experience. Oxford: Oxford University Press, 147–80. Feldman, Richard and Earl Conee. (1985). ‘Evidentialism’, Philosophical Studies, 48: 15–34. Goldman, Alvin. (1999). ‘Internalism Exposed’, Journal of Philosophy, 96: 271–93. Goodman, Nelson. (1955). Fact, Fiction, and Forecast. Cambridge, MA: Harvard University Press. Greco, John. (2001). ‘Virtues and Rules in Epistemology’, in Abrol Fairweather and Linda Trinkaus Zagzebski (eds), Virtue Epistemology. Oxford: Oxford University Press, 117–41.

of the New York University consciousness reading group, the corridor reading group, and the Rutgers epistemology group. Thanks finally to audiences at the Arché Research Centre in St. Andrews, Cambridge University, Glasgow University, Temple University, University of California Berkeley, University of California Irvine, University of Massachusetts Amherst, University of Miami, University of Warwick, York University, as well as the Epistemology of Philosophy conference at the University of Cologne, the Epistemology workshop at the University of Aarhus, the Perception, Action and Time workshop in Barcelona, the Perceptual Capacities workshop at the ANU, and the Perception and Knowledge conference at the University of Graz.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

EXPERIENCE AND EVIDENCE ABRIDGED



Greco, John. (2010). Achieving Knowledge: A Virtue-Theoretic Account of Epistemic Normativity. Cambridge: Cambridge University Press. Gupta, Anil. (2006). Empiricism and Experience. Oxford: Oxford University Press. Harman, Gilbert. (1973). Thought. Princeton, NJ: Princeton University Press. Harman, Gilbert. (1990). ‘The Intrinsic Quality of Experience’, Philosophical Perspectives, 4: 31–52. Hill, Christopher. (2009). Consciousness. Cambridge: Cambridge University Press. Jackson, Frank. (1977). Perception. Cambridge: Cambridge University Press. Johnston, Mark. (2004). ‘The Obscure Object of Hallucination’, Philosophical Studies, 120: 113–83. Julesz, Bela. (1981). ‘A Theory of Preattentive Texture Discrimination Based on First-Order Statistics of Textons’, Biological Cybernetics, 41: 131–8. Kelly, Thomas. (2003). ‘Epistemic Rationality as Instrumental Rationality: A Critique’, Philosophy and Phenomenological Research, 66: 612–40. Kelly, Thomas. (2007). ‘Evidence and Normativity: Reply to Leite’, Philosophy and Phenomenological Research, 75: 465–74. Krummenacher, Joseph, Anna Grubert, and Hermann Muller. (2010). ‘Inter-Trial and Redundant-Signals Effects in Visual Search and Discrimination Tasks: Separable PreAttentive and Post-Selective Effects’, Vision Research, 50: 1382–95. Levine, Joseph. (1983). ‘Materialism and Qualia: The Explanatory Gap’, Pacific Philosophical Quarterly, 64: 354–61. Lewis, David. (1997). ‘Finkish Dispositions’, Philosophical Quarterly, 47: 143–58. Lycan, William G. (1996). Consciousness and Experience. Cambridge, MA: MIT Press. Malik, Jitendra and Perona Pietro. (1990). ‘Preattentive Texture Discrimination with Early Vision Mechanisms Effects’, Journal of the Optical Society of America, 7: 923–32. Martin, Michael G. F. (2002). ‘The Transparency of Experience’, Mind and Language, 17: 376–425. McDowell, John. (1982). ‘Criteria, Defeasibility, and Knowledge’, Proceedings of the British Academy, 68: 455–79. McGrath, Matthew. (2013). ‘Dogmatism, Underminers, and Skepticism’, Philosophy and Phenomenological Research, 86: 533–62. McLaughlin, Brian. (2007). ‘Type Materialism for Phenomenal Consciousness’, in Max Velmans and Susan Schneider (eds), Blackwell Companion to Consciousness. Oxford: Blackwell Publishing, 431–44. Millikan, Ruth G. (1984). Language, Thought and Other Biological Categories. Cambridge, MA: MIT Press. Millikan, Ruth G. (1989). ‘In Defense of Proper Functions’, Philosophy of Science, 56: 288–302. Neander, Karen. (1996). ‘Swampman Meets Swampcow’, Mind and Language, 11: 118–29. Neta, Ram. (2003). ‘Contextualism and the Problem of the External World’, Philosophy and Phenomenological Research, 66: 1–31. Neta, Ram. (2008). ‘What Evidence Do You Have?’, British Journal for the Philosophy of Science, 59: 89–119. Peacocke, Christopher. (1983). Sense and Content: Experience, Thought, and Their Relations. Oxford: Oxford University Press. Plantinga, Alvin. (1993). Warrant and Proper Function. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



SUSANNA SCHELLENBERG

Pollock, John. (1974). Experience and Justification. Princeton, NJ: Princeton University Press. Pollock, John and Joe Cruz. (2004). ‘The Chimerical Appeal of Epistemic Externalism’, in Richard Schantz, The Externalist Challenge. Berlin: de Gruyter, 125–42. Pryor, James. (2000). ‘The Skeptic and the Dogmatist’, Noûs, 34: 517–49. Pryor, James. (2001). ‘Highlights of Recent Epistemology’, British Journal for the Philosophy of Science, 52: 95–124. Pryor, James. (2012). ‘When Warrant Transmits’, in Annalisa Coliva (ed.), Mind, Meaning, and Knowledge: Themes from the Philosophy of Crispin Wright. Oxford: Oxford University Press, 269–303. Robinson, Howard. (1994). Perception. London: Routledge. Russell, Bertrand. (1984 [1913]). Theory of Knowledge. New York: Routledge. Sagi, Dov and Bela Julesz. (1985). ‘Detection versus Discrimination of Visual Orientation’, Perception, 14: 619–28. Schellenberg, Susanna. (2011). ‘Ontological Minimalism about Phenomenology’, Philosophy and Phenomenological Research, 83: 1–40. Schellenberg, Susanna. (2013). ‘Experience and Evidence’, Mind, 122: 699–747. Shoemaker, Sydney. (2007). ‘A Case for Qualia’, in Brian McLaughlin and Jonathan Cohen (eds), Contemporary Debates in Philosophy of Mind. London: Blackwell, 319–32. Silins, Nicholas. (2005). ‘Deception and Evidence’, Philosophical Perspective, 19: 375–404. Smith, A. D. (2002). The Problem of Perception. Oxford: Oxford University Press. Snowdon, Paul. (1981). ‘Perception, Vision, and Causation’, Proceedings of the Aristotelian Society, 81: 175–92. Snowdon, Paul. (1992). ‘How to Interpret Direct Perception’, in Tim Crane (ed.), The Contents of Experience. Cambridge: Cambridge University Press, 48–78. Sosa, Ernest. (1991). Knowledge in Perspective. Cambridge: Cambridge University Press. Sosa, Ernest. (2007). A Virtue Epistemology: Apt Belief and Reflective Knowledge. New York: Oxford University Press. To, M. P., I. D. Gilchrist, T. Troscianko, and D. J. Tolhurst. (2011). ‘Discrimination of Natural Scenes in Central and Peripheral Vision’, Vision Research, 51: 1686–98. Tucker, Chris. (2010). ‘Why Open-Minded People Should Endorse Dogmatism’, Philosophical Perspectives, 24: 529–45. Tye, Michael. (2002). ‘Representationalism and the Transparency of Experience’, Noûs, 36: 137–51. Watson, Andrew and John Robson. (1981). ‘Discrimination at Threshold: Labelled Detectors in Human Vision’, Vision Research, 21: 1115–22. Weatherson, Brian. (2005). ‘Scepticism, Rationalism and Empiricism’, in Tamar Gendler and John Hawthorne (eds), Oxford Studies in Epistemology (Vol. 1). Oxford: Oxford University Press, 311–31. Wedgwood, Ralph. (2002). ‘Internalism Explained’, Philosophy and Phenomenological Research, 65: 349–69. White, Roger. (2006). ‘Problems for Dogmatism’, Philosophical Studies, 131: 525–57. Williamson, Timothy. (2000). Knowledge and Its Limits. Oxford: Oxford University Press. Wright, Crispin. (2007). ‘The Perils of Dogmatism’, in Susana Nuccetelli and Gary Seay (eds), Themes from G.E. Moore: New Essays in Epistemology and Ethics. Oxford: Oxford University Press, 25–48. Zagzebski, Linda. (1996). Virtues of the Mind. Cambridge: Cambridge University Press.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PART II

Traditional Internalism and Inferentially Justified Belief

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

6 Principles of Inferential Justification Trent Dougherty

In this chapter I will consider two extant principles of inferential justification—those of Richard Fumerton and Michael Huemer—and propose my own as an improvement over them both. Unsurprisingly, this will occasion the necessity of clarifying all the key notions. I will propose a revisionary notion of ‘inferential’ belief that avoids what I take to be a common and detrimental mistake. I begin by making a few brief remarks on the target phenomenon and then look at the two most prominent principles in turn before offering my own. According to a perfectly sensible way of picturing our mental lives, we have these things called beliefs (which I confess to finding very hard to understand), and sometimes some of these beliefs are based on others. Some examples of things we might say when talking in this mode: B1:

“My belief that America is the greatest nation on Earth is based on my belief that freedom is the greatest civic good.”

B2:

“My belief that Jane will be at the party is based on my belief that Maria will be at the party.”

B3:

“My belief in evolution is based on my belief that reputable scientists are reliable.”

While this is a perfectly sensible way to talk, I will argue that a common way of thinking about such discourse misleads us (as surface grammar often does when we give ordinary language too much weight in theorizing). I will give deflated versions of ‘inference’ and ‘basing’ which allow for a simpler theory of inferential justification which covers the data. With regard to the data, I will point out that both Fumerton and Huemer underdescribe it. A richer picture of the data favors my more parsimonious view. A theory of knowledge provides error theories for both Fumerton’s and Huemer’s intuitions.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TRENT DOUGHERTY

1 Fumerton’s PIJ 1.1 Fumerton’s case for PIJ Consider the following scenario. Astrologer Suppose one ran into an astrologer who claimed to be justified in believing that there will be many wars in the year 2000 and who offered as his evidence that Jupiter will be aligned with Saturn that year (Fumerton 1985: 39). It’s a datum that something has gone wrong, but what? Let’s bracket concerns about neurologically related mental malfunctions. For though we may want to revisit the issue, it seems even if our friend is not subject to mental illness there is something wrong. Richard Fumerton gives the following diagnosis.1 Diagnosis F We might challenge his justification for believing that Jupiter will be aligned with Saturn in the year 2000, but we would be much more likely to focus on the connection between the position of planets relative to one another and the behavior of people here on earth. In short, we would challenge his claim to have a justified belief in the one proposition (P) on the basis of the other (E) by questioning his justification for believing that E makes probable, confirms, makes likely to be true P. (Fumerton 1985: 30–40)

He notes that one might think that the problem here stems from the fact that the astrologer isn’t justified in believing that she is justified in believing P. But even controlling for that, there still seems to be a problem with the basis (or basing) of the belief itself. Fumerton captures his diagnosis in the astrologist’s failing to meet the criteria set forth in the following Principle of Inferential Justification. PIJ: To be justified in believing P on the basis of E one must not only be (1) justified in believing E, but also (2) justified in believing that E makes probable P (1995: 36). Fumerton’s account is initially plausible and explains the data, for we hardly take it that the astrologist satisfies the second clause. So far, so good.

1.2 An alternative explanation In fact, though, Astrologer underdetermines this. After all, it’s not conceptually impossible that someone could be justified in believing in astrological theories. Consider the following two expansions on Astrologer. Astrologer A An astrologer, Annie, has a normal ‘Western’ education of mostly European history, biology, chemistry, physics, and so on. Annie has been given no information about correlations between the positions of planets and the 1

Neither here nor anywhere else does Fumerton seem to me to be using Mill’s method, as claimed by Tucker (2012: 325).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PRINCIPLES OF INFERENTIAL JUSTIFICATION



occurrence of war, yet Annie claims to be justified in believing that there will be many wars in the year 2000 and offers as her evidence that Jupiter will be aligned with Saturn that year. Astrologer B An astrologer, Bonnie, has a normal ‘Western’ education of mostly European history, biology, chemistry, physics, and so on. However, Bonnie has been raised by (otherwise good) parents who have presented information affirming detailed correlations between the positions of planets and the occurrence of war, with charts going from AD 1000 to 2500, so Bonnie claims to be justified in believing that there will be many wars in the year 2000 and offers as her evidence that Jupiter will be aligned with Saturn that year. This is an important part of her parents’ worldview, and Bonnie has been raised among people—doctors, lawyers, professors—who share this belief, and have convincing explanations for why standard astronomers fail to notice and report these correlations. Before assessing the impact of these two precisifications, let’s take a detour through (versions of) a couple of old chestnuts from the internalism/externalism wars. We’ll bring some intuitions pumped there to bear on the two cases just given. So consider the following two cases. Mr. Truetemp Mr. Truetemp gets intimations about the temperature in various parts of the world. It will suddenly seem to him that it must be 77 degrees in Austin, and he will form the corresponding belief. In fact, he is right 99 percent of the time, though he is not in possession of a track record argument demonstrating this. The Chicken Sexer The chicken sexer, Sally, can quickly separate male and female chicks just by picking them up. What happens is that the chicks have a distinct scent that differs by male and female, but Sally doesn’t know this. Rather, Sally just picks them up and looks at them for a second and they either simply seem male to her or seem female to her, and she forms the corresponding belief. She has no idea how it works. She is never told how well she’s doing. These two cases were designed to be counterexamples to reliabilist theories of justification (or knowledge, but knowledge, being a rather uninteresting property, will not be considered further). For in both cases beliefs are formed via reliable beliefforming processes, and yet, we are led to believe, the beliefs are not justified. Even though originally used for the purpose of making a case for internalism, the fact is the cases are quite underdescribed and have versions perfectly consistent with externalism. Consider the naïve use of sense perception by young children (but old enough to have acquired the concepts red and round). They have no idea how their perceptual faculties work, and though they are generally reliable, they are in possession of no track record argument in support of this fact. They just look at a reddish thing and it seems to them that it is red and look at a roundish thing and it seems to them that it is round.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TRENT DOUGHERTY

No internalist can reasonably deny that this is a case of justified belief, and it is structurally similar to the cases of Mr. Truetemp and Sally the Chicken Sexer, so what went wrong in the judgment that they are unjustified by internalist lights? It seems clear to me that what has gone wrong is that people naturally fill in the gaps in the stories with a tacit closure clause “and everything else they believe is like me.” That is, we project our background information onto the subjects in the thought experiment. This background information will include information that provides defeaters for the prima facie justification one has in cases where beliefs seem forced on one by experience (to borrow a phrase used from Locke to Russell). A Mr. Truetemp with our background information will have in his possession information that makes highly improbable that any human has such powers (whether they do nor not). A Sally the Chicken Sexer with our background information (prior to learning how the phenomenon works) that makes highly improbable that any human has such powers (whether they do or not). But if we imagine Truetemp and Sally without our background information projected on them, then their prima facie justification remains intact by internalist lights, and they serve their purpose very poorly. Now let’s take these insights and bring them to bear upon the astrologer cases. My position is that Bonnie is in a sufficiently similar position to Trutemp and Sally: the object of her belief seems perfectly reasonable to her, and she has no defeater for this prima facie justification. But, quite naturally, when we hear the bare bones base case Astrologer, our minds naturally fill out the story according to variation A rather than variation B, since the A version fits our everyday experience, whereas the B version is highly unusual and requires some creativity. This leads to a diagnosis of what is wrong in Astrologer that doesn’t require committing to Fumerton’s PIJ. Recall that his defense of PIJ appealed to his diagnosis. But my alternative diagnosis is better. I appeal to a requirement on justification—a no defeater clause—that almost all epistemologists agree on. And I note that this clause is not satisfied in version A, the version that fits our own background information. Given this alternative diagnosis, we have no need to endorse PIJ to explain what is wrong in Astrologer (note that I am not arguing that it is false). This leaves us without an answer, though, to the question as to what the core conditions on inferential knowledge are. We will now look at an alternative proposal by Michael Huemer.

2 Huemer’s PIJ* Huemer’s view is best understood by first understanding his critique of Fumerton. Understanding this critique will also lay the groundwork for important distinctions to follow.

2.1 Huemer’s critique of Fumerton Huemer essentially says that PIJ is incoherent. Taking his cue from his own understanding of Fumerton’s statement that the making probable relation is like what

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PRINCIPLES OF INFERENTIAL JUSTIFICATION



Kyburg calls “partial entailment,” where entailment is the limiting case of making probable, Huemer sees the two clauses of PIJ containing the seeds of PIJ’s destruction. To see why, and I’m surprised Huemer doesn’t do this, let’s put PIJ in a more standard form. Here is PIJ as Fumerton has it. PIJ: To be justified in believing P on the basis of E one must not only be (1) justified in believing E, but also (2) justified in believing2 that E makes probable P. This infinitive construction is not the standard form of a statement of necessary conditions. Here’s a more standard version. PIJ2: S is justified in believing P on the basis of E only if (i) E is S’s basis for believing P, (ii) S is justified in believing E, and (iii) S is justified in believing that E makes probable P. Note that (i) in PIJ2 makes explicit something implicit in PIJ. But if making probable is modeled on entailment, then, as I read Huemer’s critique (he is not explicit about this), Fumerton has S reasoning like this: 1. E. 2. E makes probable P. 3. Therefore, probably P. After all, if E entailed P, then S would reason like this: 1’. E. 2’. E entails P. 3’. Therefore, P. But if that is how S is reasoning, then E isn’t the basis for S’s belief that P at all, and (i) in PIJ2 isn’t even satisfied. Huemer does not see a way Fumerton can avoid seeing E makes probable P as a premise in an argument (Huemer 2002: 332–3). In Section 3, we will see that Fumerton does give us a hint as to how this can work, but he never really comes down to the point. I will make this perfectly explicit and, ironically, my proposal will be very much in line with Huemer’s own phenomenal conservatism.

2.2 Huemer’s account of the Astrologer case Remember, PIJ was offered as an account of what’s going wrong in Astrologer. Huemer has rejected Fumerton’s explanation, but he hasn’t yet given his own. Like Fumerton’s, Huemer’s principle of inferential justification will arise out of his assessment of Astrologer. 2 Note that to be justified in believing does not imply having formed the belief in question. Thus I think Tucker speaks falsely when he says that Fumerton endorses a principle that includes that one actually has the belief (Tucker 2012: 324). At any rate, Fumerton clearly isn’t committed to that by his general approach and says many things that indicate he does not endorse that requirement.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TRENT DOUGHERTY

Huemer’s idea can be put in the form of a dilemma. Let us recall what the astrologer (let’s just name her Ashley) believes and what she cites as a basis. She believes Y2K:

There will be many wars in the year 2000.

And she cites as her basis JAM:

Jupiter will align with Mars that year.

The dilemma is this: either Ashley infers Y2K directly from JAM or she does not. If she does not, but, rather, is using the suppressed lemma, then what we are seeing is wrong in that Ashley is not justified in believing JAM makes probable Y2K. This is similar to my proposal, but whereas Huemer is merely imagining us thinking of the lemma as unsupported by our evidence, I am imagining us seeing the conflict with evidence we are aware we have. If she does infer Y2K directly from JAM, then, Huemer says, in that case the belief (that Y2K) is unjustified “Simply because the argument from [JAM to Y2K] is a non sequitur” (2002: 333). This seems wrong-headed for an internalist. For an internalist, justification is supposed to be the kind of thing we can share with our unfortunately envatted counterparts or with those under the influence of Descartes’ demon. But, as Descartes explicitly notes, deception can happen with a priori intuitions just as easily as with perceptual experiences. One could find it obvious that 1 + 2 = 7, for example. But that sense of obviousness is the very thing that justifies our belief that 1 + 2 = 3. So it is very strange for Huemer to take this externalist hard line. And, of course, Huemer himself endorses an internalist principle of evidence generation: (PC) It seems to S as if P, then S thereby has at least prima facie justification for believing that P (Huemer 2001: 99). And, clearly enough, it could also seem to one that JAM makes probable Y2K. And if it did, then, by golly, according to Huemer’s principle, they would be justified in believing it. I will return to this theme below, because the same radical externalism shows up in Huemer’s principle, and my own proposal makes use of a principle of phenomenal conservatism similar to Huemer’s PC.

2.3 Huemer’s own account flawed As we have seen, Huemer thinks that for a basis E to justify some proposition, P, based on it, E needs to actually make probable P. Thus, he modifies the second clause of PIJ to reflect this externalism. (PIJ*) S is justified in believing P on the basis of E only if (1) S is justified in believing E and (2) E confirms P. Note this is both stronger and weaker than PIJ in certain respects. It adds objective confirmation to the requirement for justified belief, but it removes the necessity of

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PRINCIPLES OF INFERENTIAL JUSTIFICATION



having an additional belief be justified. Technically, of course, removing the higherorder requirement from the second clause of PIJ is consistent with Huemer thinking there is such a necessary condition on justified belief in addition to (1) and (2). However, removing it when it was there to begin with seems to convey that he thinks there should be no such requirement on justification. That would, I think, cause additional problems, but I will focus on the problem generated by requiring objective confirmation for justified belief. Pursuant to that, consider the following case. The Forgetful Logic Student An undergraduate Anthropology major, Andy, is writing the first paper of his second philosophy class. He has been asked to give his argument in standard form. Andy took the prerequisite for the class, Logic 101, three semesters ago, and his memory is hazy. One vivid image he has is from the time when the professor, using bad judgment, discussed on the very first day how things that were fallacies in one branch of logic (‘deductive,’ but Andy can’t remember the term) are not fallacies in some other branch of logic (‘inductive,’ but Andy can’t remember the term). He has a vivid memory of the professor saying at the end of the lecture that an inference called ‘affirming the consequent’ was a good move in science and that it was used to confirm Newton’s physics. (Andy doesn’t recall where it was not supposed to work). Since Andy is writing his paper about his own major, Anthropology, he figures affirming the consequent is the way to go, so he gives the following argument. Andy’s Argument 1. If members of the Kankapachinkineoatwan tribe are polygamous, then polygamy is morally accepted in some cultures. 2. Polygamy is morally accepted in some cultures. 3. Therefore, members of the Kankapachinkineoatwan tribe are polygamous.

Let E be the conjunction of the corresponding conditional of Andy’s Argument and the conjunction of the premises 1 and 2 i.e. ((1&2) & ((1&2) ! 3)).3 And let E be the sole basis of Andy’s belief that members of the Kankapachinkineoatwan tribe are polygamous (which, as it turns out, is false, as well as unsupported by any other information Andy has). Andy, it seems clear to me, is justified in his belief. Yet, clearly enough, E does not confirm P (the conclusion of Andy’s Argument). But PIJ* (2) requires that E actually confirm P for S to be justified in believing P on the basis of E. So PIJ*(2) gives the wrong result. So PIJ*(2) is false.4 Based on his treatment of a relevantly similar case (Huemer 2002: 336), it seems clear that Huemer would agree that Andy is justified in believing P. However, he would say this is not because E justifies his belief in P, but, rather because he’s justified in believing AB: 3

I, Andy, have just proven P.

I include the corresponding conditional just to make it clear he has justification. Insisting on this seems to play into Carroll’s Paradox. I am fine saying (1&2) is his sole basis ‘in light of ’ his bearing some appropriate relation to ((1&2) ! 3). 4 The same problem befalls Tucker’s principle of inferential justification in Tucker (2012), which explicitly requires that E actually support P.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TRENT DOUGHERTY

There are a number of problems with this reply. First notice that we can stipulate that in the case Andy never actually forms the belief that AB (maybe he doesn’t even have the concept proof ), and it’s unclear that merely being justified in the belief is good enough in this kind of case. Furthermore, it’s not clear that Andy has or needs evidence strong enough for him to rightfully conclude that AB is true. It might be that Andy lacks a defeater for AB, or that AB is on balance more likely than not for Andy. There is not room to pursue this further here.

3 Chisholming Away 3.1 A sketch of a theory of knowledge As implausible as it might sound, I think it will help if I briefly sketch a theory of knowledge. I’m far from sure that there is any one, interesting, homogenous thing answering to all the things we would typically classify as knowledge. And I don’t think the concept knowledge is a good cornerstone notion for explaining other things in philosophy (such as assertion or practical reasoning or evidence). However, for understandable evolutionary reasons the concept knowledge is a useful shorthand in everyday life. Here’s one reason I think it will help for me to sketch my theory of knowledge. Externalists have also considered PIJ (Greco 1999, 2000; Bergmann 2006) and some internalists (Huemer 2002; Tucker 2012) are fans of objective support relations. I think there is a persistent tendency for externalists—who tend to be more interested in knowledge—to foist externalism on justification and for some internalists to concede too much to them. The key is getting the objective and subjective stuff in the right places. I will present a theory of justification and knowledge that gives the objective and subjective appropriate roles. The material immediately following in 3.1.1 and 3.1.2 is crucial to my account of inferential justification anyway, and, that having been covered, a few paragraphs will suffice to tie it into a theory of knowledge.

..

PERSPECTIVAL JUSTIFICATION First, justification is wholly subjective. ‘Common sense’ epistemology is often presented as the view that we really do know the kinds of things we typically take ourselves to know: our perceptual, memorial, introspective, and math/logic-y beliefs typically constitute cases of knowledge.5 But for each of these traditional sources of knowledge, there is some experience that acts as the conduit of the knowledge. That is, in each case, the belief said to constitute knowledge is a response to some kind of experience, whether a perceptual experience, apparent memory, rational intuition, or what have you. Whether beliefs about the

5

Bergmann (2012), for example, characterizes it this way.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PRINCIPLES OF INFERENTIAL JUSTIFICATION



future or even about the more than very recent past constitute knowledge is not something I’m willing to go to the mat for. But I am willing to say they are reasonable beliefs, that one has good reasons for them. So I’m willing to endorse the following principles. PERC: If it seems to S that S perceives that p, then S is prima facie justified in believing p. MEM: If it seems to S that S remembers that p, then S is prima facie justified in believing that p. UND: If it seems to S that some modal claim Mp is true, then S is prima facie justified in believing that Mp.6 It is hard to think how a parallel principle, INT, might be stated to cover introspection, but it’s clear there is a corresponding faculty. And different folks may want to divide up faculties differently, so it seems best to simply quantify over faculties and endorse the following. (PC)7 If it seems to S as if P, then S thereby has at least prima facie justification for believing that P.8 One of the virtues of this kind of principle is that it is about the only kind of reasonable principle which doesn’t exclude beliefs from justification just because they arise from dysfunctional faculties. Consider the following case. Red/Blue Due to a burst of gamma rays from Alpha Centauri, Charlie is subject to Red-Blue color inversion. Charlie, however, is none the wiser, with no noticeable effects of this malady. Shortly after the burst, Charlie walks into her kitchen where she keeps her nice collection of red and blue mugs (six of each). She fixes her gaze on a red mug, which now looks blue. The scene is perfectly familiar: a randomly checkered pattern of red and blue mugs.

If Charlie didn’t come to believe there was a blue mug in front of her, she would be dysfunctional. When it looks like there’s a blue mug right in front of you, and there is 6 ‘UND’ stands for the faculty of understanding, as Descartes used the term. ‘M’ is to range over standard alethic necessity and possibility operators. So both the diamond and the box can replace ‘M.’ It is unfortunate that ‘Mp’ in old logic textbooks is used to convey ‘Possibly, p.’ The use of this device is the best I could come up with for indicating that UND is about rational intuition, about a priori matters, as I take it all modal claims are and no non-modal claims are except in very special cases. The reader with a modicum of charity will take my meaning perfectly well. Paradigm cases of the application of UND will involve theorems and conceptual claims. One might think many of the latter are synthetic a priori, which might throw a spanner into the engine, but still, I take it that what I’m getting at here is sufficiently clear. 7 I use ‘PC’ first and foremost because it stands for the Principle of Credulity which has a long and illustrious career. It is convenient that it also stands for Huemer’s version of it which, though it needs some refining, I am happy to use here for present purposes. 8 I am inclined to require seemings as a necessary condition on justification as well, but there are many issues involved with that. One concerns whether non-occurrent mental states can count as evidence. I am inclined to say No to solve certain puzzles. For more, see Feldman (2004).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TRENT DOUGHERTY

no inkling that anything is amiss, it is perverse not to believe there’s a blue mug. Something has already gone wrong to make a red mug look blue. But given that it looks blue, it would be a further problem not to believe that it looks blue. This is my fixed point.

..

A NOTE ON SEEMINGS AND THE SUBCONSCIOUS Now a quick distinction, and then the theory of knowledge. Evidence, on the phenomenal conception, consists in experiences (see Dougherty and Rysiew 2013 for an extended defense). But on the view I’m defending, it consists in a certain species of experience. Consider a pencil in a glass of water. We naturally say it looks bent. But we aren’t even tempted to think it is bent. And note that, in this case, it’s looking bent is no significant evidence that it is bent. I call this the difference between seeming as though it is bent (the visual imagery), and its seeming that it is bent (the epistemic force the imagery has on us). I think Bergmann (2006) shows that it can’t be the sensory perceptual states, the phenomenal qualia of colors and such in a kind of geometric grid that are the evidence. For the kind of supervenience on the evidence an internalist wants, the connections between evidence and what they support must hold of necessity. And there is no necessary connection between, say, hosting phenomenal red and there being a red thing. In Martians or Alpha Centaurians (Alpha Centaurs?) hosting phenomenal red might be evidence that something is hot.9 I take it that when one’s (adult, the child case is hazy) noetic system is functioning properly, something like the following happens. We acquire concepts the application of which is triggered by sensuous experiences. Applying a concept is not something we do, it is something that happens in us below the level of consciousness, in the subconscious. I’ll go into more detail in section 3.2, but clearly a lot of unconscious ‘inference’ or calculation or computation happens in us subconsciously. For example, suppose you come home and find grocery bags on the floor with various items in them as well as new jars of Peter Pan peanut butter (I’m sure you’re not one of those Jiff people) and such on the counter. You immediately (at the conscious level) think “Oh good, someone’s been to the grocery store.” Now, in most such cases (very nearly all, I’d wager), you simply never consciously host a thought about alternative explanations or principles of induction or past experience or any of the myriad factors your blessed brain can keep track of for you subconsciously. Rather, your brain figures all this out for you, processing it subconsciously for the sake of efficiency, and just provides you with the appropriate belief via making the proposition that it is so ‘light up’ in the right kind of way. Sometimes, due to cognitive malfunction, a proposition might light up and yet you fail to form the belief. Naturally, such cases are rare.

9

Tucker (2011) makes a similar distinction.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PRINCIPLES OF INFERENTIAL JUSTIFICATION

..



KNOWLEDGE With the groundwork laid in sections 3.1.1 and 3.1.2, it is easy to set forth the theory of knowledge quite briefly. The basic idea is that knowledge occurs when our internal markers provide the right kind of causal bridge (or, perhaps more weakly, the right kind of chain of dependency) between the world and our doxastic system. As noted in section 3.1.1, internalism is partly defined in terms of the distinction between the ‘good case’ where the world is roughly how we take it to be and the ‘bad case’ of being envatted or in a demon world or the Matrix or dreaming or in Tron’s digital world or somesuch (local cases like Fake Barn County do not strike me as bad cases in the relevant sense: when you look at a barn and it looks like a barn and you believe it is a barn because it looks like a barn and you have no evidence that it’s not a barn, then you know it’s a barn). Internalists hold a perspectival theory of evidence which means that ‘what we have to go on’ in the good case and the bad case is the same (some kind of narrow mental content). However, and this is too little noticed, in the good case things are going to work just like externalists want them to work. Reacting naturally to the prompts of our internal markers will reliably lead to true beliefs via non-deviant causal chains (or chains of dependence). A well-functioning individual who is appeared to redly will form the belief there’s a red thing, and so on. But, as noted in section 3.1.2, it’s not, say, hosting phenomenal red that constitutes the evidence. It is, rather, the seeming that there is a red thing: one’s hosting the proposition that there is a red thing with the right affirmative phenomenology. The chain goes like this. There is the thing in the world with its intrinsic properties, the red mug, say. The red mug will reflect only those electromagnetic waves in the 620–740 nm range. Next (I am leaving out many steps, like the formation of the inverted retinal image, the transduction of the encoded image via nerve impulses (electrical signals), and more), there is the neural processing of this into a unified visual image in consciousness. This image all by itself could be associated with any state of affairs. It could be objectively correlated with anything you please. Martians could host phenomenal red when an object is hot, regardless of the structure of its surface. So, finally (and here I depart markedly from Huemer 2001: 71ff ), with concepts being applied we get a mental state with propositional content. This is the transition from what I have called ‘seeming as though’ (mere sensory qualities) to ‘seeming that’ (the proposition ‘lighting up’ for us). This latter state is the evidence, and a properly functioning individual will respond to this evidence with the correct affirmative kind of doxastic attitude. That is, the subconscious processing will physically causally generate something that occupies the ‘space of reasons’ (the seeming that) and a new kind of causation takes over in the reasons— responsive individual and the appearance of evidence for p in the space of reasons for S will result in S’s pro-attitude toward p. What I’ve tried to emphasize here is the interaction of the conscious and subconscious in justified belief and knowledge. We are rational animals in the world. As animals in the world, there will be unconscious and unobservable processes involved in our knowing. As rational animals, our

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TRENT DOUGHERTY

knowing will involve some reference to the space of reasons, even in the simplest cases. I say this because it even seems true in cases of simple perception. I don’t consider it a literal use of ‘knowledge’ when that term is ascribed to information we react to without any thought at all, such as avoiding cracks in a sidewalk. As a fallback position, I would countenance a distinction between two kinds of knowledge. This is a causal (or dependence) theory of knowledge. I am happy to put it in terms of explanation as well. If the fact that P is the dominant explanation of why it appears to me that P, and its appearing to me that P is the dominant explanation of why I believe that P, then I know that P. The second link is sufficient for epistemic justification, and the first link hooks up our psychology in the right kind of way to make it knowledge (the causal link between the fact that P and its appearing that P rules out most (maybe all) Gettier cases, but it also includes a ‘no essential dependence on a falsehood’ being built into a non-deviance clause on the causal requirement which will rule out most if not all Gettier cases as well).

3.2 On ‘inference’ We are almost in a position to state a principle of inferential justification which avoids the problems of both Fumerton’s and Huemer’s (and others extant in the literature). Since our target is inferential justification (and, by implication, inferential knowledge it gets us), the last piece of the puzzle is an account of the inferential. My view of the nature of inference is controversial, but, I think, necessary for a number of reasons. I will say a few words to motivate the view. Think of how we use the verb ‘catch’ and its cognates in phrases like “I caught a cold.” Catching a cold is very different from catching a train. To catch a train there’s something one does; there is an action one performs. This is not so with catching a cold. To be sure, there are actions one can perform that make it more or less likely that one will or will not catch a cold. But a cold is just something that happens to you: the faux verbal use of ‘catch’ is misleading. Something similar happens in the epistemic realm. For think how we use the word ‘believe.’ The faux verbal usage pretends that there is something one does when one believes. To be sure, there are actions one could take to bring it about that it is more or less likely that one would or would not come to hold some belief. Sometimes this is quite easily done as with the old example of turning on a light to bring it about that you believe the light is on. Sometimes it is much harder to bring about a belief. If you offered me a million dollars to believe that the moon is made of green cheese, I can’t just directly believe it, as if I were simply flexing a mental muscle. But I could try to brainwash myself or pay someone to hypnotize me or something, and that might increase the probability that I would come to believe it. Nevertheless, believing is not something one does, not an action one performs. Believing is something that happens to you. In fact, in this regard belief has been explicitly compared to catching a cold (I forget by whom).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PRINCIPLES OF INFERENTIAL JUSTIFICATION



It will be well to illustrate how this works in three kinds of common cases both a posteriori and a priori. Consider first the case of perception. You are riding on the train from St. Andrews to Edinburgh and a child asks, “Excuse me, is that a deer?” pointing out the window. You look out and have a simple, vivid sense impression associated with seeing a deer. If you are functioning properly, you’ll immediately come to believe there is a deer (and hopefully don’t forget to tell the poor kid). The simple, vivid sense impression will have ‘forced’ you to have the belief (I’m inclined to think the scare quotes unnecessary). Consider second the question whether some wff is a theorem, say, the corresponding conditional to constructive dilemma. Here again, there is, in some sense, something you ‘do’ corresponding to looking out the window: You ‘look inside’ to consider whether the wff seems valid. If you have a moderate amount of logical acumen, you will have a simple, vivid insight in which you ‘see’ that it is valid. It will just seem clearly true, and as a result, you will be convinced that it is true. ‘Convinced’ is a cool word. The ‘vinc’ in ‘convinced’ is the same root as in ‘invincible’ which means ‘unconquerable.’ Thus to be convinced, etymologically, is to be conquered or overcome by the evidence. Finally, consider memory. We have the verb ‘to remember’ as though there is something we’re doing, but that’s as misleading as ‘catch’ and ‘believe.’ Again, as with looking out the window and looking ‘inside’ there is something we do to initiate the process which will terminate in a propositional attitude being ‘sent up’ from below. We can call this ‘looking back.’ It is a species of introspection, and it is the part we have the most direct control over in the sense that we can almost always bring it about directly that we have a memory experience. But this must not be confused with our having any freedom with respect to what seems to us to have happened or (if we are functioning properly) with respect to what we believe in light of what seems true to us. That is, if we are functioning properly, then our memory will just ‘tell’ us what happened, and, having been thus told, we will immediately believe (though of course there are ‘hazy’ memories which will cause weaker doxastic attitudes than belief). In each of these three common kinds of justified belief and knowledge the computational stuff is unconscious. The only act of will is a kind of looking, and the result is something that shows up in the space of reasons to guide assent. I hasten to add that I think people use the word ‘believe’ rather loosely, and I certainly don’t have in mind here the mere storage of information. What some people call beliefs I think are mere dispositions to believe. I am not friendly to the category ‘dispositional belief.’ ‘Infer’ is just like ‘believe’ and ‘understand’ and ‘remember’ (‘judge’ works on this model as well) in relevant respects. In fact, my a priori case of belief above just is a case of the only sensible thing we could call ‘inference.’ What appear to be multi-step ‘inferences’ are simply cases where we see that some conditional is true that connects lines n and n+1, and, at the end, we remember or seem to remember that we had such an experience for each pair in the sequence constituting the proof. But neither at any individual step nor during the interval was there something we were

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TRENT DOUGHERTY

actively doing other than ‘looking’ in various respects, sending instructions, as it were, to the sub-conscious homunculus to look some stuff up in his or her logic textbook or log book of experiences and then send up to the conscious mind a proposition ‘lit up’ in the right kind of way to alert us to their (probable) truth.

3.3 ‘Inferential’ justification in light of the above In their own ways, Fumerton and Huemer go wrong in different directions, but each on the basis of a good impulse to do due diligence to one or another aspect of human cognition. Fumerton wants to be sure to include reference to the space of reasons, since he realizes that human cognition is not limited to unconscious computation and that rationality is perspectival. He is right about two things: it is not uncommon, in general, and quite common in cases that attract our attention as reflective beings, to have such higher-order perspectives.10 He is also right that we need something in the space of reasons that illuminates the target proposition. There is no epistemic justification for beliefs that just pop up unconsciously via some reliable process without any evidence (the fact that we hold the belief can in some cases be sufficient evidence if we happen to reflect on it aright). He thinks that the conjunction of support for E and for E’s supporting P is the way to light up P in the space of reasons. I have argued for the much simpler and less problematic requirement that it is P’s seeming to S to be true that gives P the appropriate standing in the space of reasons. Huemer, on the other hand, is eager to emphasize not the perspectival nature of support (not inferential support, anyway) but the part that doesn’t depend upon our perspective: objective confirmation. But problems with both Fumerton’s PIJ and Huemer’s PIJ* are due, it seems to me, to being deceived by the grammar of ‘infers’ into thinking that inference is some kind of act, something we do, as if ‘basing’ is something we decide to do. Also, both principles use the evidential variable ‘E’ to name a proposition and both model epistemic support on logical support. But as we’ve seen in the logic student cases, these two notions are quite distinct. Logical support is neither necessary nor sufficient for epistemic support. This, as Cohen (1984) points out, is one of the key insights of Chisholm.11

10

This is a weaker claim than the error theory (for the view that we must always have some kind of higher-order requirement on justification) offered by Bergmann according to which this is ‘often’ the case (2005: 419), though he later reduces it to ‘some’ (2005: 431). It seems to me that ‘often’ is untrue and ‘some’ is too weak for a successful error theory. Thus I propose that though it is not often in general it is often in cases that an availability heuristic calls to the minds of philosophers, an especially reflective lot. 11 Thus I move in the exact opposite direction of Fumerton (2004), which includes his own reply to Huemer, when he says that we are better off the closer our view of epistemic support is to entailment. Yet I do here meet Fumerton’s stated desideratum there: “we must find some proposition of the form E makes probable P that we can justifiably believe without inference” (2004: 162). On my view, any proposition of that form can be epistemically justified, and ones which have the right etiology can have doxastic justification and become candidates for knowledge. There is no room to discuss Fumerton (2004) more, for it is a long and rich paper. However, I think that much of what I say here provides the seeds for understanding how turning back to Chisholm can solve the problems Fumerton sets up there.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PRINCIPLES OF INFERENTIAL JUSTIFICATION



So instead of having ‘E’ refer to a proposition, the relatum of a logical support relation or a belief, which some take to be capable of being a relatum, I will have ‘E’ refer to an experience. And instead of building on a logical relation, I will build on the kind of (synthetic a priori? conceptual?) support relation Chisholm had at the base of his system (he called them “material epistemic principles”). This will, per my above remarks about inference not being a thing we do, change the locus of ‘inference’ from an action-like thing to a pattern or structure involved in the belief-forming process. I will make further explanatory remarks after I give the account. Finally, both seem to neglect the division of labor between the conscious and subconscious and to lack a principled connection between the objective and subjective aspects of cognition as it relates to the distinction between justification and knowledge. I propose the following in the context of the above, to characterize inferential justification. PIJ$: P is inferentially justified for S on the basis of experience E iff (i) S has E, (ii) E is its seeming to S that P, (iii) there is an inferential pattern T tokened in some subconscious computational process C in S such that (a) C maps S’s accessible, relevant, stored information I, onto T, and (b) P (the proposition, not the belief that has it as its content) is the ‘output’ of C, (iv) the fact12 that (ii) holds is non-deviantly caused by the fact that (iii) holds. PIJ$ (pronounced ‘Pidge money’) is a bit baroque (in the pejorative sense, though I adore baroque art). But note that (i), and (ii) just say that, according to PC, S is justified in believing P. Clause (iii) adds what is necessary to say that the justification is ‘inferential’ in the sense sketched above, the only sense I think makes any sense. Sometimes S will be able to ‘see’ the process and other times it will be unconscious. When S is functioning properly, P will be a pattern of reasoning that is at least pretty reliable.13,14 P might instance rules of inductive reasoning, laws of logic and

I use here the notion that facts are the causal relata only as a useful fiction. My approach, it is now obvious, is completely the opposite of that of Rhoda (2008). He says that criticisms of inferential internalism depend on what he considers an externalist conception of inference and that salvation for internalism comes from, naturally, wedding it to what he calls an internalist conception of inference, where the difference is that the latter involves a conscious perspective on the inference and the former does not. I say basically the opposite: the right way to think about internalism is to adopt what Rhoda calls an “externalist” conception of inference. 14 In effect, though PIJ$ provides an account of inferential justification as a species, I’ve simply done away with any higher-order requirement on any kind of justification. Thus I am what Fumerton calls an “inferential externalist” with respect to justification. It is common, though, I think, for ‘hard-core’ internalists to fixate on knowledge that one knows one has, which has more stringent requirements. Higher-order requirements are plausibly appropriate for higher-order knowledge. This fleshes out my error theory. Further fleshing out could be had by meditating on the point raised by Feldman (2005) that the presence of higher-order information tends to defeat and combining this with the fact that reflective individuals are often in a position of higher-order conflict about whether their evidence supports their beliefs. 12 13

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TRENT DOUGHERTY

probability, or simple heuristics that are less universal. When S is not functioning properly, P might be anything. If S actually forms the belief that P, and this fact is non-deviantly caused by the fact that it seems to her that P, then, so long as there is no defeater for this appearance, S will know that P as long as P is true. The difference, on the revisionary model presented here, between inferential and non-inferential reasoning is the presence of steps between the categorization of the sense experience and the production of a propositional seeming state. In the case of immediate justification, a sense experience of a red apple is categorized as a red apple (so long as the defeater screening mechanism doesn’t intervene) and the categorization is sufficient to trigger the seeming state. Clearly, on this view, the difference between inferential and non-inferential justification is pretty trivial as compared to the traditional inference-as-action account that I find utterly misleading. And that is just as it should be when we’ve reflected on the fact that inference isn’t something we do, it’s something that happens in us that we sometimes ‘watch’ (are aware of the steps) and ‘oversee’ (send instructions down for processing to occur). So when, for example, we believe Q because it seems to us that P and that P entails Q (or even on occasion because we host that modestly complex state wherein it seems to us both that (P & (P ! Q)) and that the latter is a good reason to believe the former) we have a special kind of inferential justification. But this ‘peek’ at the computational undergirding is strictly optional. Because we have awareness in the space of reasons we can have these peeks, unlike less rational animals. But they are not part of what justifies the belief that Q. That status is secured by its seeming that Q. Thus, the specter of Carroll’s Paradox dissolves. It is convenient to have these peeks at the process, for it might make us more able to defend our belief that Q. But showing that Q and knowing that Q are two different processes, and the former is not a condition on the latter. Having evidence does not require being able to give an argument.

4 Conclusion I have given a deflationary account of the difference between inferential and noninferential justification, which I have set within an account of inferential justification and knowledge that shows how to get the objective and subjective aspects of the doxastic and epistemic picture in perspective. It does the same for the conscious and unconscious aspects of that picture.

References Bergmann, Michael. (2005). “Defeaters and Higher-Level Requirements,” Philosophical Quarterly, 55: 419–36. Bergmann, Michael. (2006). Justification and Awareness: A Defense of Epistemic Externalism. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

PRINCIPLES OF INFERENTIAL JUSTIFICATION



Bergmann, Michael. (2012). “Commonsense Skeptical Theism,” in Kelly Clark and Michael Rea (eds), Science, Religion, and Metaphysics: New Essays on the Philosophy of Alvin Plantingae. Oxford: Oxford University Press, 9–30. Cohen, Stewart. (1984). “Justification and Truth,” Philosophical Studies, 46, 3: 279–95. Dougherty, Trent and Patrick Rysiew. (2013). “Experience First,” in Matthias Steup, John Turri, and Ernest Sosa (eds), Contemporary Debates in Epistemology. Malden, MA: Blackwell, 10–16. Feldman, Richard. (2004). “The Justification of Introspective Beliefs,” in Earl Conee and Richard Feldman (eds), Evidentialism. Oxford: Oxford University Press, 199–218. Feldman, Richard. (2005). “Respecting the Evidence,” Philosophical Perspectives, 19: 95–119. Fumerton, Richard. (1985). Metaphysical and Epistemological Problems of Perception. Lincoln: University of Nebraska Press. Fumerton, Richard. (1995). Metaepistemology and Skepticism. Lanham, MD: Rowman and Littlefield. Fumerton, Richard. (2004). “Epistemic Probability,” Philosophical Issues, 14: 149–64. Greco, John. (1999). “Agent Reliabilism,” Philosophical Perspectives, 13: 273–96. Greco, John. (2000). Putting Skeptics in Their Place: The Nature of Skeptical Arguments and Their Role in Philosophical Inquiry. Cambridge: Cambridge University Press. Huemer, Michael. (2001). Skepticism and the Veil of Perception. Lanham, MD: Rowman and Littlefield. Huemer, Michael. (2002). “Fumerton’s Principle of Inferential Justification,” Journal of Philosophical Research, 27: 329–40. Rhoda, Alan R. (2008). “Fumerton’s Principle of Inferential Justification, Skepticism, and the Nature of Inference,” Journal of Philosophical Research, 33: 215–34. Tucker, Chris. (2011). “Phenomenal Conservatism and Evidentialism in Religious Epistemology,” in Kelly James Clark and Raymond J. Van Arragon, Evidence and Religious Belief. Oxford: Oxford University Press, 52–73. Tucker, Chris. (2012). “Movin’ on Up: Higher-Level Requirements and Inferential Justification,” Philosophical Studies, 157: 323–40.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

7 Inferential Appearances Michael Huemer

1 Justificatory Dependence Sometimes, a person’s belief depends for its justification upon other beliefs held by that person. Here are three sorts of cases: Fully explicit inference: S consciously infers P from E alone, and no (non-logical) background beliefs are required to explain the justification for the inference. For example, I believe that there cannot be both a tiger and a cow on my couch (due in part to the size of the couch). I consciously, explicitly infer from this that either there is no tiger on the couch, or there is no cow on the couch (inclusive ‘or’). Here, background beliefs might of course be required to explain my ability to understand the premise and the conclusion of the inference, and to explain why I believe the premise. But no background beliefs are required to explain why, given that I understand these propositions and that I believe the premise, it makes sense for me to accept the conclusion. Inference against a background: S consciously infers P from E, where some background beliefs are required to explain why the inference was made and why it makes sense.1 Suppose I return home after a day at the office. As I enter, I see that the fabric on the corner of the couch is shredded. I recall that the cat was home all day. I infer: the cat shredded the fabric. The only conscious, occurrent beliefs of mine leading to the conclusion might be [The cat was here all day] and [The fabric around there is shredded]. But a host of background beliefs on my part are involved in the full explanation for why it makes sense for me to infer, and indeed why I do infer, the conclusion—for instance, that the cat has claws, that claws are sharp, that fabric typically does not spontaneously become shredded, that the cat was in the room before I entered, that no other animal was in the room, and so on. Unlike the explicit premises of the inference, these background beliefs need not be, and typically are not, occurrent at the time of the inference.

1

For discussion, see Rhoda (2008).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

INFERENTIAL APPEARANCES



Non-inferential dependence: S’s belief that P depends for its justification upon S’s having some background beliefs that support P, but S did not actually arrive at P via inference. For example, I believe that I live in the United States.2 I would not say that I believe this because I inferred it from something. But of course, I could not be justified in believing that I live in the United States if I did not have some other beliefs that bear on that proposition. For instance, I believe that I live in Colorado, that Colorado is in the United States, that I lived in the United States as a child, that I would notice if I moved out of the country, that I would have learned of it if the U.S. had been dissolved, and so on. My central question in this chapter concerns the first sort of case, the case of fully explicit inference. Hereafter, up until Section 6, when I speak of inference and inferential justification, I shall have in mind fully explicit inference and the kind of justification provided by such explicit inference. Here is my question: when does a belief have (explicit) inferential justification? That is, what is the supervenience base for explicit, inferential, doxastic justification? In Section 6, I shall return to the other two forms of justificatory dependence. For simplicity, assume that S has a belief that P, based upon a single premise, E. I assume that for S to have explicit inferential (doxastic) justification for P, S must believe P, S must believe E, and S’s belief that E must somehow (perhaps along with other conditions) cause S’s belief that P. But these conditions are not enough. What more is required?

2 Conditions for an Adequate Theory of Inferential Justification Following are six observations that I think an adequate account must respect. i)

S must have some attitude toward [E supports P].3

Suppose S believes E. E in fact supports P, but it supports P in a very complicated way, and S just cannot see the connection between E and P. Nevertheless, S decides to base a belief that P on her belief in E anyway. Intuitively, S does not then have a justified belief in P. In fact, I may have either misdescribed the case or described an impossible case. If S’s belief in E causes her to believe P, but S doesn’t see any connection (nor seem to see any connection, nor take herself to see any connection) between E and P, then could it really be true that S believes P on the basis of E? It is plausible to hold that 2

This example is from Leite (2008: 434–5). I use square brackets to convert an independent clause into a term denoting a proposition. Thus, ‘[E supports P]’ denotes the proposition that E supports P. Condition (i) is endorsed by many epistemologists, including Fumerton (1995: 36, 2006: ch. 6), Hookway (2000), Rhoda (2008), Leite (2008), Tragesser (2009), and Tucker (2012). 3

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



MICHAEL HUEMER

S not only lacks inferential justification in such a case; S lacks an inferential belief (cf. Fumerton 2006: 105). I express this condition on inferential justification by saying that S must have an attitude toward the proposition that E supports P. For now, I leave it open exactly what that attitude should be. Maybe S should believe that E supports P, or be disposed to believe this, or perceive E as supporting P, or take herself to see that E supports P, etc. I also leave open for now the interpretation of ‘support’; maybe S should have an attitude toward [if E then P], or [E raises the probability of P], or [P is true because E], or something else in this neighborhood. ii)

Justified belief is insufficient.

What more can be said about the attitude one must take toward [E supports P]? We can quickly see that belief (even justified belief) is insufficient. Suppose again that S believes E, and that E in fact supports P in a complicated way, but S can’t herself see how E supports P. Nevertheless, S justifiedly believes that E supports P because her epistemology professor told her that E supports P. When S is then caused to believe P (partly) by her belief in E, she will not thereby be inferentially justified in believing P on the basis of E. In fact, we should say that S does not base her belief in P on E; rather, she bases her belief in P on E and [E supports P]. Another example: ten years ago, I figured out that E supports P. I don’t remember how I figured that out and I can no longer see how the two propositions are connected; all I remember is that E somehow supports P. Now, I suddenly discover that E is true. Remembering that E supports P, I conclude that P is also true. Again, in this case I do not infer P from E simply; I infer P from E and the claim that E supports P. iii)

[E supports P] is not a premise.

The preceding examples carry a related lesson. This is that one must avoid treating [E supports P] as another premise of S’s inference to P. In the examples, [E supports P] really is a premise of the inference to P. That is different from the case where one infers P from E simply. For the case in which S is justified in believing P (solely) on the basis of E, our theory of inferential justification should be able to show that [E supports P] plays a different role in justification from the role played by E itself. If, for example, one simply says that one must justifiedly believe [E supports P], then one has not accounted for how E and [E supports P] play different roles in the justification of P. iv)

There is nothing more to put together.

Lewis Carroll’s famous story of Achilles and the Tortoise shows something about inference (Carroll 1895). Taking some liberties with Carroll’s story, suppose one accepts E, which in fact supports P, but one does not realize that E supports P. Then one is not yet in a position to justifiedly accept P on the basis of E.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

INFERENTIAL APPEARANCES



Now suppose one believes E, and one also has a separate belief that E supports P, but one simply hasn’t put those beliefs together, so one has not noticed that one has any evidence for P. In that case, again, one is not yet in a position to justifiedly affirm P on the basis of E. Something more has to happen; one must put one’s beliefs together. Once we see that, we can see that it won’t do to simply add that the subject must believe that E and [E supports P] together support P, for there would remain the possibility that the subject hasn’t yet put these three beliefs together (namely, E, [E supports P], and [E and (E supports P) support P]). The same point holds if the requisite attitude toward [E supports P] is something other than belief. Whatever condition we add to our theory of inferential justification, it must somehow rule out that S has yet to ‘put together’ his beliefs or other cognitive states in such a way as to see why P must be (or is likely to be) true. The condition must entail that, intuitively, whatever needs to be put together has been put together. And it seems that this cannot be achieved merely by a condition requiring the subject to take an attitude toward a proposition stating that E and P stand in some logical relation. For whatever attitude one takes toward such a proposition, one might still have failed to put that together with one’s belief in E (Russell 1903: section 38). v)

Genuine support is not necessary.

For S to have inferential justification for P, it is not necessary that E actually, objectively support P. E may be logically irrelevant to P. Consider the Case of the Unfortunate Mathematician: S is a skilled mathematician who has just completed an apparent proof that P, starting from the single premise E. Though the reasoning was not especially difficult, nor was the conclusion especially surprising, S checked every step very carefully, and S has no reason to suspect that anything went wrong. But suppose (as is compatible with all the foregoing) that E does not actually support P; there is a subtle error that renders the proof neither valid nor cogent.

It seems to me that it would be perfectly rational for S, in this situation, to believe P. In fact, this would be rationally required; it would be irrational for S to refuse to accept P. This seems to show that inferential justification does not require any genuine logical connection between premises and conclusion; an apparent connection is good enough. I discussed this case in earlier work, where I wrongly tried to account for it by saying that S has justification for believing P, not on the basis of E alone, but on the basis of [E, and E supports P] (Huemer 2002: 336). That approach is misguided, because it assumes that S’s belief that P is based on a belief in [E, and E supports P]. This need not be the case. We can stipulate that the Unfortunate Mathematician believes P on the basis of E, simply. I think our intuition remains that the mathematician is rational. So I see no plausible way of accounting for the case except to admit

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



MICHAEL HUEMER

that one can be justified in believing P on the basis of E, even when E does not support P. vi)

Skepticism is false.

After arguing that inferential justification requires justification for believing that one’s premise makes probable one’s conclusion, Richard Fumerton worries about how this condition can be satisfied. It could not be the case in general that we have explicit inferential justification for believing that E makes probable P, since this would generate an infinite regress. For suppose we are inferentially justified in believing [E makes probable P], on the basis of some evidence E0 . This will require that we have justification for believing [E0 makes probable [E makes probable P]]. If that justification is also inferential, then there will have to be another proposition, E00 , such that we are justified in believing [E00 makes probable [E0 makes probable [E makes probable P]]] . . . So if Fumerton’s principle is right, there will have to be cases in which we have justification for believing [E makes probable P] not on the basis of any explicit inference. Fumerton considers the possibility that we have an a priori, intuitive awareness of the sui generis, abstract ‘making probable’ relation—but he finds it hard to convince himself of this story (1995: 218). Pace Fumerton, I do not regard skepticism as a serious option. At the risk of boring the reader, I will simply make this anti-skeptical methodological point. The concept of ‘inferential justification’ derives in the first place from observation of certain cases of beliefs that we take to be justified in this way. Moreover, theoretical principles of inferential justification—for example, putative necessary conditions on inferential justification—are based on epistemological intuitions. Some of these may be intuitions directly about the principles in question; others are intuitions about particular cases of putative inferential justification. For instance, when Fumerton discusses his Principle of Inferential Justification, he motivates it by appealing to certain cases in which, intuitively, a subject lacks inferential justification for a belief—such as the case of the astrologer who thinks there will be prosperity next year because Jupiter will be aligned with Mars (1995: 86). Now, if the Principle of Inferential Justification is motivated by epistemological intuitions, but then it turns out that the principle also contradicts a vast array of the clearest, most widespread, paradigmatic epistemological intuitions—if, indeed, it emerges that the principle is such a bad fit that it cannot accommodate any of the intuitive cases of inferentially justified belief—then it seems to me that we have a failed theory of inferential justification. Thus, I think that if one cannot find another way to avoid skepticism, then one had just better reject Fumerton’s Principle of Inferential Justification.4

4 For other ways of avoiding skepticism, see Hookway (2000); Rhoda (2008); Leite (2008: 433–8). For objections to these proposals, see Alexander (2012).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

INFERENTIAL APPEARANCES



3 An Appearance-Based Theory of Inferential Justification 3.1 Inferential appearances I have elsewhere written about the epistemic import of appearances, where these are understood as a special sort of non-doxastic, non-factive propositional attitude, which—so I’ve argued—normally provides justification (in the absence of defeaters) for beliefs (Huemer 2006, 2007, 2013). Perhaps when we speak of ‘seeing the connection’ between E and P, we are speaking of a species of appearance (cf. Tragesser 2009; Rhoda 2008: 220). If so, this species of appearance might aptly be dubbed an inferential appearance. Here is some more context. There are at least four species of appearances: sensory experiences, memory experiences, introspective appearances, and intellectual appearances. Only the last of these is particularly relevant here. Intellectual appearances require the exercise of the faculty of reason, otherwise known as ‘the intellect’ or ‘the understanding’; have conceptual content; and are normally the product of reflection on relatively abstract questions. There are at least two species of intellectual appearances: intuitions, which typically (always?) directly represent some abstract, necessary truth, and inferential appearances, which, roughly, represent some thing to be true given, or in the light of, some other presumed truth. For instance, I know that my couch is not large enough to fit both a tiger and a cow. So I believe [It is not the case that there are a tiger and a cow on the couch]. Now, it seems to me that, in light of the fact that there is not both a tiger and a cow on the couch, there must be either no tiger on the couch or no cow on the couch. This last appearance is an inferential appearance; it represents that a certain conclusion must be correct, in the light of a certain premise that I already accept. A few comments about inferential appearances. There are two species of inference, deductive inference and non-deductive inference (including induction, inference to the best explanation, and perhaps other non-deductive forms of reasoning). Accordingly, there are two species of inferential appearances. When one deductively infers P from E, one’s inferential appearance represents that in light of E, P must be the case. When one non-deductively infers P from E, one’s inferential appearance represents that in light of E, P is probably the case (or at least is more likely to be the case). The ‘must’ or ‘probably’ here is conditional. When P is deduced from E, it is not that P seems to be necessary, simply. It is that P seems to be necessary given E; that is, E seems to necessitate P. Similarly, when P is non-deductively inferred from E, E seems to probabilify P. At the same time, however, we don’t want to say that the appearance merely represents that E and P stand in a certain relation to each other. That view incurs Lewis Caroll’s problem and fails to satisfy condition (iv) from Section 2 (‘there is nothing more to put together’). Rather, we want to say that the

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



MICHAEL HUEMER

appearance presupposes the truth of E, and given that assumption (‘in light of E’), represents that P is either definitely or probably true. Why speak of ‘presupposition’ here? In the process of reasoning to P, the subject is not neutral with regard to the truth of E; he is assuming the truth of E. And we do not want to say merely that the subject believes E (though of course that will be true), because that leaves open the possibility that the subject has yet to put together his belief in E with the fact that E supports P (again, recall the Lewis Carroll problem). So the representation that shows the connection between E and P must also in some sense already take into account the (alleged) truth of E. But we also don’t want to say that the inferential appearance straight-out represents that E is true—it is not as though, when one starts reasoning from E to something else, one immediately acquires a new source of justification for E (viz., the inferential appearance). Thus, the attitude that the inferential appearance takes toward E is more akin to presupposition than to direct assertion: given E, which we’ve already assumed, something else ought to be true. The inferential appearance doesn’t have anything new to say about the truth of E.

3.2 Conditions for explicit inferential justification Here, then, is my account of the conditions for explicit inferential justification. S’s belief that P is inferentially justified to some degree, if and only if there is some E such that a) S is justified to some degree in believing E; b) S has an inferential appearance that, in light of E, P must be (or: is likely to be) the case; c) S’s justification for E does not depend on S’s having justification for P; d) S lacks robust defeaters for P; and e) S’s belief in E causes S’s belief that P via the inferential appearance, by a nondeviant causal chain. Why do I speak of one’s being inferentially justified ‘to some degree,’ rather than being inferentially justified full stop? Well, when one performs an inference, the conclusion typically winds up at least slightly less justified than the premise. This is why, if one performs a long series of calculations, starting from a single premise, one will often be overall unjustified in believing that the last step is correct, even if the first step was strongly justified and each succeeding step was rational. Still, the subject will have at least some inferential justification for believing the last step. Clause (a) in my account is obviously needed; otherwise, one could adopt the arbitrary belief, say, that there are six purple unicorns living in the Andromeda galaxy, and thence justifiedly deduce that there are more than five purple unicorns. Clause (b) is motivated by the conditions laid down in Section 2. (b) is my account of the sense in which the subject, in making an explicit inference, sees the connection (more accurately, seems to see the connection) between the premise and the

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

INFERENTIAL APPEARANCES



conclusion. I have elsewhere defended the idea that appearances are a source of justification. In earlier work, I primarily had in mind non-inferential justification; the present view is intended to extend/correct my view so as to cover inferential justification.5 Clause (c) is needed to rule out question-begging inferences. Clause (d) is needed because even if the previous conditions are satisfied, the subject will lack justification for P if, for example, the subject has independent proof of ~P, or the subject knows that his reasoning about P is completely unreliable. I take the range of defeaters to include things that provide S with justification for ~P (rebutting defeaters), as well as things that provide S with justification for thinking the belief that P lacks warrant (undercutting defeaters).6 Why the qualifier ‘robust’ in condition (d)? My account only gives conditions for a belief ’s being inferentially justified to some degree. Even if S has very strong (but not absolutely conclusive) grounds for denying P, or strong but inconclusive grounds for denying that S’s belief that P is warranted, S could still retain some justification for P. A defeater must be very powerful in order to render one’s belief completely unjustified. I shall not here try to describe exactly what such strong defeaters must be like, but the qualifier ‘robust’ is intended to capture the idea. Clause (e) is motivated by the following thought: when one is inferentially justified in believing P on the basis of E, it is not enough that one’s belief that E cause the belief that P in some way or other. There cannot just be a brute causal link. Rather, the subject must infer P from E because the subject seems to see a logical connection between E and P. Here is the reason for the qualifier about non-deviant causal chains. Suppose I believe E, and I see how E supports P, but I still refuse to accept P, because the thought of P makes me unhappy. However, an epistemically benevolent brain scientist with advanced technology shows up and scans my brain. The scientist detects my belief in E, my inferential appearance, and my refusal to accept P. This causes the scientist (who is offended by epistemic irrationality) to use his brainmanipulation machine to induce in me a belief that P. Here, my belief that P is caused by the belief that E and the inferential appearance (via the brain scientist), but intuitively, I lack an inferentially justified belief. The problem, roughly speaking, is that this is not a normal way for these things to cause a belief; it is a ‘deviant’ causal chain. Deviant causal chains can afflict virtually any causal condition. Killing involves causing a death, but not just any way of causing a death counts; there can be deviant causal chains by which one’s action causes a death, such that one does not count as a killer. Similarly, breaking a vase is causing it to be broken by a non-deviant causal 5

Cf. Huemer (2013: 238–41). Here, I use ‘warrant’ in Plantinga’s (1993: 3) sense, for the property that must be combined with true belief to yield knowledge. 6

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



MICHAEL HUEMER

chain. And when a subject perceives an object, the object must cause the subject’s perceptual experience by a non-deviant chain. Naturally, I am not going to try to define ‘deviant causal chain’; I leave that at an intuitive level. For that reason among others, my account is not a full explanation of the conditions for inferential justification. The account is nonetheless an interesting, substantive thesis.

4 The Inferential Appearance Theory Satisfies the Conditions for a Theory of Inferential Justification In Section 2, I advanced six principles that an adequate theory of inferential justification should respect. To what extent does my Theory of Inferential Appearances satisfy these constraints? i)

S must have some attitude toward [E supports P].

This is satisfied by the inferential appearance that in light of E, P must be or is likely to be true. ii)

Justified belief in [E supports P] is insufficient.

The theory does not advert to a belief (justified or otherwise) that E supports P; it requires an appearance, which is distinct from belief. iii)

[E supports P] is not a premise.

In the Inferential Appearance Theory, the role played by E is quite different from the role played by [E supports P]. E must be justifiedly believed; it is then presupposed by the inferential appearance, which represents that P is definitely or probably true because of E. Thus, the theory does not at all treat [E supports P] in the same way that it treats the actual premises of the inference. This differs, for example, from Fumerton’s theory, which merely states that both E and [E supports P] must be justified. iv)

There is nothing more to put together.

The inferential appearance resolves Lewis Carroll’s problem of Achilles and the tortoise. When S bases a belief that P on a belief that E, according to the present theory, S needs an inferential appearance to bridge the gap between the belief that E and the belief that P. But one might wonder: why is there not also a need for a second appearance to bridge the gap between the inferential appearance and the belief that P?7 Why couldn’t Lewis Carroll’s tortoise have the inferential appearance but fail to see the connection between that appearance and the conclusion P?

7

Chris Tucker first raised something like this issue with me (personal communication, 2012).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

INFERENTIAL APPEARANCES



In reply, the reason why something is needed to bridge the gap between E and P is that we have the intuition that a subject must see the connection between E and P, in order to acquire explicit inferential justification for P on the basis of E. We further intuit that the subject must believe P in part because the subject sees that connection. But I take it that we do not similarly have an intuition that a subject must see the connection between the connection between E and P and P. Once the tortoise has the inferential appearance, he has all he needs. Having the inferential appearance just is having it seem to you that P must be true, given E. The tortoise obviously cannot be in this state while simultaneously failing to see why P must be true. And there is no need for the tortoise to see a connection between the inferential appearance and P; the tortoise need only see the connection between E and P, which is just a matter of having the inferential appearance. v)

Logical support is not necessary.

The theory does not require that E actually support P; it only requires an appearance which represents E as supporting P. vi)

Skepticism is false.

Appearances, in the absence of defeaters, provide justification for believing their contents. It is therefore natural to hold that an inferential appearance that in light of E, P must be true would provide justification for believing P in the absence of defeaters. There is, however, the additional requirement that E must itself be justified. This stems from the fact that the inferential appearance does not simply represent that P, full stop; it represents that P must be or is likely to be true given E, which is presupposed. That presupposition must therefore be justified in order for P to receive any justification from this source. What should we say about Fumerton’s scruples concerning inferential justification? Fumerton finds it difficult to convince himself that he has an intellectual insight into an abstract, sui generis ‘making probable’ relation. I myself do not. Here are some observations that might help explain my lack of concern on this score: First, I hold that we grasp abstract objects (namely, universals) all the time. When I think the humble proposition [Squirrels are furry], I deploy the concepts SQUIRREL and FURRY, each of which constitutes a grasp of a certain universal (squirrelhood and furriness). Second, I think there is an enormous number of undefinable concepts, including most lexical concepts (concepts expressed by single words in a natural language). Obviously, there is a great deal more to be said about each of these first two points, but the draconian length limit imposed by this volume’s editors prevents me from saying any of it.8 Here, I will just report that in view of these first 8 One of the editors has objected that the length limit is not actually draconian. I do not have space to address this. I don’t even have enough space to tell you all the things that I don’t have space to tell you. For more on realism about universals, see Armstrong (1978); on the indefinability of most concepts, see Fodor (1998); Huemer (2015).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



MICHAEL HUEMER

two points, I find no difficulty in the idea that, in thinking about probability, one is grasping a unique abstract object, whose nature cannot be precisely explained using other concepts. Third, I am not postulating some new mental state that no one has ever thought of before. I am talking about the state of seeing the connection between the premises and the conclusion of an inference, an experience with which I assume all my readers are very familiar. I have made some theoretical claims about that mental state, claims that could be disputed. But given an awareness of this kind of experience, I don’t think any of my claims are particularly outlandish. Fourth, being aware of an F does not entail being aware that the thing in question is an F. Thus, it might be true that, in thinking about probability, one is aware of an abstract, logical relation, even if one does not realize that this is the sort of thing one is aware of. And one need not appeal to direct reflection on the concept of probability to establish that this is the sort of thing it denotes; one can simply argue that the hypothesis of an abstract, sui generis logical relation best accommodates a variety of widely shared intuitions about probability. Thus, in defending the logical interpretation of probability, I would argue that neither the frequency interpretation, nor the subjective interpretation, nor the propensity interpretation of probability is adequate to our intuitions about certain cases (even apart from considerations about avoiding inferential skepticism).9 That, of course, would be a topic for a separate paper (again, draconian length limits). But now suppose I am wrong about that. Suppose that either (a) there is no such thing as logical probability, or (b) there is such a thing, but we do not have appearances about logical probability when we make explicit inferences. Should we then become inferential skeptics? No, we should not. Fumerton holds that, to be justified in believing P on the basis of E, one needs justification for believing that E makes probable P. I hold that, to be (explicitly) inferentially justified in believing P on the basis of E, one needs an inferential appearance which represents that in light of E, P must be or is likely to be true (which I take to provide justification for believing that E supports P). These alleged requirements cannot rationally support inferential skepticism, because the only sound motivation for these requirements—either Fumerton’s requirement or mine—is something that is incompatible with the inferential skeptic’s argument. To explain: the sound motivation for Fumerton’s and my requirements is that they are ways of accommodating the intuition that in a genuine case of explicit inferential justification, the subject sees the connection between the premise and the conclusion. That intuition is best brought out by contrasting the normal case of explicit inferential justification—where the subject does see that connection—with cases where the subject ‘makes an inference’ without seeing the connection (e.g., the subject just 9

What about the epistemic interpretation? I would seek to merge the epistemic and logical interpretations of probability in the manner suggested in Section 6 below.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

INFERENTIAL APPEARANCES



whimsically decides to infer P from E, or the subject infers P because someone else told him that E supports P). If no one ever saw the connection between a premise and a conclusion, then there would be no such contrast, and there would be no intuition that seeing the connection was required for inferential justification. So whatever condition we introduce to accommodate this intuition, it should be a condition that is satisfied in the cases that we normally describe as cases of one’s ‘seeing the connection’ between a premise and a conclusion. If you think those cases do not involve having a representation of logical probability, then you should not require a representation of logical probability as a condition on inferential justification.

5 Revisiting Fumerton: Acquaintance vs. Appearance As noted, Richard Fumerton holds that in general, one is inferentially justified in believing P on the basis of E, only if one (i) is justified in believing E, and (ii) is justified in believing that E makes probable P. I assume that clause (ii) is to be understood in such a way as to require that one have propositional justification for [E makes probable P], but not to require that one have a justified belief that [E makes probable P]. I assume also that ‘E makes probable P’ is a notational variant of ‘E supports P.’ So understood, my own views now (contrary to my earlier work) entail the truth of Fumerton’s principle. This is because the inferential appearance that I take to be required for inferential justification is also something that, in my view, provides justification (in the absence of defeaters) for [E supports P] (and the absence of defeaters is also included among the conditions for inferential justification). So Fumerton is right. But while I agree with Fumerton, I think he will not agree with me. That is, I have gone beyond Fumerton’s principle in ways that he will reject. I have stated a specific way in which the subject must have justification for [E supports P]; other ways of having justification for [E supports P], in my view, do not count. For instance, if one has testimonial justification for [E supports P] or memorial justification for [E supports P], this is not enough to enable one to acquire an inferentially justified belief in P on the basis of E. I suspect that Fumerton will agree with me about the testimony and memory cases. But he will disagree with my claim that an inferential appearance confers justification for [E supports P]. He will say that one must instead be acquainted with the support relation (perhaps among other things).10 Here is my argument for preferring an appearance-based account over an acquaintance-based account of inferential justification. One salient difference between appearances and states of acquaintance is that acquaintance is by definition successful, whereas appearances can be either veridical or unveridical. In the present 10 Fumerton (1995: 75) has proposed that in general, to be non-inferentially justified in believing that P, one must be acquainted with the fact that P, with the thought that P, and with the correspondence relation between the fact and the thought.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



MICHAEL HUEMER

case, this means that one cannot have the relevant acquaintance unless E really does support P. But one can have the relevant appearance state even if E does not really support P. Thus, a test case would be one in which a subject has the relevant appearance, all the other conditions in my account of inferential justification are satisfied, but E does not genuinely support P. In such a case, my account will take the subject’s belief to be inferentially justified, while Fumerton would take the subject’s belief to be unjustified. We have already seen such a case. I called it “the Case of the Unfortunate Mathematician.” The mathematician believes P on the basis of E, where E seems to entail P and he has no reason for doubting this. Unbeknownst to the mathematician, his argument contains a subtle error, rendering it invalid. I take it that the mathematician’s belief in this case is justified. This can only be accounted for in terms of the appearances. We can extend the conclusion to normal (successful) proofs as well. Suppose the same mathematician also constructs a proof of Q. This proof seems to him just as good as the proof of P, and the proofs are relevantly similar in all other ways, except that the proof of Q does not contain any mistakes, so it is valid. Everyone will agree that the mathematician’s belief that Q is justified. We can now construct an argument parallel to one Fumerton himself deployed in another context:11 1. The mathematician is justified in believing P in the same way that he is justified in believing Q. 2. The mathematician’s justification for believing P does not derive, even in part, from acquaintance with a support relation between the premise and the conclusion of his argument. 3. Therefore, the mathematician’s justification for believing Q also does not derive, even in part, from acquaintance with a support relation between the premise and the conclusion of his argument. To put the central idea more simply: if the mathematician’s belief that P is justified because of an inferential appearance (or series of appearances), rather than a state of acquaintance (or series of acquaintances), then it just is very plausible to say that the mathematician’s belief that Q is also justified because of inferential appearances, rather than states of acquaintance. And so states of acquaintance are not needed to explain inferential justification in any case.

6 Inexplicit Justificatory Dependence I have so far considered only cases of fully explicit inference. Here I comment on the other forms of justificatory dependence mentioned in Section 1 above. 11

See Fumerton (1985: 79–80), arguing against perceptual direct realism.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

INFERENTIAL APPEARANCES



Start with the case of inference against a background. S infers P from E, where some complex mass of background information, B, is required to explain why S is justified in making the inference. Suppose, as is normally the case, that S lacks occurrent awareness of most of this background information at the time of the inference. Just as in the case of fully explicit inference, however, S must have an inferential appearance. What should be the content of the appearance? Should it represent that E supports P, or that E & B together support P? We should not demand an appearance that E & B supports P, because it is not plausible that this appearance occurs in normal cases of inference against a background. When I see the shredded couch fabric and infer that the cat shredded the couch, it seems to me that the current state of the couch is evidence that the cat shredded the couch. I do not have an appearance representing that the state of the couch, together with the facts that the cat was present all day, that the cat has claws, that no other animal was present, that fabric usually does not spontaneously become shredded, that shredding a couch usually requires being in the same room with it, etc.—all of that taken together—is evidence that the cat shredded the couch. Most of that background information does not come into my mind at all, neither in the form of belief nor in the form of appearance. So it looks as though we should require of the subject only an appearance that E supports P. But then we face another problem: it could be argued that E by itself in fact does not (to any significant degree) support P. If we think of support as an abstract logical relation, it is far more plausible to hold that [E & B] supports P than it is to hold that E alone supports P. So it looks as though inferences made against a background will be inferences based in part on false appearances. This is logically compatible with the subjects’ being justified in making the inferences (if they do not know that the inferential appearances are non-veridical). It would, however, prevent subjects from gaining knowledge through these inferences. So we seem to face a dilemma: if we demand an appearance that [E & B] supports P, we lose inferential justification because it is implausible that we actually have such appearances; if we demand only an appearance that E supports P, we lose inferential knowledge because our inferential beliefs will be based on false appearances. The case of non-inferential justificatory dependence poses a parallel dilemma. In this case, S has a belief that P, which is supported by S’s background information, B, but S has not actually inferred P from anything—as in the case of my present belief that I live in the United States. If we demand that S have an inferential appearance that shows how B supports P, we lose justification, because it is implausible that we actually have such appearances. If we demand only that S have an appearance to the effect that P is likely, we lose knowledge, because P is not in fact (unconditionally) likely; P is only likely given B. My best suggestion to resolve this dilemma is to relativize such things as ‘support’ and ‘likelihood’ to one’s background information. Consider ordinary cases in which we have a sense that something is likely, that something is unlikely, or that some

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



MICHAEL HUEMER

thing renders another thing more likely. These are all probability assessments. What sort of probability do they invoke? My suggestion is that these thoughts typically refer to logical probability conditioned on one’s own background information. How do our thoughts count as referring to this kind of probability, when we are not conscious of all that background information? The answer is twofold: (a) our probability thoughts are generated by brain processes that in fact are shaped by all (or almost all) of the relevant background information, including information that we are not conscious of at the moment; and (b) we are disposed, if the issue arises, to acknowledge the relevance of that background knowledge to our probability thoughts. Thus, when the issue arises as to what country I occupy, it seems to me very likely (to put it mildly) that I live in the United States. This sense of likelihood is shaped by a host of past experiences and beliefs of mine that are relevant to where I live, including many experiences and beliefs that I do not now remember. That is part of why my sense of likelihood counts as referring to the probability of my living in the United States conditional on all those past experiences and beliefs. But reference is not solely determined by this sort of external factor; it is also a matter of how I would react if some of these past experiences or beliefs were raised in connection with the question of how likely it is that I live in the United States. For example, if the question arises, I will view the fact that Colorado is in the United States (which I believe) as supporting the claim that I ‘probably’ live in the United States. And so for all of my other relevant background information. These dispositions show that my sense of ‘probability’ refers to probability given my background information. To return to my account of inferential justification, I proposed the following necessary condition on explicit inferential justification: b)

S has an inferential appearance that, in light of E, P must be (or is likely to be) the case.

How should we extend this to the case of an inference made against some complex background information, B? In this sort of case, the subject should have an inferential appearance representing that E makes probable P, where this refers to probability given that subject’s background information—and the subject may perfectly well have such an appearance even though the subject is not conscious of most of that background information at the time. Thus, when I see the shredded couch, it seems to me that the shredded state of the couch renders it likely (given my background information) that the cat shredded the couch, even though I am not thinking all of that background information at the moment.

7 Conclusion According to the theory of inferential appearances, inference involves a special kind of appearance, which represents the conclusion of the inference as being either necessarily or probably true given the premises, relative to the subject’s current

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

INFERENTIAL APPEARANCES



background information. This appearance presupposes the truth of the premises while representing the conclusion as likely because of those premises. This explains the sense in which, in a successful inference, one sees the connection between the premises and the conclusion. Contrary to other theorists, I have argued that a justified belief that the premises support the conclusion is not only unnecessary but insufficient, since one can justifiedly believe that E supports P without actually seeing how E supports P, and in such a case, one is not in a position to infer P from E alone. Regardless of what one believes, justifiedly or not, one needs the inferential appearance. The theory thus appropriately treats [E supports P] differently from the premises of the inference, which need (only) be justifiedly believed. The inferential appearance theory avoids Lewis Carroll’s problem of Achilles and the tortoise, by positing a mental state that simultaneously takes into account the truth of the premises and the logical connection between the premises and the conclusion—a mental state representing the conclusion as likely given the premises. This mental state rules out the possibility that one fails to see why P must be true. Unlike theories based upon the notion of acquaintance, the appearance-based theory rightly accommodates the possibility of an inferentially justified belief based upon a mistaken (invalid and uncogent) argument. Provided that an argument seems perfectly good to the person and they have no grounds for doubting this, a person can be justified in accepting the conclusion based on the premises, whether or not the argument is in fact cogent. Finally, the inferential appearance theory, when combined with Phenomenal Conservatism, satisfies our goal of avoiding skepticism. Since we are justified in trusting our appearances unless and until we acquire grounds for distrusting them, we have an easy explanation of why most inferential beliefs are justified. More demanding theories of justification, such as acquaintance-based theories, leave significant room for doubt as to whether we ever satisfy the conditions for justified belief, particularly in the case of inferential beliefs. But it is scarcely open to doubt that it often seems to us that some conclusion is rendered probable by some premises. It would then be the skeptic’s burden to show the appearance to be untrustworthy.

References Alexander, David. (2012). “Weak Inferential Internalism,” Journal of Philosophical Research, 37: 357–77. Armstrong, David. (1978). Nominalism and Realism. Cambridge: Cambridge University Press. Carroll, Lewis. (1895). “What the Tortoise Said to Achilles,” Mind, 4: 278–80. Fodor, Jerry. (1998). Concepts: Where Cognitive Science Went Wrong. New York: Oxford University Press. Fumerton, Richard. (1985). Metaphysical and Epistemological Problems of Perception. Lincoln: University of Nebraska Press.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



MICHAEL HUEMER

Fumerton, Richard. (1995). Metaepistemology and Skepticism. Lanham, MD: Rowman and Littlefield. Fumerton, Richard. (2006). Epistemology. Malden, MA: Blackwell. Hookway, Christopher. (2000). “Scepticism and the Principle of Inferential Justification,” Philosophical Issues, 10, Skepticism: 344–65. Huemer, Michael. (2002). “Fumerton’s Principle of Inferential Justification,” Journal of Philosophical Research, 27: 329–40. Huemer, Michael. (2006). “Phenomenal Conservatism and the Internalist Intuition,” American Philosophical Quarterly, 43: 147–58. Huemer, Michael. (2007). “Compassionate Phenomenal Conservatism,” Philosophy and Phenomenological Research, 74: 30–55. Huemer, Michael. (2013). “Phenomenal Conservatism Uber Alles,” in Chris Tucker (ed.), Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. Oxford: Oxford University Press, 328–50. Huemer, Michael. (2015). “The Failure of Analysis and the Nature of Concepts,” in Christopher Daly (ed.), The Palgrave Handbook of Philosophical Methods. New York: Palgrave Macmillan, 51–76. Leite, Adam. (2008). “Believing One’s Reasons Are Good,” Synthese, 161: 419–41. Plantinga, Alvin. (1993). Warrant: The Current Debate. Oxford: Oxford University Press. Rhoda, Alan R. (2008). “Fumerton’s Principle of Inferential Justification, Skepticism, and the Nature of Inference,” Journal of Philosophical Research, 33: 215–34. Russell, Bertrand. (1903). The Principles of Mathematics (Vol. 1). Cambridge: Cambridge University Press. Tragesser, Robert S. (2009). “Inference,” in Ernest Sosa, Jonathan Dancy, and Matthias Steup (eds), A Companion to Epistemology (2nd edition). Hoboken, NJ: Wiley-Blackwell, 444–5. Tucker, Chris. (2012). “Movin’ on Up: Higher-Level Requirements and Inferential Justification,” Philosophical Studies, 157: 323–40.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

PART III

Traditional Internalism and Skepticism

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

Responding to the Skeptic

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

8 The Costs of Demon-Proof Justification Sanford C. Goldberg

1 Imagine an epistemologist who becomes obsessed over his career with answering skeptical doubts.1 Let us call this epistemologist Ricardo. Ricardo is aware of all of the responses that people make in the face of such doubts. He is aware of attempts to use externalist accounts of justification and knowledge in an attempt to respond to skepticism (see e.g. Sosa 1994); he finds that all of them are inadequate. He is aware as well of attempts to use externalist accounts of attitudinal content as responses to skepticism,2 and he finds that they too are inadequate. What Ricardo really wants is the sort of epistemic assurance one can have from the armchair—a sort of assurance that one’s epistemic house is in order which, when it is in order in this way, one can ascertain that it is by mere reflection alone. To this end, Ricardo sets out to frame an epistemic assessment that answers to this constraint. Ricardo reasons as follows. Suppose you perceive that p, and on the basis of this perceptual experience you come to believe (and so, presumably, to know) that p. Now imagine your envatted doppelganger. (S)he, too, forms the belief that p, on the basis of what (s)he takes to be the perception that p.3 Only (s)he is a BIV, and so doesn’t know. Even so, it is natural to suppose that your doppelganger’s justification is as good as is yours; it’s only that the world hasn’t complied, and so your BIV doppelganger fails to know. But this sort of failure has no power to affect the sort of justification which, intuitively, his/her belief enjoys. What we want from the property of justification, then, Ricardo thinks, is something that depends on factors that cannot be affected by anything within the Demon’s control.

1 2 3

With thanks to Mike Bergmann, for comments on an earlier draft of this chapter. See e.g. the discussions in Brueckner (1992) and Brown (2004). Let’s agree to waive worries about content-determination.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



SANFORD C . GOLDBERG

On this basis Ricardo proposes a recipe to handle the effects of the Demon, and in this way to render justification ‘Demon-proof.’ Identify the set of things the Demon can affect—the truth-value of one’s belief, the presence (or not) of luck in the environment, the reliability of one’s cognitive processes, and so forth—and then simply regard this very fact, and indeed all facts of this sort, as ‘external’ to epistemic justification, where being ‘external’ to justification amounts to being irrelevant to assessments of justification. The goal will then be to characterize a notion of justification whose obtaining (or not) in no way depends on the ‘externals.’ Ricardo’s recipe, then, is to regard anything that the Demon can affect as relevantly ‘external,’ and in this way to carve off a range of factors—the non-external, or internal ones—as constituting the supervenience base for justification. Since the internal factors are identified precisely by their being beyond the reach of the Demon, any notion of justification that supervenes on factors that are internal in this sense will be secured against demonic interference.4 I will call any notion of justification that conforms to this recipe Demon-proof justification. Ricardo was hankering after Demon-proof justification. He devotes his great philosophical talents to formulating a notion of justification that is Demonproof, and to showing how it bears on the variety of types of belief many take to be justified: perceptual belief, inferentially supported belief, memorially supported belief, etc. Clearly, Demon-proof justification is a species of internalist justification; and any internalist whose theory proposes an account of justification on which justification is Demon-proof I will call Demon-proof internalism. (In what follows, when I speak of internalism it will always be Demon-proof internalism I have in mind, unless otherwise noted.) Ricardo’s research agenda is to formulate and defend a version of Demon-proof internalism. Now Ricardo is a fictional character—one I just made up. (All characters appearing in this work are fictitious. Any resemblance to real persons, living or dead, is purely coincidental.) Even so, I can say that I have been deeply influenced by a real philosopher who in some important respects is like Ricardo. I refer, of course, to Richard Fumerton, author of Metaepistemology and Skepticism (1995). This chapter is dedicated to him. In it, I try to pay off a debt that I have owed him for some time: to say why I think that any theory of justification on which justification is Demon-proof will prove to be unacceptable. Elsewhere I have tried to argue that there can be no coherent notion of Demon-proof justification—not, at any rate, if justification is to play the role that epistemology assigns to it (see Goldberg 2012). I now worry that such an argument trades on an illicit understanding of epistemic luck (see Dutant unpublished). Here I will be arguing for a weaker claim: Demon-proof justification does not have the value that we expect from a notion of epistemic justification. 4 I should acknowledge that this characterization of ‘internal’ is non-standard; but it should be relatively clear, if difficult to spell out explicitly, how this non-standard notion relates to more traditional ‘accessibilist’ notions of the internal.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

THE COSTS OF DEMON - PROOF JUSTIFICATION



To bring this out, I will be arguing as follows. Suppose that epistemic justification is Demon-proof. In that case we must allow that all internal duplicates of a subject S’s are equally well-justified, so that S’s belief enjoys no more justification than her epistemically worst-off internal duplicate does. When we consider the variety of the kinds of demonic intervention that are possible, the result, I submit, is that we will see just how weak Demon-proof justification actually is. I will bring this out by arguing that Demon-proof justification will not have the value we typically associate with doxastic justification. The fact that the Demon can be used to bring this out, then, suggests that the Demon may have the last laugh after all—Ricardo’s strategy notwithstanding.

2 Your envatted doppelganger is such that most (or all?) of her empirical beliefs about her environment are false. Still, many epistemologists stand with Ricardo in embracing the idea that in other respects your envatted doppelganger is doing well, epistemically speaking. They insist that there is an intuitive sense in which (s)he is and has always been responding appropriately to the evidence of his/her senses. Standard versions of internalism use this idea to motivate the claim that correct epistemological theory will regard your doppelganger as exactly as well-off, justification-wise, as you are. What is more, insofar as the Demon cannot affect one’s evidence,5 it would then seem that the Demon cannot affect one’s justification. In which case we have arrived at a Demon-proof kind of justification. Unfortunately, things are not so simple: the Demon has far greater powers than what is seen in BIV cases, and when we appreciate this, we will appreciate just how great is the cost of Demon-proofing the notion of justification.6 In what follows I want to describe three more types of ‘demonic interventions,’ each of which bears on how well the subject is responding to her evidence. Or at least it would seem that way; the internalist may want to describe matters differently, of course, but my point will be that she does so at some cost. The first type of demonic intervention we might describe as a case of ‘partial envatment.’ To bring out this sort of case, imagine that there you sit, looking out the window, coming to the perceptual knowledge that the sun is low on the horizon. Your envatted doppelganger forms the same belief; but her belief is false. But, for reasons to be revealed, the Demon isn’t happy merely to have produced subjects who have always been and always will be envatted. Rather, the Demon wants cases involving subjects whose periods of (something like) envatment obtain at irregular

5

This is a matter to which I will return, below. When I say this, I have in mind the endorsement of such a notion as one’s only notion of epistemic justification. 6

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



SANFORD C . GOLDBERG

intervals, and last for an arbitrarily long period of time. Hence the curious case of Ms. Appelbaum: APPELBAUM Ms. Appelbaum happens to be attending her middle child’s swim meet when the Demon strikes. The result of the demonic intervention is that for 25 minutes Ms. Appelbaum is an internal duplicate of yours. In the midst of this interval, she remembers nothing about her life before that 25 minutes, has illusions of memories corresponding to your memories, and will remember nothing about this 25 minutes after the interval is all over, returning to her normal programming, etc. (Yes, the Demon is that good.) At one point in the interval, Ms. Appelbaum takes herself to be looking out of a window, and forms the belief that the sun is low on the horizon. If justification is Demon-proof, then Ms. Appelbaum’s belief is as justified as yours is. Still, things are worse for Ms. Appelbaum. For in that 25-minute interval, she seems to recall doing all sorts of things none of which she has done, seems to remember all sorts of friends none of whom actually know her, seems to recognize a variety of people she has never in fact met, seems to remember having committed herself to all sorts of future plans, and so forth. So not only is she wrong about her immediate environment (in this way she is like your envatted doppelganger). In addition, she is wrong about both her past and her future; and, what may be the same thing, she is systematically wrong about the very course her experience has taken prior to the 25-minute interval itself, and she is systematically wrong in the predictions she makes regarding the course her experience will take going forward (in moments beyond the 25-minute interval). For this reason, most or all of her temporal beliefs which make reference to times outside the present 25-minute interval are false.

Ms. Appelbaum is in a terrible situation. Of course, if internalism is right, her belief that the sun is low on the horizon is exactly as well-justified as is your belief to that effect, her belief that she recognizes that fellow over there in the corner is exactly as well-justified as is your corresponding belief regarding a corresponding fellow near you, her belief that she has committed to going to a barbecue next month is exactly as well-justified as is your belief to that effect, and so forth. But this now seems a cost to the internalist view. In particular, since she is not in a particularly happy epistemic situation, the insistence that these beliefs of hers are as well-justified as yours implies that you are not in a particularly happy epistemic situation. In a moment I will say why I think that Ms. Appelbaum is not in a particularly happy epistemic situation. Before I do, however, it will be good to contrast Ms. Appelbaum’s situation with that of your BIV doppelganger.7 Whereas both Ms. Appelbaum and the BIV have systematically false memorial beliefs, Ms. Appelbaum is unlike your BIV doppelganger in that her faulty memories include false memories of her own past mental states. In this respect Ms. Appelbaum is akin to Russell’s subject who came into existence 5 minutes ago (with implanted ‘memories’ of a lifetime). But Ms. Appelbaum differs from Russell’s 5-minute-old subject in an important respect as well: whereas in the case of Russell’s 5-minute-old subject the false ‘memories’ are implanted, in the 7

With thanks to Mike Bergmann, for suggesting that I consider this case.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

THE COSTS OF DEMON - PROOF JUSTIFICATION



case of Ms. Appelbaum the false ‘memories’ supplant real ones. And this point brings me to my claim that Ms. Appelbaum is not in a particularly happy epistemic situation. To see this, consider that there is a mass of evidence to which Ms. Appelbaum is utterly insensitive: namely, the evidence she had prior to the start of the 25-minute interval, which evidence will become accessible to her once again after the 25minute interval is over. Notice that neither your BIV doppelganger nor Russell’s 5-minute-old subject suffer from a corresponding condition. Whatever one wants to say about the epistemic value of their ‘memory’-sustained evidence, those subjects aren’t insensitive to evidence they’ve acquired in the past. But this evidenceinsensitive condition is the condition Ms. Appelbaum is in, when she is within the 25-minute interval of demonic possession. Now the internalist epistemologist will think to reply that when Ms. Appelbaum is in that 25-minute interval she doesn’t actually have the evidence from her past, and so the claim that she is insensitive to that evidence (or that she forms beliefs in a way that does not conform to that evidence) is irrelevant to whether she is responding properly to the evidence she does have.8 But, it should be pointed out, this underscores the very reason that the Demon wanted cases of partial envatment of the Ms. Appelbaum sort in the first place: it enabled the Demon to mess with a subject’s relation to her previously amassed evidence. And in any case the internalist epistemologist who reacts in this way has in effect acknowledged a cost of the view. For it now seems that the internalist is forced to understand what it is to ‘have’ evidence in such a way that the Demon could take your evidence away from you at a moment’s notice, and replace it with very different evidence. To be sure, you could always respond appropriately to the newly given evidence; but it is hard to see how your doing so puts you in a particularly happy epistemic situation, given that this evidence is worthless when it comes to revealing your actual environment (understood to be invariant across the Demon’s manipulations). It is also unclear whether such a view puts us in a position to appreciate why (epistemically speaking) it is a good thing to collect more evidence: if evidence is the sort of thing that the Demon can take away at will, then insofar as epistemology is in the business of assessing a Demon-proof condition, what is the value of collecting more evidence? Perhaps the defender of internalism will try to avoid this line of reasoning altogether, by denying that Demon-proof internalism is committed to the verdict that Ms. Appelbaum’s belief (that the sun is low on the horizon) is as justified as yours is. After all, the internalist can say that Ms. Appelbaum and you are not internal duplicates. Prior to the interval, there are all sorts of differences in your respective courses-of-experience. And after the interval, there will be further differences in your respective courses-of-experience. But (the internalist insists) two subjects are internal duplicates only if their whole internal histories are duplicates. Unfortunately for the internalist, this reply comes at the cost of internalism itself. 8 See for example Feldman (1988), where the view defended is (roughly) that one only had evidence one can access through memory.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



SANFORD C . GOLDBERG

This is for the simple reason that the factor being cited—whether one’s whole courseof-experience is a duplicate of another’s whole course-of-experience—is not itself Demon-proof. The proof of this is simply that the Demon can manipulate this condition, making it seem to one as if things are one way, when in fact they are not. Indeed, this is the situation with poor Ms. Appelbaum when she is within the 25-minute interval. I should add, too, that no one who endorses an ‘accessibilist’ version of internalism should be tempted by this move, for the simple reason that the fact in question—pertaining to the relation between the entirety of your course-ofexperience compared to that of Ms. Appelbaum—is something that is beyond the ken of Ms. Appelbaum when she is within the interval itself. This fact is one that is not reflectively available to Ms. Appelbaum while she is within the interval itself—nor indeed when she is without the interval (as the Demon ensures that she has no memory of the switch). Moving on from the case of Ms. Appelbaum, there is a second type of demonic intervention that we must consider. Here, the Demon is not satisfied to wreak havoc on a subject’s relation to ‘her’ evidence by way of ‘partial envatment’ cases. The Demon also wants to mess with a subject’s sense of himself as a competent reasoner, one capable of reasoning to conclusions e.g. about how likely it is that he is in a Ms. Appelbaum-type situation. Here the Demon targets the subject’s capacity to assess his own reasoning competence, as witnessed in the strange case of Mr. Fitzgerald: FITZGERALD Mr. Fitzgerald is an ordinary man minding his own business and living an ordinary life in Sewanee, Tennessee. One day, the Demon decides to affect Mr. Fitzgerald’s cognitive life as follows: while Mr. Fitzgerald’s memory and perceptual faculties are intact, his ability to appreciate the strength of his reasons is undermined, so that his appreciation for (and judgments regarding) the strength of his reasons is (are) utterly at odds with the actual strength of those reasons. So he systematically takes the weaker reasons to be stronger, and the stronger reasons to be weaker. Unfortunately, his metacognitive abilities are also affected: it appears to him that all is well with his ability to discern the strength of his reasons, even after the demonic intervention.

Now clearly Mr. Fitzgerald is not your doppelganger (unless you reside in a place that is indistinguishable from Sewanee, Tennessee, and lead a rather boring ordinary life). But his case is instructive nevertheless: it highlights yet another power of the Demon. Corresponding to this power, we will have another set of doppelgangers for you: these are people who are internal duplicates of yours,9 yet, owing to the Demon’s fooling around with their external environment, what they take to be weighty reasons are weak (not highly reliable indicators of truth), what they take to be weak reasons

9

As we will see, when it comes to deductive reasoning, we will need to be more qualified in how we bring out the point I wish to bring out.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

THE COSTS OF DEMON - PROOF JUSTIFICATION



are actually weighty (highly reliable indicators of truth), and yet this is indiscernible to them: they think their reasons-assessing-faculties are in perfect order. It is important to see how there can be doppelgangers of the sort I am describing here. It will be allowed on all sides that the Demon can mess with contingent environmental regularities. Suppose that tracks of a certain appearance are highly correlated with the presence of an animal of a certain kind around here, in our world. (Members of that species or kind make tracks with that distinctive appearance.) Well, the Demon can mess with these things, so that at another world tracks of that appearance are highly correlated with the presence of another kind of animal, or perhaps are not correlated with the presence of one type of animal at all. Now imagine that your twin, D, resides in a world in which tracks of this appearance do not correlate with any one type of animal: they are caused by a random set of events (none of which include being caused by the imprint of the hooves of the type of animal whose hooves regularly make tracks like that here, in your world). Then D sees tracks of this appearance, and D infers (as do you) that there must be an animal of the relevant sort present. Only in D’s world tracks of this sort have never been caused by this type of animal, so D’s grounds are not reliable indicators of truth, despite D’s taking them to be such. Of course, since D is your internal duplicate, D has had precisely the same course-of-experiences you have had; but the Demon has ensured its being the case that in each occasion on which you confirmed that the tracks were caused by the animal in question (i.e., by seeing the animal cause the tracks), your doppelganger D suffered from a corresponding (internally indistinguishable) hallucination in which it seemed to D that the tracks were so caused. But they weren’t. The end result is that when D’s inductive practice proceeds on the supposition that tracks of that appearance are good (inductive) grounds for believing that the animal in question caused them, D’s inductive practice is weak (the supposition being false), even as your corresponding inductive practice is strong (the supposition being strong in your case). And D is none the wiser for this. Notice that insofar as the Demon does this for all of D’s inductive practices, D’s world begins to seem very much like that of the BIV—D being systematically deceived. But still we can imagine the Demon being somewhat selective in his interventions; sometimes D’s inductive practice is strong. The difficulty for D is that D can’t discern which is which, worse, can’t discern even that there is something wrong in some cases with D’s own inductive practices. In fact, insofar as the internalist is correct, there is nothing wrong with D’s own inductive practices. (I will return to this below.) It is also worth noting that the Demon can mess with D’s deductive competence as well, albeit given the nature of deductive reasoning this is a bit trickier to bring out (and the argument that does so will be a bit more controversial). Consider the phenomenology of its seeming to you that a deductive inference is valid. This phenomenology is present in cases of short valid inferences whose validity is captured by simple valid inference forms; in these cases the inference seems to you to be

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



SANFORD C . GOLDBERG

valid. But the Demon can mess with the appearances: he could make it the case that a subject experiences the seeming validity of an inference even in cases in which the inference itself is not valid. Thus, the very same ‘seeming validity’ that you experience when you engage in (valid) modus ponens reasoning might be experienced by a subject when she engages in (deductively fallacious) reasoning involving affirming the antecedent. Of course, such an incompetent reasoner is not your internal duplicate: after all, there is a difference in your course-of-experiences, as (s)he reasons (invalidly) from the consequent to the antecedent, whereas you reason (validly) from the antecedent to the consequent. But we can say this: you two are duplicates at least along the dimension of how things seem to each of you regarding the validity of the inference itself. And we might think that, at least by the lights of internalism, this dimension plays a role in the justification of beliefs acquired through deductive inference. That is, we might think that the internalist will hold a view like the following: a belief acquired through inference is internalistically justified (if and) only if (i) the premises are justified, (ii) the subject is justified in regarding the inference as valid, and (iii) the subject acquires the belief in question through her appreciation of (i) and (ii). And we might think that the materials that determine whether condition (ii) is satisfied will be the materials pertaining to how the inference seems to the subject, and in particular whether it seems to her to be valid. But if this is so, then the cost of Demon-proof inferential justification is that the Demon can ensure that a belief acquired through even a very simple deductive inference can be Demon-proof justified even when the inference itself is invalid—indeed, obviously so (from our non-duped perspective). To be sure, the deductive case differs from the inductive case in this particular: given an inductive inference, a difference in the world at which the inference was made (for example, in the regularities that hold in that world) can affect the inductive strength of the inference at that world, whereas the same does not hold for deductive inference. (No difference in the world can affect the validity of a deductive inference—at that world, or indeed at any world.10) For this reason, whereas we can have doppelgangers who differ only in that one of their inductive inferences is inductively strong whereas the other’s is inductively weak (since they might inhabit different worlds with different environmental regularities), we can’t have doppelgangers who differ only in that one of their deductive inferences is valid whereas the other’s is invalid (since any difference in validity will require a difference in premises or conclusion, rendering them different in this way as well). This makes it more difficult to bring out the point about how the Demon can affect the epistemology of deductive inference. More difficult, but not impossible: the point remains that insofar as condition (ii) will need to be part of an account of the epistemology of deductive inference, the Demon clearly can mess with the relations between the appearance and 10

Here I assume that this holds after all context-sensitive elements in the determination of the propositions involved in the inference have been resolved.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

THE COSTS OF DEMON - PROOF JUSTIFICATION



the reality of validity, thereby putting one in a position in which what seem to be valid inferences are no such things. Whereas such a person will not be your doppelganger—your deductive inferences not only seem valid but are valid, we will suppose—even so, such a person will have precisely the same phenomenology you have when presented with a simple valid inference, only that person will experience that phenomenology in the presence of what you and I would regard as an obviously invalid inference. Not a good thing to sanction from an epistemic perspective; but it would appear that the internalist has to sanction this if she wants her justification to be Demon-proof and she endorses the picture above (on which the epistemology of inferentially supported belief involves (i)–(iii)). At this point the internalist might simply return to The Given. Doing so, she might think to herself: OK, one can’t be sure one is unenvatted; OK, one can’t be sure that one isn’t in a Ms.-Appelbaum-type situation (in the midst of a short interval of subjective indistinguishability of what we would pre-theoretically call “someone else’s experience”); OK, one can’t be sure that one isn’t in a Mr.-Fitzgerald-type situation (in which one systematically misapprehends the strength of one’s reasons, or the support they provide for a conclusion, even as all seems right to one); but even in the midst of this chaos, there is always one’s apprehension of the sensorily given. Perhaps this is enough of a basis on which to erect a plausible account of epistemic justification. Or so the internalist might think, as she pursues the materials out of which to construct Demon-proof justification. But things are even worse for the internalist, and this brings me to the third of the three types of demonic intervention I want to consider. For the Demon has read both Sellars and Wittgenstein, and has come up with a devious plot to undermine the plausibility of even so restricted an internalism. To bring this out I present the bizarre case of Ms. Zeta-Jones: ZETA-JONES Ms. Zeta-Jones has the full range of sensory phenomenology: visual images, touch impressions, tastes, sounds, and smells. Indeed, she also enjoys various kinaesthetic impressions, as of her body’s being in various arrangements and moving in various ways. What is more, Ms. Zeta-Jones has made the determination never to judge beyond the phenomenological—she is pure phenomenologist. (Husserl has nothing on her!) At first she thinks: red-here-now; rose-smell-now; cold-feel-here-now; salty-taste-now; etc. But soon she recognizes that in doing so she is taking on a risk she need not take on. For whereas in naming the various sensations she is committing to what all sensations of a given (named) type have in common, she proposes to be safer. So, rather than naming these impression-types with reference to colors, shapes, tastes, etc., she will simply assign each of her impression-types from the various sensory modalities into an equivalence class—thereby prescinding any attempt to say what the members of the class have in common. Thus she groups her sensation into visual color 1, visual color 2, . . . , touch impression 1, touch impression2, etc. Indeed, she goes further and does away with any reference to the sensory modality itself, merely grouping sensations into classes: class 1, class 2, and so forth. That way, the only

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



SANFORD C . GOLDBERG

sensory judgment that she is making is to put her sensory impressions in an equivalence class—without saying what the items in the class have in common, or even the sensory modality through which they were apprehended. She feels safe. But the Demon simply laughs! He realizes that with only a small change in her memory system, he can get her to change her similarity judgments entirely. He begins by bringing this out in connection with her within-modality judgments: he brings it about that what she previously would have put in visual color #27 she now puts in visual color #28 (which we would describe as a nearby color on the color wheel); what she previously would have put in taste #15 she now puts in taste #16 (whereas previously she would have distinguished the ever-so-slightly-saltybitterness of #15 from the mere bitterness of #16; and so forth). And the Demon realizes with a chuckle that with a bit more messing with her memory he can mess with her within-modality categorizations even more drastically: she now puts a smell, which yesterday she would have put in olfactory #15 (what we would call “the smell of freshly baked bread”) into olfactory #278 (“the post-rain smell of worms”), and would now put tactile sensation #6 (which we would call “a sharp pain on the bottom of the foot”) into #111 (“a feather-like touch on the arm”), thinking all the while that she is preserving the categories she has had all along. And when the Demon gets really bored he messes with her memory even more, to get the result that she ends up mixing sensations from different sensory modalities altogether. So class 1 contains some token visual sensations, some token auditory sensations, some token olfactory sensations, some token kinesthetic sensations, etc.; and class 2 contains (different) token visual sensations, (different) token auditory sensations, etc. All in a day’s work.

The upshot, of course, is that Ms. Zeta-Jones’ classes are jumbled, and do not make for intelligible ‘categories.’ The natural interpretation is that, insofar as we assume that the original categories were meant to be equivalence classes defined by sameness-of-sensory-impression-type (if such are even possible11) are now a jumble of arbitrary sensory impressions whose sole ‘similarity’ with the other members of the category is that Ms. Zeta-Jones placed them there. It is natural to think that Ms. Zeta-Jones is not making false judgments at this point—she is not making intelligible judgments in the first place. Hardly an epistemologically sound resting place from which to (resist the Demon and to) construct a notion of epistemic justification. Of course, Ms. Zeta-Jones is not an internal duplicate of yours. But we can imagine someone—call this person D—where D is just like you save that, for some time, D has been slowly expanding or contracting the set of objects to which D applies his/her predicates, albeit in ways that escape D’s own recognition. There are in fact two versions of this: one in which the predicates in question apply to purely phenomenological features of one’s sensory experience, the other involving predicates that are meant to apply to ordinary worldly properties and objects. It is worth developing both; I will take them up in reverse order. Imagine that you and D have precisely the same phenomenological experiences (insofar as it is possible for two distinct people to have the same phenomenological

11

The worry is that there might be a sorities looming. I will waive this here.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

THE COSTS OF DEMON - PROOF JUSTIFICATION



experiences). But imagine that the Demon decides to intervene regarding the nature of D’s external environment: in particular, the Demon acts so as to ensure that the external items to which D aims to be applying his/her predicates are a grab-bag, forming nothing you or I would recognize as an intelligible category of objects.12 The result will be a scenario in which D’s perspective evolves into unintelligibility, where the unintelligibility in question concerns D’s purported judgments regarding external objects and their properties. The Demon ensures this as follows. Keeping D’s own phenomenology fixed to yours (at least insofar as phenomenology supervenes on what is relevantly internal), the Demon rearranges the external world in such a way that there is a slow expansion or contraction in the applications of D’s predicates, allowing the scenario to evolve to the point where D’s applications are no longer intelligible as amounting to judgments in which external objects are judged to fall into recognizable categories. D is aware of none of this, of course. Now this situation, in which the demonic intervention concerns items in the external world, is like the BIV scenario in one respect: as D’s course-of-experience evolves over time, it comes to pass that D systematically misapplies the predicates in her/his language—at least as these are standardly interpreted. But unlike in the BIV case, at a certain time in the evolution of D’s course-of-experience, there is no intelligible relation between D’s experience, his/her application of the predicates, and the items to which (s)he aims to apply the predicates. If the external world is the purported target, D is unintelligible. And yet D is your doppelganger. Now at this point the proponent of Demon-proof justification might think to diagnose the purported unintelligibility of D’s ‘judgments’ as the result of a certain semantic assumption which (it is urged) we should give up. The assumption in question is a content externalist assumption, to the effect that the contents of judgments are fixed by appeal to the external objects and properties to which one purports to apply the referring expressions and predicates of one’s language. The proponent of Demon-proof justification might urge us to reject this assumption, and to replace it with a view according to which the contents of judgments are fixed by states internal to the subject himself/herself. If we do so, then insofar as the Demon cannot affect D’s internal states, the Demon cannot affect the contents of D’s judgments, and so, trivially, cannot render D’s ‘judgments’ unintelligible. To this move I have two replies. First, it would be striking indeed if it turned out that the only way to defend the doctrine of Demon-proof justification would be by endorsing a thoroughgoing internalism about content; given the popularity of externalist views of content, and the hope we might have for thinking that epistemic internalism might be able to accommodate semantic externalism,13 this would surely count as a cost. Of course, there might be internalists around who regard this as a 12 Of course, the category is intelligible as ‘things to which D applies such-and-such a predicate,’ but this is too subjective a basis on which to render all of D’s various predicate-types intelligible. 13 For a clear-headed expression of this, see Brown (2007).

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



SANFORD C . GOLDBERG

result to be embraced;14 but this brings me to my second, and more important, reply. There are reasons to think that the retreat to internalist semantics will not ultimately work: even if (for the sake of argument) we allow that the Demon can’t affect the contents of judgment, still, the Demon can affect the truth, and even the very intelligibility, of ‘judgments’ regarding one’s own phenomenal states. This brings me to the second, purely phenomenological version of the doppelgangers suggested by the case of Ms. Zeta-Jones. Imagine that you and D have precisely the same phenomenological experiences (insofar as it is possible for two distinct people to have the same phenomenological experiences). Now imagine that at a particular time t, both of you experience a certain color (that is, both of you have a certain color sensation). Because you are doppelgangers, the token color sensation involves a token color (or color-quale, if you like) of the very same color type (color-quale-type)—say, the experienced color is that of fire-engine red. Now, given internalism, for any proposition regarding the experienced color itself, and for any doxastic attitude A that one might take towards that proposition, you and D are equally justified in taking A towards the proposition.15 This is an epistemic ‘fact’ holding of you and D, given that you are internal duplicates who are experiencing tokens of the same color-type. But now the Demon acts so as to affect D’s color similarity space—that is, acts so as to affect what colors D regards as similar to what other colors. As a result, there are certain color propositions about her current experience that D finds credible, but, with respect to the corresponding color propositions regarding your current experience, you reject these propositions (or at least remain agnostic). Thus, where c is the phenomenal color-token in D’s experience and c* is the phenomenal color-token in your experience, and where c and c* are of precisely the same shade of the same color, even so, D accepts, of c, that it is slightly orange, whereas you reject, of c*, that it is slightly orange. And this, even though you are internal duplicates at the level of phenomenology. Since only one of you at most can be right, this shows that the Demon can mess with the truth of judgments regarding one’s own phenomenological experiences. But the Demon can do more: he can mess with the intelligibility of those ‘judgments.’ To do so, the Demon need only manipulate D’s memories over time, further warping D’s color similarity space, so that there comes a time at which the divergences in your and D’s dispositions to accept propositions on the basis of a particular color experience differ in more dramatic ways than that just illustrated. For example, if the Demon messes with D’s memory, so that at a certain point D is willing to apply the predicate ‘is slightly orange’ to things that are in fact of the same color as grass is (i.e., phenomenal green to you and me), it will no longer be clear what proposition D is judging. Worse

14 I suspect that both Ricardo and Richard Fumerton are in this category. What most of us regard as an unacceptable retreat from the world they regard as the price for Demon-proofing one’s epistemic home. 15 Here I bracket the fact that we are talking about two distinct color experiences, yours and D’s. I will only raise this where relevant.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

THE COSTS OF DEMON - PROOF JUSTIFICATION



still if the Demon messes with D’s memory even more, so as to induce cross-modal groupings (as when D is willing to apply ‘is slightly orange’ to what in fact is a ticklesensation). In such circumstances, D’s would-be experiential ‘judgments’ are no longer intelligible at all. Yet at least at the level of the phenomenology of experience, you and D remain internal duplicates. Care must be given in handling this sort of argument. For it is clear—and the internalist will no doubt insist—that at the level of explicit judgment the two subjects above are not internal duplicates, as they differ in the explicit judgments that each makes about his or her own color (or other) experiences. This much is true. But a reply is near-to-hand: the ideology of the Given is supposed to be that judgments are justified, or not, on the basis of a mere surveying of sensory experience, where identity at the level of sensory experience is supposed to be sufficient to determine justification. If the internalist wants to give this up, so be it, but then it is no longer clear whether there is a stable resting place for her at this point in the dialectic— which is to say that it is no longer clear whether she has materials with which to construct a notion of Demon-proof justification. One other feature of the phenomenological version of the case of Ms. Zeta-Jones is worth highlighting. In all of the other cases I have presented so far—that of Ms. Appelbaum, Mr. Fitzgerald, and even the first (non-phenomenological) version of the case of Ms. Zeta-Jones—the challenge for the proponent of Demon-proof justification was that her theory predicted that the subject enjoyed such justification, albeit under conditions in which intuition suggests that the subject himself/herself was not in a happy epistemological situation. Thus, I argued that the Demon-proof internalist will have to say that Ms. Appelbaum’s beliefs enjoy Demon-proof justification, even when she is in the midst of a 15-minute interval in which she is utterly detached from her current surroundings (and lacks any evidence she acquired more than 15 minutes ago); the Demon-proof internalist will have to say that Mr. Fitzgerald’s inferentially based belief enjoys Demon-proof justification, even though he is oblivious to his own de facto reasoning competence (which is actually quite poor); and the Demon-proof internalist will have to say that Ms. Zeta-Jones’ perceptual ‘judgments’ about items in her external environment enjoy Demon-proof justification, even when those ‘judgments’ cannot be rendered in a coherent way (at least not in terms of the items in her surrounding environment). In this case, however, the problem is different. It is that, given a subject whose memory and color-similarity space has been manipulated (as in the phenomenological case of Ms. Zeta-Jones), the Demon-proof internalist will be forced to recognize that the Demon can mess with the truth, and even the intelligibility, of the judgments themselves. For this reason, the Demon-proof internalist appears to face a dilemma, according to whether she endorses or rejects the thesis that the experiential ‘judgments’ of the victim of this sort of demonic intervention enjoy Demon-proof justification. If our proponent of Demon-proof internalism accepts this thesis, she in effect is giving up on the idea that The Given has any justificatory role to play. But

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



SANFORD C . GOLDBERG

if she rejects this thesis, she is in effect giving up on the idea that motivates her internalism—the idea that whether one is justified or not is something one can discern (through searching reflection) from the armchair. Neither option should be attractive.

3 OK, the fun and games are over; the Demon exits stage left. Now it is time for the lesson. If internal type-identity determines degree of justifiedness, as Demon-proof justification would have it, then one is only as justified as one’s epistemically worst-off doppelganger. Of course, it’s also true that one’s epistemically worst-off doppelganger is as epistemically well-off as the epistemically most well-off doppelganger. But the point I have been seeking to make in the cases above is that there are a variety of Demon-induced maladies we might suffer from, and so, given the variety of demonic interventions, the result is that enjoying Demon-proof justification would not appear to be such a valuable property to have—at least it is not as valuable as we would normally take doxastic justification to be. This is for a simple reason: even if a belief enjoys Demon-proof justification, the claim that it enjoys such a justification is compatible with the claim that the believer suffers from any number of Demoninduced maladies, and indeed it is compatible with the claim that the belief itself was formed through such maladies. It would seem, then, that whatever internalist (Demon-proof) justification is, it is not the sort of thing that is very valuable from an epistemic perspective. Since I assume that doxastic justification is very valuable from an epistemic point of view, we should reject (Demon-proof) justification as a candidate for doxastic justification. In this penultimate section I develop the case for thinking that this is a lesson for which we have the Demon to thank. Let us begin with the question why we should care about Demon-proof justification. I suspect that those who do value Demon-proof justification will respond in something like the following way: we value such justification because we (have reason to) value assessments of the degree to which a subject’s epistemic house is in order. Their idea is that a belief ’s enjoying Demon-proof justification is tantamount to its having been formed and sustained in a way that conforms to the standards governing the well-orderedness of the subject’s epistemic house. Now, while I agree that we have reason to value assessments of the degree to which a subject’s epistemic house is in order, I submit that this will not do as a rationale for valuing assessments of Demon-proof justification. Simply put, we value assessments of the well-orderedness of a subject’s epistemic house only insofar as we presuppose certain things about what it is to have one’s epistemic house in order—things having to do with the epistemic robustness of the perspective under assessment. Unfortunately, a belief ’s being Demon-proof justified is compatible with the falsity of each of these presuppositions, and so is compatible with the subject’s failing to be epistemically robust in each of these ways. Let me explain.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

THE COSTS OF DEMON - PROOF JUSTIFICATION



What is it to have one’s epistemic house in order? Well, to begin, one must do well with the evidence one has, adjusting one’s doxastic attitudes in an epistemically appropriate fashion as new evidence comes in. Now you might think that a subject whose belief enjoys Demon-proof justification may be said to be represented as doing just this: after all, a belief ’s enjoying Demonproof justification is a matter of its being properly apportioned to the evidence available to the subject (or so I will assume for the sake of argument16). Still, we need only recall the case of Ms. Appelbaum to see that, even if we admit that a belief ’s enjoying Demon-proof justification is a matter of its being properly apportioned to the evidence available to the subject, this is not all that is wanted when we assess how well a belief squares with the evidence. For we typically presuppose that an epistemic subject’s evidence is the evidence she has collected over some significant stretch of time. If we were to find out that there is a positive epistemic status that a subject’s belief can have in virtue of squaring with the evidence available to her at the moment, where this assessment is neutral with respect to how well the belief squares with the evidence this same subject has collected in the very recent past (say, half an hour ago), we would be decidedly unimpressed, I think. What value does the mere squaring with available evidence have, if by hypothesis one’s available evidence need not include evidence one acquired even in the very recent past?17 Indeed, I suggested in Section 2 that it is not even clear whether such a view can say why collecting evidence is a good thing, epistemically speaking: since the Demon can always take away your evidence at a moment’s notice (and replace it with other evidence), we might well wonder why, on any view on which justification is Demon-proof, more evidence is an epistemic good. Nor should the proponent of Demon-proof justification think to respond to this problem by insisting that, when it comes to demonic interventions of the sort in play in the case of Ms. Appelbaum, there are actually two (or more) distinct epistemic subjects inhabiting the single body (in this case, all bearing the name ‘Ms. Appelbaum’). While such a response would save the proposal from the objection just described, it would only shift the bump under the rug from one place to another. After all, if we were to find out that there is a positive epistemic status that a subject’s belief can have in virtue of squaring with the evidence available to her at the moment, where this assessment is neutral with respect to how many distinct epistemic subjects inhabit the very body inhabited by the subject under assessment, we would remain decidedly unimpressed.

16 I note that this was called into question in one of the horns of the dilemma presented in the phenomenological version of Ms. Zeta-Jones. 17 Perhaps the proponent of Demon-proof justification will think to object that this is the best we can do: we can do no better than assessments of how well one’s doxastic attitude squares with the evidence currently available to one. But this begs the question. Assessments of externalist justification purport to do better. To be sure, they do so on the presumption that in fact we are not in a Demon-infested world; but that is another matter.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



SANFORD C . GOLDBERG

There is a second dimension of having one’s epistemic house in order which is not captured by assessments of Demon-proof justification. Namely, one whose epistemic house is in order is one whose (deductive, inductive, abductive) reasoning is relevantly competent. Now you might think that a subject whose inferentially supported belief enjoys Demon-proof justification may be said to be represented in just the desired way. After all, an inferentially justified belief ’s enjoying Demon-proof justification is a matter of its seeming to the subject (on searching reflection) to be properly inferred from well-supported premises. Of course, if the case is like that of Ms. Appelbaum, then it may well be that what seem to the subject to be wellsupported premises are no such thing; but let us waive this. For there is another difficulty facing the proponent of Demon-proof justification. We need only recall the case of FITZGERALD to see that the Demon can affect one’s sensitivity to the de facto competence with which one reasons. Once again, if we were to find out that there is a positive epistemic status that a subject’s inferentially supported belief can have in virtue of its seeming to the subject to be properly inferred from what appear to be well-supported premises, where this assessment is neutral with respect to how competent the inferences actually were (not to mention how well-supported the premises themselves were), once again we would be decidedly unimpressed. Finally, there is a third dimension to having one’s epistemic house in order that is not captured by assessments of Demon-proof justification. One whose epistemic house is in order is one who makes intelligible judgments—judgments that can be ascribed determinate (or at least judgeable) contents. (I am not really sure that this is properly called ‘a dimension’ of having one’s epistemic house in order, since arguably it is presupposed by the possibility of any epistemic assessment at all; but more on this below.) Now you might think that a theory of Demon-proof justification which appeals to The Given can represent its subjects as doing just this. But the case of ZETA-JONES calls this into question. Suppose that the Demon really can mess one up so much that one’s sense of similarity changes radically, albeit imperceptibly, over time, so that, with respect to what purports to be a sensation-based similarity judgment at a particular time, there is no determinate category of sensation-type to which the current sensation-token is being judged to fall under. Insofar as we were to find out that there is a positive epistemic status that such a ‘judgment’ could have, where this assessment is neutral with respect to whether there is a coherent category of sensation-types in terms of which the present sensation-token is being classified, once again we would be decidedly unimpressed. In each of these cases of substantial cognitive deficit, the subject suffering from such a deficit is oblivious to the fact that she is so suffering. For this very reason, I have been supposing that any Demon-proof internalism will yield improper verdicts regarding the epistemic justification enjoyed by the beliefs. This assumption of mine is based on the idea that the cases above will be treated as Demon-proof internalism would treat cases of a BIV who forms (would-be) perceptual beliefs. Those internalists who hanker for Demon-proof justification have always taken pride

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

THE COSTS OF DEMON - PROOF JUSTIFICATION



in the fact that internalism itself can deliver verdicts of ‘justified’ in cases of the BIV who is your internal duplicate; and such internalists have always counted it as a strike against externalist theories that they cannot do so (or can do so only at the cost of some serious and perhaps ad-hoc complications in the theory). Here I have been trying to suggest that this attitude comes at a cost. If one really opts for a Demonproof kind of justification, one will have one’s hands tied in the various cases of demonic intervention described above—only here, verdicts of ‘justified’ do not enjoy intuitive appeal. Now of course the cases of Ms. Appelbaum, Mr. Fitzgerald, and Ms. Zeta-Jones are far-fetched. But internalist-minded epistemologists should not take solace in this fact. For it seems to me that there are all sorts of real-life cases which, while not anywhere near as radical as these cases (!), nevertheless exhibit much more modified versions of these ‘flaws,’ and I think it is perfectly proper for epistemological assessment to assess the extent to which these flaws affect the epistemic ‘goodness’ of the subject’s beliefs.18 For consider: many people suffer from memory losses of various kinds. While these losses are nowhere near as radical as that suffered by Ms. Appelbaum (during the interval in which she is an internal duplicate of yours), still they affect a variety of what we might call the subject’s cognitive competences: her ability not only to recall, but also to judge, to draw inferences, to recognize, to categorize, and so forth. We might hope that epistemic assessment distinguishes between those beliefs formed by people who so suffer and those formed by people who do not— and we might hope that such a distinction is present even in cases in which the memory loss and its effects on one’s own cognitive competences are hidden from the subject’s own powers of discernment. Or (to take another sort of ‘flaw’) consider: if contemporary psychology is to be trusted, people suffer from all sorts of implicit biases which affect our reasoning in a great many cases. We might hope that that epistemic assessment distinguishes between people whose inferentially formed beliefs (say, about the candidates for the job) reflect their implicit biases, and those people whose inferentially formed beliefs do not (perhaps because the people subjected their reasoning to various external checks). And we might hope that such a distinction is present even in cases in which the presence of implicit bias and its effects on one’s own competences are hidden from the subject’s own powers of mental discernment. Finally, consider: given ordinarily fallible memory, it would seem that over time a subject might well expand or contract what she will recognize as falling within a single sensory-kind category. While this might not be sufficient to render her unintelligible—in this respect she is unlike Zeta-Jones from the example above—even so, we might wonder precisely what sensory kind it is that she is judging the sensation-token presently before her mind to be a token of. Insofar as we can’t make sense of this, it still seems that we might well recoil at a 18

this.

Here, ‘goodness’ is meant to be good in the way of epistemology, however we are to understand

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



SANFORD C . GOLDBERG

positive epistemic status that obtains with respect to a (would-be) sensory judgment merely in virtue of how things seem to her at the moment, where how things seem to her at the moment is compatible with some significant fluctuation in the boundaries of the categories invoked in her judgment.

4 The story I have told here is not a new one. Many people (especially those of us inclined towards more externalist theories of knowledge and justification) have suspected that internalist justification is not an epistemically interesting property. My claim in this chapter is slightly weaker, in two ways. First, I have not targeted all versions of internalist justification. Rather, I have restricted my attention to those seeking to honor the New Evil Demon intuition (to the effect that internal duplicates have the same degree of epistemic justification). Second, I have not argued that such a property is utterly epistemically uninteresting. Rather, I have argued that judgments to the effect that a belief enjoys such a property do not capture various things we are typically interested in when we engage in epistemic assessment. If my claim on this score is not entirely novel, perhaps the new illustrations of the Demon’s reach, and the distinctly metaepistemological way the key point itself was made, can succeed in making this point more psychologically vivid than it has been. If so, we have the Demon to thank for that.

References Brown, Jessica. (2004). Anti-Individualism and Knowledge. Cambridge, MA: MIT Press. Brown, Jessica. (2007). “Externalism in Mind and Epistemology,” in Sandford Goldberg (ed.), Internalism and Externalism in Semantics and Epistemology. Oxford: Oxford University Press, 13–34. Brueckner, Anthony. (1992). “Semantic Answers to Skepticism,” Pacific Philosophical Quarterly, 73: 200–19. Dutant, Julien. (Unpublished). “What ‘Gettiered’ means.” Feldman, Richard. (1988). “Having Evidence,” Philosophical Analysis, 83–104. Fumerton, Richard. (1995). Metaepistemology and Skepticism. Lanham, MD: Rowman and Littlefield. Goldberg, Sanford. (2012). “A Novel (and Surprising) Argument against Justification Internalism,” Analysis, 72: 239–43. Sosa, Ernest. (1994). “Philosophical Scepticism and Epistemic Circularity,” Proceedings of the Aristotelian Society, Supplementary Volumes: 263–307.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

9 Acquaintance and Skepticism about the Past Ted Poston

How long does it take you to read this sentence? Did you rely on memory at all in reading that sentence?1 What is the most complex thought you can entertain without relying on memory at all? These questions raise a fundamental epistemological issue concerning our ability to justify our extensive reliance on memory. Nearly every thought relies on memory. Even simple thoughts we entertain in the fleeting present—e.g., ‘green here now’—rely on our apparent memory that the meanings of our terms are constant and that the ‘I’ which now thinks is the same ‘I’ that thought a moment ago. My goal in this chapter is to consider the epistemological problem of how our beliefs about the past can be justified within an acquaintance theory. Fumerton explicitly acknowledges that the problem of justifying our beliefs about the past is the most fundamental epistemological problem (1985: 119), and yet his solution to the problem relies on acquaintance with the quasi-logical relation of making-probable which he strongly suspects is an illusion (1985: 218). I argue that an acquaintance theory does not offer an adequate solution to memory skepticism. At the same time I am not a skeptic and honesty requires a reply to memory skepticism. As Fumerton acknowledges, the problem is stark and the answers are few (1985: 185). I defend another response to memory skepticism which Fumerton rejects. I will argue for an epistemic conservative response to memory skepticism by arguing that the theoretical economy of a conservative epistemology combined with its virtue of actually addressing memory skepticism gives us a reason to accept it. The structure of the chapter is as follows. In the first section I present the problem of skepticism about the past. In the second section I explain Fumerton’s acquaintance

1

It is my pleasure to write an essay honoring Richard Fumerton’s contributions in epistemology. I have long admired the clarity and forthrightness of Richard’s philosophy. Richard brings lucid argumentation and searing honesty to difficult and fundamental philosophical problems. For comments on a previous draft I thank Michael Bergmann, Matthew Frise, Richard Fumerton, and the audience at the 6th Orange Beach Epistemology Workshop.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TED POSTON

theory, his solution to the skepticism about the past, and the problems with his solution. In the final section I explain epistemic conservatism, how it handles skepticism about the past, and argue that it has more virtues than Fumerton’s arguments to the contrary suggest.

1 Skepticism about the Past Skeptical arguments thrive on the distinction between appearance and reality. External world skepticism begins with the premise that our awareness of an external world is indirect. What we are aware of by sensation is not an external world because we can have qualitatively identical sensations in dream experience. The appearance/reality distinction assumes that we have direct awareness to appearances. We might put the appearance/reality distinction in a different form: there is a distinction between what we are directly aware of and what we are not directly aware of. Then we can say that skeptical arguments thrive on the distinction between what we are directly aware of and what we are indirectly aware of. The skeptical challenge is that many of our prephilosophical beliefs wrongly assume that we are directly aware of things which we can only be indirectly aware of: the past, the future, the physical world, other minds, and theoretical entities of science.2 Putting the skeptical issue in terms of the distinction between direct awareness and indirect awareness highlights the role of epistemic internalism in classical skeptical arguments. Internalists about epistemic justification claim that justification supervenes on internal states of a subject, where these internal states are accessible to direct awareness. I bypass a discussion of the nature of internal states or what direct awareness amounts to. Rather we can approach the issues here by focusing on the problem of skepticism about the past, a problem that arises because of the confines of the fleeting present. However one understands the nature of internally accessible states, the problem of skepticism about the past assumes that our evidence for our beliefs about the past is restricted to whatever we are aware of within the fleeting present.3 This way of casting the problem also highlights the nature of the externalist response to skepticism. Externalists need not be concerned at all with the appearance/reality distinction nor the distinction between direct and indirect awareness. Externalists rather claim that knowledge requires only the presence of an appropriate external relation—e.g., reliably formed belief, truth-tracking, lawful connection, or counterfactual dependence—and the absence of defeaters. Given the nature of externalism, it’s possible that one has knowledge of reality even if all one has to go on are appearances. 2

This list comes from Fumerton’s presentation of the logical order of skeptical problems (see 1985: 119). For a discussion on whether episodes of consciousness have temporal depth and if so, for how long see Dainton (2014). Also for a good historical discussion on the notion of the specious present see Anderson and Grush (2009). 3

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND SKEPTICISM ABOUT THE PAST



Put in this way, it illustrates one of Fumerton’s central claims in his criticism of externalist epistemology: its irrelevance to traditional philosophical concerns. As Fumerton explains, Given paradigm externalism, it is not clear that a philosopher qua philosopher is even in a position to speculate intelligently on the question of whether or not we have noninferentially justified belief in any of the propositions under skeptical attack. Because the externalist has reduced the question of what is noninferentially justified to questions about the nature of the causal interaction between stimuli and response, and particularly to the processes of the brain that operate on the stimuli so as to produce the response, the search for noninferential justification would seem to be as much in the purview of the neurophysiologist as the philosopher. (1995: 162–3)

If externalism is true then it can easily handle skepticism about the past. Externalists need not agonize over the limitations of the fleeting present because all that matters for knowledge is the presence of the appropriate external relation and an absence of any defeaters. However, those who find the appearance/reality distinction to be of central epistemological interest, the problem of skepticism about the past is vexing. The problem ties very able philosophers in knots. Consider, for instance, C. I. Lewis’ brief reply to skepticism about the past. Lewis writes, [K]nowing takes place in the epistemological present; a present in which what is sensuously given is surrounded by or embedded in a mass of epistemically pertinent surrogates of past experience, in the form of memories or of the sense of past experience as having been so and so; and that such present-as-past items are capable of being elicited by attention and reflection and brought into relations with one another and with the sensuously given—all without going beyond the bounds of what is genuinely present now. (1946: 331–2)

Lewis, working within confines of the fleeting present, tries to fill it with as many items of past experience as possible to afford an adequate justification of our past beliefs. But his brief remarks on the present surrogates of past experience are not phenomenologically plausible. These ‘present-as-past items’ are best conceived of as present recollections, which afford the only data for justifying our belief in the past. It is implausible that the fleeting present can be filled with enough of these recollections. To be sure, one can have a recollection in the present ‘now,’ but it is doubtful that there is a “mass of epistemically pertinent surrogates of past experience.” Additionally, it is dubious that the attention and reflection that Lewis discusses can be accomplished within the fleeting present. Reasoning, attention, and reflection all occur across a temporal interval that extends beyond the present ‘now.’ Indeed, as Fumerton remarks “without relying on memory . . . we seem to be prisoners of an all too fleeting present that simply allows no time for the kind of reasoning necessary to gain justification” (2006: 128). Once the data are limited to the present ‘now’ the opportunities for resisting a wide-ranging skepticism are stark. To illustrate the grim prospects for adequately addressing skepticism about the past, consider an inductive justification of past beliefs. Consider the belief that “I had

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TED POSTON

oatmeal for breakfast this morning.” What data do I have for this belief within the fleeting present? I believe it. How could this provide some material for an inductive justification of the belief? Perhaps, we have the following argument: 1. I believe I ate oatmeal for breakfast this morning. 2. In the past I’ve largely been right about beliefs of this kind. So, 3. Probably, I ate oatmeal for breakfast this morning. The problem with this argument is premise 2. In arguing for the reliability of one belief about the past it relies on the truth of a host of beliefs about the past. What is the possible justification one could have for premise 2 within the fleeting present? 1. I believe premise 2. 2. In the past beliefs of this kind have been proven right. So, 3. Probably, premise 2. This argument hardly helps. Any inductive argument for skepticism must include a premise about a correlation between past beliefs and present belief which cannot be justified within present ‘now.’ The problem of skepticism about the past gets much worse by severely restricting the kind of content knowable within the fleeting present. A fundamental presupposition in our thinking is that we have an ability to reidentify objects and properties over time. Consider, for instance, the simple thought ‘green here now.’ To the extent this thought differs in content from the bare indexical thought ‘this here now,’ it involves the content that this color is the same color that was present a moment ago.4 To the extent our thoughts differ from this indexical thought, we rely on justified beliefs about the past. Similarly, our belief in an enduring self relies on the past. The problem isn’t fundamentally one of justifying belief in a self that endures throughout one’s lifetime or even sizable temporal segments of it. Rather the fundamental skeptical problem is justifying belief in anything that extends beyond the fleeting present.

2 Fumerton’s Acquaintance Theory and Skepticism about the Past I begin with a brief review of Fumerton’s acquaintance theory. Then I turn to explain two proposals Fumerton considers to address skepticism about the past. Finally, I present several problems with his solution. 4

See Poston (2014a) for a defense of this claim.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND SKEPTICISM ABOUT THE PAST



2.1 Fumerton’s acquaintance theory The acquaintance theory is a theory of non-inferential justification. According to Fumerton, non-inferential justification for believing that p requires that one be acquainted with (i) the fact that p, (ii) the thought that p, and (iii) the relation of correspondence which holds between (i) and (ii).5 One of the central motivations for his acquaintance theory is that it provides a kind of philosophical assurance for the truth of one’s beliefs that is not available on other accounts of non-inferential justification. As he explains, “When one is acquainted with a fact, the fact is there before consciousness. Nothing stands ‘between’ the self and the fact” (1995: 76). To the extent we want to satisfy the distinctive kind of epistemological curiosity philosophers feel, an acquaintance theory may well be the only game in town. A central question for the acquaintance theorist is “what is the nature of acquaintance?” Acquaintance is a relation that holds between a self and a thing, property, or fact (Fumerton 1995: 74). The relation itself is sui generis; it is not capable of analysis into more fundamental constituents. We can learn about this relation by standing in this relation to things, properties, or facts. When one feels an intense pain one is directly acquainted with the pain itself; the pain is there before consciousness. Additionally, Fumerton suggests that one can be acquainted with the relation of acquaintance (1995: 77). Acts of acquaintance occur within the fleeting present. This feature of an acquaintance theory significantly limits the data available for non-inferential justification. We have seen above how stark the problem of skepticism about the past is when one limits the data to what is available in the fleeting present. This feature of the acquaintance theory implies that it is impossible to have noninferential justification for any memory belief. Non-inferential justification requires that one is directly acquainted with the relevant fact. In memory experience the relevant fact lays outside of the fleeting present. Fumerton considers and rejects a disjunctivist move that veridical memory experiences are different mental states than non-veridical ones. This move could enable one to maintain that in veridical memory experience one is directly acquainted with a fact that lies outside the fleeting present. Fumerton rejects this by arguing that the phenomenological similarity between veridical and non-verdical memory experiences makes it implausible to maintain that one is directly acquainted with a past fact (1995: 186). Consequently, on an acquaintance theory one can have at best inferential justification for a memory belief, a justification that must occur within the fleeting present. Furthermore, the limitations imposed by an acquaintance theory raise some puzzling questions. One question is whether we can be acquainted with the fact

5

Fumerton (1995: 75). However, see Poston (2010); Fumerton (2010) for a discussion of acquaintance with similar facts and the issues it poses for an acquaintance theory.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TED POSTON

that a self endures through time. Acquaintance is a relation between a self and a thing, property, or fact. When one is acquainted with a fact, the fact is there present to consciousness—nothing stands between the self and the fact (Fumerton 1995: 76). But is one acquainted with the self? Hume answered “No.”6 Russell held that we are not directly acquainted with a self (1993 [1914]: 81). Fumerton might follow Hume and Russell’s lead, but then the acquaintance theory of non-inferential justification would imply that when one stood in the relation of direct acquaintance to a fact, while nothing stood between that fact and the self, one could never be acquainted with the self. This feature of an acquaintance theory raises a more perplexing question about the distinction between what is the self and what is not, a question Russell says is “a very difficult one” (1993 [1914]: 81). If one can never be acquainted with a self, what is the basis for supporting the distinction between a self and a thing, property, or fact? Another question concerns the basis within the fleeting present for distinguishing between thoughts. Within the fleeting present what is my justification for thinking that the thought “this is green” is distinct from the thought “this is red.” Perhaps, within the fleeting present one can entertain both thoughts together with the judgment that “this is not thus.” But if that isn’t possible, is there any hope for grounding even minimal distinctions between thought contents? I return to these troubling questions below.

2.2 Fumerton’s replies to skepticism about the past The challenges Fumerton’s acquaintance theory faces from skeptical issues about the fleeting present are grave. The conditions on non-inferential justification imposed by an acquaintance theory combined with the severe confines of the fleeting present make it dubious that there is any completely satisfactory response. Fumerton offers two responses to memory skepticism, but he explicitly acknowledges the limitations of these responses.

..

REPLY : SELF - REFUTATION Fumerton’s first response to skepticism about the past occurs in a larger discussion about the charge that some skeptical arguments are self-refuting (1995: 43–53). A crucial distinction for this discussion is the distinction between local and global skepticism. A local skeptic takes for granted some knowledge or justified belief and then argues for a particular kind of skepticism. For instance, a skeptic about theoretical entities of science can take for granted much of our commonsense knowledge to argue that we do not have knowledge about unobservable entities that figure in current scientific theories. A global skeptic, however, does not take anything for granted. This skeptic attempts to argue that we lack any knowledge or justified belief. However, it is difficult to

6

See the Treatise: Book I, Part IV, sec. 6, “Personal Identity” (Hume 2000 [1738]).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND SKEPTICISM ABOUT THE PAST



understand how the skeptic can argue that global skepticism is reasonable since his own arguments for global skepticism are impugned by the global skeptical position. In light of the self-defeating nature of global skepticism, Fumerton remarks that skeptics “have almost always presupposed a kind of unproblematic access to some foundational empirical data and to the legitimacy of the reasoning on which their skeptical conclusions depend” (1995: 50). Skepticism about the past is such an extreme skeptical position that he suggests the self-refutation charge may stick against the skeptic about the past (Fumerton 1995: 49). Thus, Fumerton suggests, only local skepticism which presupposes some unproblematic knowledge, can pose genuine skeptical arguments. However, the charge of self-refutation against the skeptic about the past is not straightforward. In fact, Fumerton says, The charge involved here is not the charge of formal epistemic self-refutation. The claim is not that the skeptic’s conclusion entails that the skeptic has no reason to believe his conclusion. The alleged self-refutation is more subtle. It amounts to the claim that in embracing a skeptical conclusion as a serious philosophical position the skeptic is implicitly engaging in behavior that makes sense only against the backdrop of a set of beliefs that are incompatible with radical skepticism. (1995: 52)

At best, then, the charge against the skeptic about the past is that he is guilty of some kind of inconsistency between his skeptical conclusions and his practical activity. This conclusion does not provide an epistemic reason against an extreme skeptic. Furthermore, it misrepresents the nature of the skeptical position. Fundamentally, skepticism is a view; it is not a position anyone need be arguing for. If we, nonskeptics, think skepticism is false, then we should be able to explain our reasons. Insofar as we embody the skeptic, he is best conceived as someone who holds up the skeptical view with an eye to eliciting our reasons for thinking that it is false. A skeptic, then, is more of a gadfly than a theoretician who aims to maintain a skeptical position. As I read Fumerton, in the end he agrees with this understanding of the skeptical argument and consequently the ineffectiveness of the self-refutation charge. He writes, It may be that the philosopher is interested in and wants a kind of justification that ordinary people do not even think about in their day-to-day lives. The philosophical skeptic may best be construed as telling the philosopher that this kind of justification is unavailable. In every other walk of life people must get used to the idea that they cannot have everything they want, and the skeptic might maintain that it is a kind of perverted optimism to suppose that the kind of justification that would satisfy the kind of curiosity that afflicts the epistemologist is there to be found. (1995: 52)

I conclude that the self-refutation charge does not provide an epistemic reason to discount skepticism about the past.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TED POSTON

..

REPLY : ACQUAINTANCE WITH THE RELATION OF MAKING-PROBABLE The other response Fumerton offers for memory skepticism appeals to acquaintance with the quasi-logical relation of ‘making-probable.’ In his 1985 book Fumerton highlights the special move we make when we rely on memory. He writes,

One is sorely tempted to suggest that the ‘inference’ involved in taking for granted that what we seem to remember having happened, happened is sui generis. It is an inference we make (if only in the sense that when we do seem to remember X we do take for granted X), but it is difficult to see how the ‘rule’ ‘Infer X from your seeming to remember X (ceteris paribus)’ can be subsumed under some more general rule or pattern of inference. (1985: 180)

The skeptical challenge here is made more demanding by Fumerton’s adoption of inferential internalism, the view that “necessarily, S is justified in believing p on the basis of e only if S has justification for believing e and has justification for believing that e makes p probable.”7 Applied to memory experience, this principle requires that we have some justification for thinking the memory experiences make probable that the attested events really occurred. But the only way to do that within the fleeting present is to have non-inferential justification for the principle that memory experiences make probable that the attested events actually occurred. According to Fumerton, “Unless we are non-inferentially justified in believing such a principle we have no (philosophically relevant) justification for believing propositions about the past” (1985: 185). But, at least in his earlier work, he expresses some doubts about this possible solution writing, But it is not at all clear to me that I am noninferentially justified in believing that my seeming to remember doing X makes probable my having done X. I have the uneasy suspicion . . . that we are simply programmed to believe certain things about the past given the relevant memory ‘cues.’ This suspicion . . . is not one I claim to be justified (in the philosophically relevant sense) in holding—it is simply a suspicion I report having. And if it should turn out to be true, it may be that it is impossible to discover any philosophically relevant justification for believing propositions about the past and, hence, the physical world. (1985: 185)

Fumerton’s later philosophical writing continues this theme. He stresses that acquaintance with relation of ‘making-probable’ is the only hope for a traditional foundationalist response to skepticism (2006: 133, 1995: 190). What is the relation of ‘making-probable’ that one can be directly acquainted with? It is a relation weaker than entailment (Fumerton 1995: 190). Unlike an entailment relation, the makingprobable relation is non-monotonic. If E makes p probable then it doesn’t follow that E + E1 makes p probable. Furthermore, this is an internal relation that holds with necessity between a memory experience and its object (Fumerton 1995: 199–201). An internal relation is one that holds between two things in virtue of the intrinsic natures

7

See Fumerton (1995: 85–9) for a discussion of inferential internalism.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND SKEPTICISM ABOUT THE PAST



of those propositions. For example, “is darker than” is an internal relation that holds between the propositions “this is brown” and “this is yellow.” With memory experiences, the claim is that the intrinsic nature of “its seeming to one as if p occurred” implies that “it is probable that p occurred.” Fumerton locates this internal relation in the Keynesian conception of probability. According to the Keynesian, the probability that some p gives to q is not explicable in terms of pure logic, frequencies, or subjective beliefs. Rather the relation of ‘q making probable p’ is an objective, epistemic relation. It can be discovered by reason, but it exists independently of any particular person’s beliefs about it. Fumerton’s remark that the relation of ‘making-probable’ is an internal relation needs to be significantly qualified to get the right result that the relation is nonmonotonic. A memory experience does not always make it probable that its attested event really occurred. I seem to recall going fishing on Monday, but I know it was on Tuesday. Even though memory can make probable its object, the combination of a memory experience with other experiences may not make it probable. Yet standard cases of an internal relation are monotonic. “This is brown” and “that is yellow” implies that “this is darker than that.” The addition of any other information doesn’t change the fact that “this is darker than that.” The solution to this difficulty is to hold that the internal relation of ‘making-probable’ is a relation between one’s entire body of evidence at a time and its object. Strictly speaking, it is not a relation that holds between two atomic propositions; rather it is a three-place relation between a body of evidence, a basing proposition, and a target proposition. A sympathetic rendering of Fumerton’s claim is thus that within the fleeting present, if one has just the apparent recollection that p together with a background body of evidence then one may be acquainted with the fact that, given the background body of evidence, the apparent recollection that p makes p probable.

2.3 Difficulties with Fumerton’s response to memory skepticism The self-refutation charge misrepresents the nature of radical skepticism about the past and, at best, lays a practical charge against ‘the skeptic.’ Fumerton’s second response to skepticism about the past is the most promising. The appeal to acquaintance with the quasi-logical relation of ‘making-probable’ would explain how one may have inferential justification for beliefs about the past. In the following I discuss three problems with this solution: the phenomenological problem, the problem of background evidence, and the content problem.

..

The phenomenological problem is that it is doubtful that one is in fact acquainted with the facts required for justified memory beliefs. Fumerton’s acquaintance theory implies that one can have at best inferential justification for any memory belief. One cannot be directly acquainted with facts that occur outside of the fleeting present. To satisfy Fumerton’s conditions for inferential justification for believing p on the basis of e, one must be justified in THE PHENOMENOLOGICAL PROBLEM

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TED POSTON

believing (i) e and justified in believing that (ii) e makes p probable. Applied to memory this requires that one is justified in believing that (im) one is undergoing a memory experience and justified in believing that (iim) memory experience makes probable its object. (im) is a plausible candidate for non-inferential justification. It requires that you are directly acquainted with the thought that you are undergoing a memory experience, the fact that you are, and the correspondence between the two. (iim) requires that you are directly acquainted with the belief that the memory experience makes its object probable, the fact that it does, and the correspondence between the two. Thus, inferential justification for (e.g.,) the belief that I had oatmeal for breakfast on the basis of the relevant memory experience requires be directly acquainted with six items: (i) the thought that I am undergoing the relevant memory experience; (ii) the fact that I am; (iii) the correspondence between the two; (iv) the thought that this experience makes it probable that I did have oatmeal this morning; (v) the fact that this experience makes it probable; and (vi) the correspondence between the two. The phenomenological problem is that it doesn’t seem that one is acquainted with these facts within the confines of the fleeting present. Fumerton’s account of justification requires that one simultaneously realizes all six conditions such that they are directly present to the mind. Speaking for myself, within the fleeting present, I find it difficult to hold simultaneously all six conditions before my mind. Moreover, condition (v) does not seem to obtain. I have beliefs to the effect that memory experience makes probable its object but I do not seem to be acquainted with the fact that memory experience makes probable its object. Similarly, I don’t appear to be acquainted with the correspondence between the thought that apparent memory makes probable its object and the fact that it does. Fumerton writes about phenomenological difficulties with virtuous probity. One must be scrupulously honest in one’s use of phenomenological appeal. There is nothing wrong with introducing the sui generis, but one must be absolutely sure that one understands that about which one talks. One must be certain that one can isolate in thought a relation of making probable holding between propositions describing current memory states and propositions describing past events before one can in good conscience appeal to such a relation in order to avoid skepticism. And in the end, I strongly suspect that the probability relation that philosophers do seek in order to avoid skepticism concerning inferentially justified beliefs is an illusion. (1995: 218)

If one can be acquainted with the relation of making probable that holds between memory experience and its object then one has excellent reasons for resisting skepticism about the past. But the skeptic asks curious questions and it may well be that the non-skeptics among us cannot answer the skeptic’s reasonable questions. The skeptic does not foist questions upon us that occur out of the blue. Rather, recognizing that knowing is limited to the fleeting present, the skeptic naturally inquires about how we manage this feat.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND SKEPTICISM ABOUT THE PAST



..

THE PROBLEM OF BACKGROUND EVIDENCE We observed above that the relation of making-probable is an internal, non-monotonic relation. Internal relations hold with necessity. E.g., the relation greater than holds between 3 and 2. No additional facts undermine that. Yet additional information can undermine the relation makes probable that holds between a memory experience and the fact that is attested to. We capture the relevance of additional information by conceiving of the relation of making probable as a three-place relation between a background body of evidence, a basing proposition, and a target proposition. Thus, given a suitable background body of evidence, the apparent memory that p makes probable the fact that p. The challenge to Fumerton’s account of how memory beliefs can be inferentially justified is straightforward. To be inferentially justified in believing p on the basis of e requires that one is justified in believing both e and e makes p probable. Yet this latter fact itself requires reference to a suitable background body of information, k such that e+k makes p probable. Once k is added to the evidential situation, Fumerton’s principle of inferential justification requires that one is justified in believing k. But k is a background body of evidence and one is not non-inferentially justified in believing an entire background body of evidence within the confines of the fleeting presence. Moreover, given the fact that inductive inference always requires a background body of evidence, it is not possible to eliminate the role of k.8 We noted above that one may have non-inferential justification for believing that one has a specific memory experience. This provides justification for one of the conditions of inferential justification for a memory-based belief. But if we must add justification for a background body of information then it is not plausible that we are acquainted with the fact that we have this body of evidence at a specific time. This problem afflicted Laurence BonJour’s coherence theory of justification. BonJour required that one was aware of the fact that one had an entire system of beliefs at a time, but he couldn’t see how this could be done. He stipulated the “doxastic presumption,” that one is aware that one has an entire body of beliefs at a time as a condition for questions of justification to be raised.9 With respect to an acquaintance theory, the problem of acquaintance with background beliefs is no less severe. By their nature, background beliefs are not present to consciousness and hence aren’t the kinds of things one is presently acquainted within the fleeting present.

..

The promise and burden of an acquaintance theory of non-inferential justification is that every belief is justified entirely on the basis of the materials available within the fleeting present. I argued that this promise may not be fulfilled since we do not seem to be acquainted with the required truths

8 9

THE CONTENT PROBLEM

See Poston (2014b: ch. 3) for an argument that induction is always against a background of beliefs. BonJour (1985: 101). My own solution to this problem is presented in Poston (2014b: ch. 6, sec 3).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TED POSTON

and, moreover, that facts about probabilities always require a background body of information. The content problem strikes to the core of an acquaintance theory by undermining the possibility of cognitively significant thoughts within the fleeting present. The issues here require some theory to bring into sharper focus. I draw upon David Chalmers’ recent attempt to defend a traditional doctrine of the given.10 The content problem concerns the ability to have cognitively significant thoughts within the fleeting present. Frank Jackson’s thought experiment concerning Mary provides fertile ground for framing the content problem (1982). Mary is a neuroscientist specializing in color vision who knows every physical fact about color vision. Mary has been raised in a black and white room and has never seen a red tomato in normal conditions. One day, Mary leaves the black and white room and sees for the first time a ripe tomato. Upon having this experience, Mary learns something new; she learns what it is like to see to see a red object. That is, she learns about the phenomenal character of typical red experiences. What is the content of Mary’s new knowledge? To answer this question Chalmers distinguishes among several types of phenomenal concepts (2010: 254–60). When Mary steps outside the monochromatic room and attends to her new experience, her experience instantiates the property of phenomenal red, R. Chalmers distinguishes two relational phenomenal concepts about R. First, there is the community relational concept, redC. This concept indicates “the phenomenal quality typically caused in normal subjects within my community by paradigmatic red things” (Chalmers 2010: 255). The second type of phenomenal concept is the individual relational concept, redI. This concept indicates “the phenomenal quality typically caused in me by paradigmatic red things” (Chalmers 2010: 255). RedC and redI are distinct concepts. An abnormal subject may have red-green color inversion in which case her concept redI picks out a different phenomenal quality than redC denotes. The phenomenal property R can also be picked out by using a demonstrative concept, denoted by the phrases ‘this quality’ or ‘this sort of experience.’ Let us refer to this demonstrative concept as thisE. This demonstrative concept picks out whatever quality is present on the specific occasion. Chalmers claims that each of these concepts fixes the reference to phenomenal redness relationally, either through external objects or acts of ostension. He then argues that there is a fourth phenomenal concept that picks out phenomenal redness “directly in terms of its intrinsic phenomenal nature” (2010: 256). He terms this a “pure phenomenal concept.” Chalmers argues that there are pure phenomenal concepts by reflection on the case of Mary. When Mary steps outside the black and white room she learns that red experiences have “such and such a quality.” She learns that red experiences cause experiences of such and such quality and she learns that the quality now extending is

10

Chalmers (2010: chs. 8 and 9). The following draws upon material from Poston (2014b: ch. 5).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND SKEPTICISM ABOUT THE PAST



such and such. Chalmers refers to this as “Mary’s ‘such-and-such’ concept” (2010: 256). This is Mary’s pure phenomenal concept R. This concept R picks out the phenomenal quality R. Chalmers then argues that the concept R is distinct from the concepts redC , redI, and thisE. His argument uses cognitive significance tests for difference between concepts. When Mary steps outside the monochromatic room and sees a red object in normal conditions she gains the following beliefs: redc=R, redi=R, and thisE=R. The first two beliefs are cognitively significant. She learns that the quality typically caused in normal subjects in her community by paradigmatic red things is R. Similarly, for the second identity, Mary learns that the quality caused in her by paradigmatic red things is R. However, on an acquaintance theory the first two beliefs are not non-inferentially justified since the content of such beliefs involves how things appeared in the past. A crucial question for an acquaintance theorist is whether the belief that thisE=R is cognitively significant. The belief expressed by this identity is the claim that “the quality she is now ostending is such-and-such” (Chalmers 2010: 257). Chalmers needs to successfully argue that this thought differs in content from the trivial thought that this quality is whatever it happens to be. A problem for the cognitive significance of thisE=R is that the lifetime of the pure phenomenal concept R is restricted to the lifetime of the experience that constitutes it (Chalmers 2010: 272). R is a non-relational concept that picks out the very quality exemplified in the experience. Because of its non-relational character, there are no natural persistence conditions of concept R. Moreover, the concept R is not what is expressed in public language. The terms in a public language, e.g. ‘red,’ pick out relational concepts. Thus, the kind of belief expressed by the sentence ‘R is phenomenal red’ is not a belief that is non-inferentially justified within the fleeting present. The consequence that direct phenomenal beliefs exist only within the lifetime of the relevant experience puts significant pressure on Chalmers’ insistence that direct phenomenal beliefs are cognitively significant (2010: 282). Chalmers argues for the significance of these beliefs by claiming that a direct phenomenal belief constrains the class of a priori epistemic possibilities (2010: 282). His thought is that when Mary forms the belief that ‘thisE=R’ her belief is false at all worlds (considered as actual) in which Mary is not experiencing phenomenal redness. Mary’s belief is cognitively significant because her new experience significantly constrains the possible worlds prior to having that experience. For instance, prior to leaving the black and white room it was epistemically possible that Mary form the belief that ‘thisE=G,’ but now, having had the relevant experience, that thought is no longer epistemically possible.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TED POSTON

This sounds as if direct phenomenal beliefs are cognitively significant, but appearances are deceptive. The direct phenomenal beliefs exist only within the fleeting present, and so they only constrain epistemic possibilities within the present ‘now.’ Mary’s thought that ‘thisE=R’ constrains epistemic possibilities only for a fleeting moment; let it pass and it is an epistemic possibility that a similar thought ‘thisE=G’ is true. What makes this an epistemic possibility is that judgments of identity, similarity, and difference are not direct phenomenal beliefs. When one thinks that ‘R is phenomenal red’ one identifies a present quality with a relational quality. One’s evidence that this identity is true relies on a host of background information that is not contained within the fleeting present. If the relational quality is redC then one needs evidence both that this quality exists (i.e., that the members of your community do not experience different properties when faced with red things) and that the majority of other people experience the same quality as you do. There is a real puzzle here about how to understand the significance of Mary’s knowledge when she leaves the black and white room. She gains a new belief ‘thisE=R’ which exists for the present ‘now’ of her experience. Suppose Mary stares at the ripe tomato thinking “wow, this is what it’s like.” On Chalmers’ account, this is a direct phenomenal belief only within the fleeting present. If Mary were to look away and attend to another red object, her belief that ‘thisE=R1’ would be a different direct phenomenal belief. But her belief that ‘R=R1’ is not a direct phenomenal belief. If Mary’s knowledge is restricted to direct phenomenal beliefs then she has no way to knowingly identify or compare any pure phenomenal qualities. To the extent that Mary can knowingly identify and compare phenomenal qualities she must rely on a wealth of information that is not contained within the fleeting present. She relies on her beliefs that memory is reliable, that sensations do not change faster than she realizes, and that other people experience the same sensations. None of these beliefs are given in the fleeting moment. But to the extent Mary gains new knowledge, she relies on these beliefs. If you take these beliefs away and consider only what Mary knows in the fleeting present via direct acquaintance with a phenomenal quality, the best one can do is get a belief like ‘thisE=R.’ But, this belief has little cognitive significance. To the extent it constrains epistemic possibilities it constrains them momentarily. The space of epistemic possibilities contracts and expands with every passing moment. Mary cannot hook up this new belief with any other beliefs, at least apart from coherence considerations. She cannot, for instance, reason that ‘phenomenal redness is R’ because ‘phenomenal redness’ is a public language term. The conclusion I draw is that while Mary can latch onto the property of phenomenal red this property can have little cognitive significance for her. She cannot use her latch to the property to compare or contrast that property to other phenomenal properties across time; at least, not on the basis of what an acquaintance theory of non-inferential justification allows. Since this conclusion holds for the ideal case of phenomenal properties, it holds, mutatis mutandis, for thoughts about other properties.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND SKEPTICISM ABOUT THE PAST



3 Epistemic Conservatism and Skepticism about the Past I have argued that skepticism about the past poses grave problems for an acquaintance theory. Yet I’ve also pressed the line that skeptical issues about the past are natural. Concerns about skepticism about the past arise from the simple observation that awareness occurs within the fleeting present. How is it that we know so much about the past when the data that we have are limited to what we are aware of in the fleeting present? This is a question whose simplicity requires an answer, and yet the answers given are quite few. There is in Fumerton’s epistemological writings an alternative answer to this question which I think is the more adequate of the two. I turn to that now. Fumerton’s first solution to the problem of skepticism about the past occurred within a distinction between global and local skepticism. Global skepticism is a complete, wide-ranging skepticism about all propositions. Local skepticism is a limited form of skepticism. Skepticism about the past is a very strong form of local skepticism in which the range of propositions under skeptical attack are all those that lie outside what is knowable within the fleeting present. The propositions within the fleeting present are treated as immune to challenge for the purposes of the skeptic’s argument. These propositions are presuppositions which are taken for granted for the purpose of skeptical argument. Fumerton recognizes that some propositions are simply taken for granted. These propositions are not amenable to rational defense independently of any presuppositions. Fumerton writes, The vast majority of skeptics . . . have actually presupposed knowledge or justified belief with respect to some class of propositions. Skeptics in the empiricist tradition almost all seemed to presuppose unproblematic access to occurrent mental states. Indeed, the presupposition was so complete that one rarely even finds the Modern philosophers raising the question of whether or not one can know that one is in a certain subjective mental state. Furthermore, almost all skeptics seemed to presuppose knowledge of at least logical relations. They seemed to presuppose that one can recognize or ‘see’ contradiction, at least some simple necessary truths, and at least some simple entailments. (1995: 31)

The two presuppositions are that we can correctly identify our own subjective mental states and that we have the ability to identify simple logical truths. The latter of these presuppositions is fundamental to any rational argument. Because the ability of an argument to convince takes for granted that one can appreciate simple logical truths, it’s impossible to present an argument that one has this ability without begging the question. Suppose it’s in doubt whether one’s memory is reliable. An argument for the reliability of memory that depends on the reliability of memory will not help assuage one’s doubts. Similarly, if it’s in doubt that one has an ability to appreciate simple logical truths, an argument that presupposes one has that ability will not help.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TED POSTON

The upshot of this is that the activity of philosophy presupposes that we have certain abilities (Fumerton 1995: 52). What possible justification could there be for these presuppositions? The line that Fumerton considers is that there is not a justification for these fundamental presuppositions. Rather these propositions lack epistemic authority (i.e., they are not known or justifiedly believed), and it is only by taking them for granted that we can make sense of epistemic authority. If we presuppose that we can identify simple logical truths and identify our subjective states then we can make sense of classical representative realist arguments that we have some epistemic authority for believing that there is an external world. Similarly, if we presuppose that memory is reliable then we can make sense of arguments for the reliability of induction. But it is a consequence of this view that the ultimate foundations for knowledge and justification rest on unjustified presuppositions.11 Is there a way to resist this conclusion? I think there is. A resolution to this problem must take seriously that some propositions lack defenses by way of evidence and arguments designed to rationally persuade someone in doubt of those propositions. These propositions are not merely presupposed for the purposes of inquiry. It is not as if people think “let’s suppose that memory is reliable for the purposes of argument and let’s see what follows,” or “let’s take it for granted that we have an ability to intuit simple logical truths, and see what follows.” Rather these fundamental presuppositions are things that people actually believe. We believe that we can identify simple logical truths, believe that we can reidentify objects over time, believe that memory is reliable, and so on. These beliefs we take to be confirmed over and over again in common experience, but reflection can bring us to recognize that there is no confirmation apart from already presupposing the truth of those beliefs. An argument that one can reidentify objects over time must not presuppose that we can identify properties of subjective experience over time. That would beg the question. An epistemically conservative position takes seriously this epistemic predicament. An epistemic conservative holds that the mere fact that a subject believes a proposition gives that proposition some positive level of epistemic merit. This positive level of merit is weak. It is not strong enough to warrant assertion or use in reasoning. But this merit can be combined with other merited propositions to generate justification strong enough for use in warranted assertion and reasoning.12 The conservative must provide a plausible story of how conservative merit can be combined to generate justification but this story is possible to tell (see van Cleve 2011; Poston 2014b). There are several arguments for epistemic conservatism. One appeals to the fact that people are conservative. When we are trying to figure out what to believe, we follow the maxim of minimal mutilation (Quine 1990: 14). This is followed by nearly everyone from the simplest folk to the most sophisticated theoretician. Another 11 12

Cf. BonJour’s doxastic presumption BonJour (1985: 101). See McGrath (2007); McCain (2008); Poston (2014b) for recent defenses of epistemic conservatism.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND SKEPTICISM ABOUT THE PAST



argument for epistemic conservatism argues that it is required to avoid wholesale skepticism.13 I’ve argued that some presuppositions are so fundamental to rational belief that it’s not possible to give arguments or evidence in favor of them. If these presuppositions lack any epistemic authority by themselves, it is hard to see how one should resist skepticism. Finally, another argument for conservatism appeals to the egocentric nature of epistemic inquiry. It is reasonable for a person to believe what is supported by her perspective. But a person’s perspective relies on some unquestioned beliefs, and if those beliefs fail to have any epistemic authority it’s difficult to see how reasonable belief is egocentric.14 In addition to those arguments, there is an elegance argument for epistemic conservatism. A central goal of inquiry is to construct theoretically elegant accounts of some phenomena. In normative epistemology, we aim for a theoretically elegant account of knowledge and justification. Elegance is a property of a system in virtue of which it explains a great deal by way of simple postulates. Newton’s theory of motion by way of the four laws is a prime example of an elegant system. Fumerton himself appeals to elegance as a reason for an acquaintance theory, arguing that “one of the great virtues of an acquaintance theory is that it can provide a unified account of non-inferentially justified belief and knowledge” (1995: 199). But as we’ve seen, Fumerton’s acquaintance theory is unable to explain how we can have any knowledge or justification that is not available within the fleeting present. By contrast, an epistemically conservative position combined with a coherentist dimension has greater elegance than an acquaintance theory. Like an acquaintance theory, it offers a single explanation of both a priori and a posteriori knowledge. Knowledge, on a conservative coherentist account, is ultimately justified true belief that does not rely on any false presuppositions. Knowledge, whether a priori or a posteriori, is the same. Explanatory coherentism relies on the notions of belief and explanation, of which we have an adequate grasp. In contrast, an acquaintance theory requires unjustified background assumptions and a problematic conception of the internal relation of making-probable. This brief presentation is inadequate to defend the view. However, I present it as a viable option to address a fundamental epistemic problem within a commitment to epistemic internalism. We’ve seen that options for addressing skepticism about the past are quite limited. Yet honesty requires either a skeptical response or another view. I am not a skeptic. I take the fact that I believe memory to be reliable to be part of an overall justification of the fact that memory is reliable. Within the fleeting present, background beliefs together with coherence considerations generate 13 See Fumerton (2008) for a discussion of this argument. I take Fumerton’s discussion here to be more subtle and slightly more sympathetic to epistemic conservatism than an initial reading suggests. See, for instance, Fumerton’s closing remarks on the relationship between Keynesian probability and epistemic conservatism. One might be tempted to think that an underlying conclusion of Fumerton’s article is that one must brace oneself for skepticism if one completely rejects epistemic conservatism. 14 See Poston (2014b) for a defense of this argument.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



TED POSTON

justification. We are not aware of all these beliefs, but, on my mentalist view, justification does not require current access to all the justifiers. What we are aware of within the fleeting present is only a small part of all the mental resources we have to go on. If those resources are good then justification is present. This solution is not ideal. We would like to have it that every belief has a proper grounding in some distinct justified belief that is available within the fleeting present. But that is not possible. Some propositions are part of the background. Some propositions function as hinges on which the doors of inquiry turn (see Wittgenstein 1969). Either those propositions get justified by the fact that we believe they are true or they don’t. I honestly do not see how apart from conservatism, these propositions get any justification.15 We can say that we have an entitlement to these fundamental presuppositions, but plausible defenses of the entitlement position amount to an epistemically conservative position. Philosophers who defend entitlements seek to trim those propositions that we have entitlements to those that figure in skeptical arguments. But I don’t see how this is not anything but reverse engineering. In contrast, the epistemic conservative takes any proposition to have some basic merit simply in virtue of being believed. The contours of this position are not defined explicitly to limit the skeptical assault. If you find yourself not believing those fundamental presuppositions, then those propositions have no merit for you. In such a case, your epistemic position is stark.

4 Conclusion The problem of skepticism about the past is as difficult as it is easy to appreciate. Knowing occurs with the fleeting present yet the materials available within the fleeting present are quite limited. I’ve argued that an acquaintance theory faces insurmountable problems with skepticism about the past. The only way out of these problems for an acquaintance theorist is to make certain presuppositions that are not justified on an acquaintance theory. This move opens the door for an epistemic conservative epistemology, which takes seriously the idea that the fact that people believe a proposition bestows on it some positive level of epistemic merit. I have argued that conservatism provides a more elegant solution to the problem of skepticism about the past. It is the only simple and effective solution to the positions that take seriously the bounds of the fleeting present.

References Andersen, Holly and Rick Grush. (2009). “A Brief History of Time-Consciousness,” Journal of the History of Philosophy, 47: 277–307. 15

Of course, externalist views would provide justification of these kinds of propositions if the relevant external fact is present. But, at this point, I am working within an internalist conception of justification.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

ACQUAINTANCE AND SKEPTICISM ABOUT THE PAST



BonJour, Laurence. (1985). The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press. Chalmers, David. (2010). The Character of Consciousness. New York: Oxford University Press. Dainton, Barry. (2014). “Temporal Consciousness,” Stanford Encyclopedia of Philosophy (Spring), Edward N. Zalta (ed.), at . Fumerton, Richard. (1985). Metaphysical and Epistemological Problems of Perception. Lincoln: University of Nebraska Press. Fumerton, Richard. (1995). Metaepistemology and Skepticism. Lanham, MD: Rowman and Littlefield. Fumerton, Richard. (2006). Epistemology. Malden, MA: Blackwell. Fumerton, Richard. (2008). “Epistemic Conservatism: Theft or Honest Toil?,” in Tamar Szabo Gendler and John Hawthorne (eds), Oxford Studies in Epistemology (Vol. 2). Oxford: Oxford University Press, 63–86. Fumerton, Richard. (2010). “Poston on Similarity and Acquaintance,” Philosophical Studies, 147: 379–86. Hume, David. (2000 [1738]). A Treatise of Human Nature, edited by David Fate Norton and Mary J. Norton. New York: Oxford University Press. Jackson, Frank. (1982). “Epiphenomenal Qualia,” Philosophical Quarterly, 32: 127–36. Lewis, Clarence Irving. (1946). An Analysis of Knowledge and Valuation. Chicago, IL: Open Court. McCain, Kevin. (2008). “The Virtues of Epistemic Conservatism,” Synthese, 164: 185–200. McGrath, Matthew. (2007). “Memory and Epistemic Conservatism,” Synthese, 157: 1–24. Poston, Ted. (2010). “Similarity and Acquaintance: A Dilemma,” Philosophical Studies, 147: 369–78. Poston, Ted. (2014a). “Direct Phenomenal Beliefs, Cognitive Significance, and the Specious Present,” Philosophical Studies, 168: 483–9. Poston, Ted. (2014b). Reason and Explanation: A Defense of Explanatory Coherentism. New York: Palgrave MacMillan. Quine, W. V. (1990). In Pursuit of Truth. Cambridge, MA: Harvard University Press. Russell, Bertrand. (1993 [1914]). Our Knowledge of the External World. London: Routledge. Van Cleve, James. (2011). “Can Coherence Generate Warrant ‘Ex Nihilo’? Probability and the Logic of Concurring Witnesses,” Philosophy and Phenomenological Research, 82: 337–80. Wittgenstein, Ludwig. (1969). On Certainty. New York: Harper and Row.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

Skepticism and Circularity

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

10 On Metaepistemological Scepticism Duncan Pritchard and Christopher Ranalli

1 Introduction Since at least the mid-1990s, there has been a wave of ‘anti-externalist’ replies to externalist responses to scepticism.1 Richard Fumerton has been one of the most influential figures in this wave, and alongside Barry Stroud his contributions to this debate are the most well-known. Indeed, one often finds Fumerton and Stroud being jointly identified as defenders of a position known as ‘metaepistemological scepticism’ (for example, see Bergmann 2008; Fumerton 2006: 181–2; Pritchard 2005: 208–13). In this chapter, we seek to revise this tradition, for although there are important similarities between their views, there are also important differences. Finding out what those differences are will help us to better appreciate their individual contributions. In particular, we focus on three important points of contact, which we can summarize as follows: (1) Externalist theories permit a kind of non-inferential, perceptual justification and knowledge of the external world, but it’s not sufficient for philosophical assurance—that is, it’s not ‘philosophically satisfying noninferential justification’ (Fumerton 2006: 190). (2) Direct acquaintance with facts in some domain D is sufficient for philosophical assurance about D, and can thereby be appealed to explain how knowledge in D is possible.2

Note that throughout this chapter, when we refer to ‘externalism’ we will have epistemic externalism in mind. 2 More specifically, Fumerton maintains that S has ‘philosophically satisfying’ non-inferential justification to believe that p iff S believes that p, p is true, S is acquainted with the fact that makes p true, and S is acquainted with the correspondence relation between p and the fact that makes p true. 1

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



DUNCAN PRITCHARD AND CHRISTOPHER RANALLI

(3) We don’t have direct acquaintance with facts about the external world (see Fumerton 2014: 121, 2006). Fumerton accepts each of (1)–(3). However, while Stroud accepts (1), interestingly he disagrees with Fumerton on both (2) and (3). The interesting disagreement between them, then, is that while Fumerton thinks that it’s at least possible that direct acquaintance with facts about the external world yields philosophical assurance, Stroud does not think this. Put another way, if only it were possible for us to be directly, perceptually acquainted with facts about the external world, then we would be in a position to achieve a philosophically satisfying response to the sceptical challenge. In Section 2, we provide an exposition of Fumerton’s central argument against externalist responses to scepticism, and the role of philosophical assurance in that argument. In Section 3, we present Stroud’s anti-externalist argument, and along the way review and reject two interpretations of Stroud’s argument in this regard. In Section 4, we return to the contrast between Stroud and Fumerton.

2 Fumerton on Metaepistemology and Scepticism According to Fumerton, one of the principal concerns of his book Metaepistemology and Skepticism is to: explore the implications of accepting various accounts of internalism and externalism for the way in which one should understand and respond to the traditional skeptical challenges. (Fumerton 1998: 906)

In particular, Fumerton argues that ‘there is something wrong with paradigm versions of externalism’, because of the ‘ease with which externalists can and should ignore skeptical challenges at all levels’ (Fumerton 1998: 906). Here we are exposed to Fumerton’s distinctive strand of metaepistemological scepticism, which targets externalist responses to the ‘traditional skeptical challenges’, such as the problem of the external world, the problem of other minds, the problem of induction, and so on. According to Fumerton, one of the central problems facing externalist responses to these kinds of sceptical challenges is that the externalist ought to permit a question-begging track-record argument in order to support their second-order beliefs about the epistemic credentials of their first-order beliefs. But as Fumerton himself summarizes the problem with these kinds of arguments: You cannot use perception to justify the reliability of perception! You cannot use memory to justify the reliability of memory! You cannot use induction to justify the reliability of induction! Such attempts to respond to the skeptic’s concern involve blatant, indeed pathetic circularity. (Fumerton 1995: 177)

The basic idea, then, is that if externalism were true, it would be possible for one to know that their perceptual faculties were reliable on the basis of a track-record

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

ON METAEPISTEMOLOGICAL SCEPTICISM



argument, refuting the sceptical claim that it’s not possible for one to know that their perceptual faculties are reliable, because for all one knows one is just dreaming or a brain in a vat, situations in which one’s perceptual faculties are unreliable. But Fumerton thinks that these kinds of track-record arguments are at best a reductio of externalism (see also Fumerton 2006: 179–80). Here one might wonder what’s missing from the externalist’s explanation of how we can know that our perceptual faculties are reliable. Fumerton’s suggestion is that what’s missing is some kind of philosophical assurance that those faculties are reliable (for example, see Fumerton 2006). After all, we might appeal to the alleged fact that there have been several occasions in the past where whenever we had a perceptual experience as of p, p turned out to be true, and whenever we have a perceptual experience as of q, q turned out to be true, and so on. From this, we could then formulate a track-record argument for the conclusion that perception is reliable. But couldn’t we have just been dreaming that p was true, and all the while we were having the perceptual experience as of p nevertheless, and so on for the rest of our perceptual experiences? In short, how can we assure ourselves that this is not so, among countless other possibilities? In order to make good on this kind of question, Fumerton’s suggestion is that the philosopher seek philosophical assurance, and philosophical assurance, we are told, is not what the externalist’s explanation gives us. Here we are led to at least three questions, which we will consider in turn: what is philosophical assurance; how can one get it; and why can’t externalism allow one to get philosophical assurance? Consider the first question. According to Fumerton, philosophical assurance is the state in which one’s philosophical curiosities are satisfied, because one has an assurance of the truth of the claim in question. For comparison, consider a child who asks her parents ‘why is the sky blue?’ As Fumerton explains, one might answer with something like a school-level scientific explanation, appealing to wavelengths and the fact that wavelengths of that kind look blue to people with normal colour vision. But then the child might persist: ‘why do wavelengths of that sort look blue to us?’ The child’s curiosities haven’t been satisfied. And this would continue to be so if the parent got to the point in the exchange with an answer like: ‘That’s just how it is’.3 What the child seeks is some kind of explanation—an explanation which should no longer yield any more ‘why?’ questions. In this way at least, philosophers are like children. We want our philosophical curiosities satisfied, just as children want their curiosities satisfied. In particular: The epistemologist . . . wants to know why we can legitimately conclude that a certain way of forming belief is legitimate, and the epistemologist’s philosophical curiosity isn’t going to be satisfied by being told at any stage of the game that it just is. (Fumerton 2006: 184)

3

This example is borrowed from Fumerton (2006: 184–5).

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



DUNCAN PRITCHARD AND CHRISTOPHER RANALLI

So what is sufficient for philosophical assurance? In general, following Fumerton, we might say that if S’s philosophical curiosities are settled with respect to p, then S has no further philosophical reason to ask ‘why p?’ Philosophical assurance simply is what one gets when one’s philosophical curiosities are satisfied. So, to ask what’s sufficient for philosophical assurance with respect to p is to ask what’s sufficient for settling one’s philosophical curiosities with respect to p. And to ask whether one’s philosophical curiosities with respect to p are (or ought to be) settled is to ask whether one has any further philosophical reason to ask ‘why p?’ According to Fumerton, the relation of acquaintance is sufficient for philosophical assurance, because it’s the kind of epistemological relation that brings the truthmaker of one’s belief before one’s mind.4 On this view, acquaintance with p ought to provide one with no further philosophical reason to ask ‘why p?’ In particular, acquaintance is a sui generis relation of ‘direct awareness’ of the truth-maker for the corresponding belief that p.5 What’s important about acquaintance is that (i) S’s being acquainted with the fact that p, (ii) their belief that p, and (iii) the correspondence relation between them, is jointly sufficient for non-inferential justification that satisfies philosophical curiosity—that is, it generates philosophical assurance.6 On Fumerton’s view, then, all it takes for us to acquire philosophical assurance for our beliefs about the external world is for us to be acquainted with facts about the external world. Once this question is settled, the philosophical problem of the external world will be resolved. Can we be acquainted with the truth-makers of some of our beliefs? And if we can, what is the range of our acquaintance? As it turns out, Fumerton thinks we can acquire philosophical assurance for a certain limited range of propositions, because we can be acquainted with the truth-makers of those propositions. For example, consider the propositions that I’m in pain now. On Fumerton’s view, when I’m in pain, I can come to know that that proposition is true because ‘I am directly aware of the pain itself—the very truth-maker of my belief ’ (Fumerton

4

For an explicit statement of this, see Fumerton (2006: 188). An interesting question here is what Fumerton has to say about objections to his acquaintance theory from proponents of the ‘Myth of the Given’, such as BonJour (1985) and McDowell (1994). Roughly put, how could a non-representational (or more narrowly, non-propositional) relation like acquaintance yield justification for one’s beliefs, a representational state (or more narrowly, a propositional attitude)? It seems like it could do so only by entailment, or probabilification, each of which seems to be a relation that can hold only between representational states. Note that this kind of objection doesn’t obviously arise for proponents of an acquaintance theory, such as Campbell (2009), and perhaps also Travis (2004, 2005), who maintain that acquaintance is a non-representational relation, but that it’s epistemic role is to give one non-propositional knowledge of external things. 6 See Fumerton (1995: 74–6, 2006: 187–90) for his work on acquaintance and its role in philosophical assurance. Note that here we represent Fumerton as thinking that acquaintance is sufficient for philosophical assurance, but elsewhere he seems to think it’s necessary as well (e.g., Fumerton 2014: 123). Note, however, that some other philosophers, such as Campbell (2009), support a different theory of acquaintance. 5

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

ON METAEPISTEMOLOGICAL SCEPTICISM



2006: 189). But now consider a proposition about the external world, such as that I have hands. Can I be acquainted with the fact that I have hands? According to Fumerton, I cannot, because since one can possess in hallucinatory experience the same justification that one possesses in veridical experience for believing truths about one’s physical environment, I don’t think one can be directly acquainted with facts about physical objects. (Fumerton 2014: 123)

As we have interpreted Fumerton, acquaintance with p is sufficient for philosophical assurance about p. The prospects of acquiring philosophical assurance for our external world beliefs cannot therefore come from acquaintance with facts about the external world. But that leaves open whether it is strictly necessary. Can we acquire philosophical assurance for our beliefs about the external world even if we cannot be acquainted with facts about the external world? A second question is whether or not an externalist can integrate acquaintance within their stock of noninferential reliable processes. The thought here is that if what explains the epistemological properties of acquaintance are properties that can be explained along externalist lines, then it’s not clear what would prevent Fumerton from being committed to the thesis that either acquaintance is not sufficient for philosophical assurance, or else that externalists can accommodate the requirement that we can acquire philosophical assurance for our beliefs about the external world. Consider the second question. At least as far as Fumerton (e.g., 2006: 188) is concerned, a stable causal, reliable, or modal relation between one’s state of mind and a fact is not identical with acquaintance with that fact, even if it is sufficient for noninferential justification to believe that fact. If this is right, we might wonder why acquaintance is the kind of non-inferential relation that can generate philosophical assurance, whereas the externalist’s explanation of our non-inferential perceptual processes, for example, cannot generate philosophical assurance. After all, that the relation of acquaintance is sui generis does not entail that the explanation of its epistemological properties is sui generis as well—that the epistemological properties of acquaintance admits of no further explanation. The closest we get to an explanation of why acquaintance with facts enables it to yield philosophical assurance is that it’s a relation in which the fact in question is before the subject’s mind, and that when a fact is before the subject’s mind, and the subject’s belief which represents that fact is also before their mind, this ought to settle the question of its truth. Fumerton’s thought here is that there is simply nothing more to want in terms of assurance of truth than having the fact of the matter ‘before one’s consciousness’. As he asks: ‘What more could one want as an assurance of truth than the truth-maker before one’s mind?’ (Fumerton 2006: 189). Now one might think that none of this implies that the externalist is barred from thinking that veridical perceptual experience is not the kind of relation that brings the fact in question before the subject’s mind, settling the question of its truth. On the

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



DUNCAN PRITCHARD AND CHRISTOPHER RANALLI

other hand, Fumerton gestures towards this sort of position. In an exposition of the nature of non-inferential justification, Fumerton tells us that: The state that constitutes noninferential justification is a state that contains as constituents both the bearer of truth value and the truth-maker. (Fumerton 2001: 14, our emphasis, see also 2014: 123)

The reason this passage suggests that Fumerton ought to think that the externalist is barred from holding that perceptual experience is a relation that can bring facts about the external world before the subject’s mind is that if perceptual experience could do that, it would then be sufficient for philosophical assurance. After all, this is what’s supposed to explain the epistemological significance of acquaintance. On the other hand, we might wonder how the externalist could accommodate the view that what constitutes non-inferential justification is a state that contains both the truth-bearer and the truth-maker. Here’s the problem: if it is constitutive of noninferential justification that the target justification-conferring state have the truthbearer and the truth-maker as constituents, ipso facto the state is not the kind of state that one could have in the absence of the truth-maker. And here’s the reason: consider a perceptual experience as of having a hand. Suppose that S has a hand, and that she believes this because she has a perceptual experience as of having a hand. However, let us suppose that it’s possible for S to have that kind of perceptual experience, an experience with the very same nature as the one she’s having now, even if she did not have a hand. Still, this is compatible with S’s perceptual experiences as of having a hand being a reliable indicator that she has a hand. So, if a version of externalism such as reliabilism were true, then this state would be sufficient for non-inferential justification, modulo defeaters. But the proposal from Fumerton in the previous passage suggests that it simply couldn’t be sufficient for noninferential justification—at least, not ‘philosophically relevant justification’ (Fumerton 2006: 188). On this account, it’s not just that the externalist cannot account for philosophical assurance, but that the externalist cannot really account for non-inferential justification either. So, what’s really wrong with an externalist response to the traditional sceptical challenges, such as the problem of the external world, is that their explanation of how our non-inferential justification is possible is philosophically dissatisfying. And what grounds the dissatisfaction we ought to have with their explanation is that what they call ‘non-inferential justification’ isn’t the kind of epistemic state, relation, or process that is even capable of settling the question over the truth of the target belief. But ‘philosophically relevant justification’ has, at the very least, to be capable of settling the question of truth of the target belief. If it is not capable of settling the question of the truth of the target belief, then we will not be in a position to acquire philosophical assurance for that belief.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

ON METAEPISTEMOLOGICAL SCEPTICISM



3 Stroud’s Metaepistemological Scepticism There are at least two interpretations of the source of Stroud’s metaepistemological scepticism in the literature. The first interpretation—the internalist reading— maintains that Stroud argues for internalism about knowledge, and that his antiexternalist view flows from his internalism. The second interpretation—the circularity reading—maintains that Stroud argues that externalism permits an unsatisfactory circular explanation of how knowledge of the world is possible. Let’s begin with the internalist reading. If people know that there are tables, then people know that there are external things. This would be a positive answer to G. E. Moore’s (1939) question of whether the existence of external things is known, rather than just an article of faith. However, Stroud reminds us that: [i]f many people knew that there are external things, but no one knew that anyone had that knowledge, then no one would know the answer to the epistemological question. There would be a positive answer—the existence of external things would be something known and not just an article of faith—but no one would know it . . . The truth of the answer they accept would not give them the understanding they seek unless they could recognize that they know or have good reason to believe that answer. (Stroud 2004: 166)

At first glance, this passage makes Stroud look like an internalist. But our suggestion is that Stroud is not expressing a straightforward internalist thesis about knowledge or justification. Instead, he’s expressing a constraint on understanding how knowledge of the world is possible. What it tells us is that we cannot understand how our knowledge of the world is possible unless we ‘recognize’ that we have knowledge of the world. That thesis does not entail that knowing that p requires having to ‘recognize’ that one knows that p. While there might be an interesting relationship between first-order internalist theses and Stroud’s thesis about philosophical understanding, failing to distinguish them encourages a misreading of Stroud, and so a failure to appreciate his metaepistemological views. A related internalist interpretation of Stroud is due to Fumerton (2006). According to Fumerton, we can distinguish between modest and immodest internalist epistemological goals. The goal of each type of internalist is to philosophically understand how knowledge of some sort is possible, and to do so by respecting internalist principles. According to Fumerton (2006: 181), Stroud is our paradigmatic immodest internalist, because he ‘seems to locate the epistemologist’s target as second-level knowledge (or understanding)’. While it’s true that Stroud locates the epistemologist’s target at second-level knowledge or understanding, it’s not clear that this need give expression to internalist desiderata. After all, Stroud’s core thought seems to be that one of our epistemological goals is to understand how knowledge of the world is possible in the face of various sceptical problems which makes it look impossible. In the face of this challenge, what we want is not just knowledge of the world, but an

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



DUNCAN PRITCHARD AND CHRISTOPHER RANALLI

understanding of how it’s possible given those problems which made it look impossible. We can bring this element of Stroud’s views into sharper relief by considering his views on the relationship between perceptual knowledge and scepticism. On Stroud’s view, perceiving that p is a form of knowing that p, and he thinks that we can come to facts about the external world on the basis of this kind of perception. In fact, he tells us that this ‘is one answer to a question about our knowledge of external things; what might be called the most straightforward answer’ (Stroud 2004: 166). What Stroud means by the straightforward answer to the epistemological question of whether knowledge of the external world is possible is thus that we can come to know that there are external things on the basis of seeing or perceiving that there are. However, Stroud believes that while the straightforward answer is true, it cannot offer us a philosophical understanding of how knowledge of this kind is possible. This might seem odd, since Stroud thinks that the straightforward answer is true, and that its truth implies that we know about the external world. One puzzle, then, is what’s wrong with the straightforward answer, if it’s not that it’s false? Let’s consider two possible deficiencies. On the first account, what makes the straightforward answer deficient is that it presupposes that perceiving that p is a form of knowing facts about the world. What we need to know is that that’s what it is. On the second account, what makes it deficient is that it’s somehow too close to what we want to explain for it to be a satisfying explanation. After all, one might think that an answer to the question ‘how is it possible for one to know that p by perception?’ by saying that one can perceive that p is like saying one can know that p by knowing that p. For Stroud at least, we can know that we perceive that p, because we can perceive that others perceive that p: If perception is indeed a way of coming to know something about external things, then I can also know by perception that that answer to that epistemological question is correct. I can often see that someone right in front of me sees that there is a table in the room and thereby comes to know that there is a table in the room . . . The truth of the straightforward epistemological explanation is something I can know to be true by perceiving that it is true. (Stroud 2004: 166–7)

So Stroud thinks that not knowing that the straightforward answer is true is not what makes it deficient, because we can know that it’s true. What about the second? There is the temptation to protest that knowing that there are external things on the basis of perceiving that there and knowing that other people know that there are external things on the basis of perceiving that that is true, is a circular explanation of how knowledge of the external world is possible. This point is related to Stroud’s (2000a: 201) desideratum that a satisfying explanation of how knowledge of the world is possible must be ‘general in several respects’. For example, Stroud argues that one problem with the straightforward answer is that

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

ON METAEPISTEMOLOGICAL SCEPTICISM



if we thought of sensory knowledge as itself knowledge of material objects around us we would not get an appropriately general explanation of how any knowledge of any [material] object is possible by means of the senses. We would be explaining knowledge of some material objects only on the basis of some others. (Stroud 2000b: 105)

Now, while the transition from seeing that p to knowing that p is not a circular transition, because it’s not an inferential transition, there can still be an issue with appealing to the putative fact that we know that we perceive that p and that it’s a form of knowing that p, because it’s a circular explanation rather than a circular inference. The straightforward answer tries to explain how our knowledge of the world is possible on the basis of our knowledge of the world.

4 Fumerton and Stroud Fumerton maintains that acquaintance with facts is sufficient for philosophical assurance of one’s beliefs about those facts, and is therefore committed to the thesis if we could have direct acquaintance with facts about the external world, then we could have philosophical assurance for our beliefs about the external world. The problem with externalism is that S can be non-inferentially justified in believing that p while their philosophical curiosities about whether or not p are still wide open, so that one lacks philosophical assurance about p. Although Stroud doesn’t express perceiving that p in terms of direct acquaintance with or perception of p, this is a natural understanding of the nature of that state. It’s a state with a mind-to-world direction of fit; one in which it’s impossible to be in unless the content p is true and known to be true. At least, Stroud has endorsed this kind of view in his recent work (see Stroud 2009, 2011). But Stroud wouldn’t want to claim that the immediate perception of external world facts allows one to attain a philosophical understanding of how our knowledge of the world is possible. In this fashion, Stroud can allow that we have assurance of the truth of our beliefs about the world, and even that we have assurance of the truth of that proposition as well. What remains in question is whether this would be sufficient for Fumerton’s philosophical assurance. It seems like it should be, since facts about the external world, on Stroud’s view, can be brought right before our minds during certain kinds of perceptual states. What’s perplexing here is the disagreement between Fumerton and Stroud. Both Fumerton and Stroud seem to think that rule-circular explanations are not good explanations. Stroud thinks the straightforward answer of how we know that there are external things is true, and known to be true. However, his explanation of how we know that it’s true appears to be rule-circular. So, one might think that Fumerton and Stroud are not too far apart after all, since both seem to bar rule-circular explanations. The thought here is that what’s wrong with the straightforward answer, and externalist answers, for example, is that these answers are rule-circular.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



DUNCAN PRITCHARD AND CHRISTOPHER RANALLI

As it turns out, however, things are far more complex. Fumerton holds that acquaintance, unlike being caused to believe that p, or the belief that p being the product of a reliable belief-forming process, is sufficient for philosophical assurance. Of course, one could ask ‘can we be acquainted with the fact that we are acquainted with p?’ But Fumerton thinks we can ask and answer this: we can be acquainted with that fact! However, the externalist will have to keep moving up levels. As he expresses the point: The matter is quite different, I think, with belief-forming processes that may or may not be reliable . . . Am I noninferentially justified in thinking that I am in pain when I stub my toe? The reliabilist, for example, says that I am provided that my belief is caused by a process that is unconditionally reliable. The philosopher can’t resist, at this point, asking the obvious next question. But is my belief caused in the right way? (Fumerton 2006: 189)

This echoes Stroud’s view that we can perceive that we perceive that there are external things. The philosopher can’t resist asking ‘but can we perceive that there are external things—how can we know that the straightforward answer is true?’ But according to Stroud, this question can be settled, just as the question of whether we can know that we are acquainted with certain facts can be settled through acquaintance. Stroud thinks that the reason some philosophers will be dissatisfied with the straightforward answer to the problem of the external world is that these philosophers will simply think it’s false: no one can see that such-and-such is so in the external world if seeing that p is a form of knowing that p. What’s interesting about this suggestion, in connection with our puzzle about what Stroud thinks is wrong with the straightforward answer, is what he thinks the relationship is between the straightforward answer and the problem of the external world: how do we get knowledge by perception of the existence of such external things, given that we never, strictly speaking, see or otherwise perceive that they exist? This is perhaps what has come to be called the philosophical problem of the external world. But with this understanding of the restricted deliverances of unaided perception, the word ‘external’ takes on new significance. It no longer just denotes things to be met with in space, like tables and trees. It now applies to everything that is not . . . perceived to be so; and what is perceived turns out to be much less than might originally have been thought. The problem then is how we can come to know or have reason to believe anything about what is ‘external’ to us, or beyond that limited domain of what we strictly speaking perceive. (Stroud 2004: 168)

This passage gives expression to one of Stroud’s central views about scepticism and the problem of the external world. As Stroud presents it, the philosophical problem of the external world depends upon the denial of the thesis that we can perceive that there are external things. On his account, the phrase ‘the external world’, as it is expressed in ‘the problem of the external world’, means everything that cannot be perceived to be so, and so everything that cannot be known on the basis of perception alone. For Stroud, the denial of our being able to perceive that there are external things is what presents us with the ‘deep and challenging’ epistemological problem.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

ON METAEPISTEMOLOGICAL SCEPTICISM



Here we see that Stroud’s dissatisfaction with the straightforward answer to the problem of the external world is not that it is not true, but that without presupposing that it isn’t true, we won’t have a challenging philosophical problem to answer. Contrast this view with the stronger claim that we just won’t have the philosophical problem of the external world unless the straightforward answer is false. It’s not clear which interpretation is correct. However, in favour of the stronger claim, Stroud tells us that: its source [the problem of the external world] lies somewhere within the familiar and powerful line of thinking by which all of our alleged knowledge of the world gets even temporarily split off all at once from what we get in perception, so that we are presented with a completely general question of how perception so understood gives us knowledge of anything at all in the physical world. If that manoeuver cannot really be carried off successfully, we have no completely general question about our knowledge of the world to answer. (Stroud 2000c: 153)

This passage suggests the stronger reading, because he makes clear that in order to be presented with the problem of the external world, it has to be that ‘all of our alleged knowledge of the world’ gets ‘split off all at once’ from a paradigm case of perceptual knowledge. Now we are in a better position to understand Stroud’s argument against externalism. Let us suppose that Stroud is right about the problem of the external world. On his view, the problem depends upon the view that we’re unable to perceive that there are external things. Now in order to achieve a philosophical understanding of how knowledge of the external world is possible, we have to know that there are external things, and know that we know this. But then how could the straightforward answer be a genuine answer to the philosophical problem of the external world, if there even being some such problem requires that that answer is false? On this view of the problem of the external world, then, it’s the conditional-problem whereby if the perception of facts about the external world is impossible, how then is it possible for us to know anything at all about external things? Let us suppose that the externalist philosopher believes that externalism can provide a satisfying philosophical answer to this challenge, given what Stroud thinks is required for it: that we can’t perceive that there are external things. According to Stroud, the externalist philosopher ought therefore to be in a position to explain how (a) we know that there are external things, and (b) how we know that (a) is true, without the explanation presupposing that we do know about external things. Again, recall that Stroud thinks the proponent of the straightforward answer can explain how (a) is true and how (b) is true. But Stroud wants to draw a distinction between the proponent of that thesis and the externalist at the metaepistemological level. On his view, while the proponents of the straightforward answer can explain how (a) is true, and how (b) is true, their explanation is not intended to be—nor could it be—a bona fide answer to the philosophical problem of the external world.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



DUNCAN PRITCHARD AND CHRISTOPHER RANALLI

However, for the proponent of externalism, their explanation of how (a) is true and how (b) is true is intended to be an answer to the philosophical problem of the external world. For this reason, our assessment of each view will have to answer to different demands: for the former explanation is not even in the market to answer the philosophical problem of the external world, while the other is. In rough outline, Stroud’s argument against externalism is that an externalist account of how we know that we can have perceptual knowledge of the world is no better than an astrologer’s explanation of how astrologers know that we have astrological knowledge of the world. Of course, it might be that there are reliable, co-variations between our perceptual experiences and the facts that we form beliefs on that basis, whereas this is not so for the astrologer’s crystal-ball images. But since the philosophical problem requires that we see their respective epistemic bases— perceptual experience in the one case, and crystal-ball gazing in the other case—as being impoverished and restricted, we have to conclude that both are on epistemic par, as far as understanding how their putative knowledge is possible. In order to appreciate this argument in more detail, consider two kinds of theorists: an externalist about crystal-ball gazing knowledge and an externalist about perceptual knowledge. In each case, the theorist wants to explain how (a) is true and how (b) is true: how we know that there are external things, and how we know that too: The Externalist Crystal-Ball Gazer: I know that I am looking into the crystal ball, and I believe that if there are reliable connections between what happens in the ball and the world I form beliefs about, then I can know about the world I am forming beliefs about. I also believe that there are reliable connections between what happens in the ball and the world I form beliefs about. The Externalist Perceiver: I know that I am having perceptual experiences as of an external world, and I believe that if there are reliable connections between what happens in (or is represented by) these kinds of perceptual experiences and the world I form beliefs about, then I can know about the world I form beliefs about. I also believe that there are reliable connections between what happens in our perceptual experiences and the world we form beliefs about.

Here we need to distinguish between at least three questions we might ask about the ball gazer and the perceiver. The first question is whether or not these theorists are right about there being reliable connections between their epistemic bases and their target beliefs. To this question, one might argue that: The difference between the positions of the two theorists lies only in the believed-in connections between the relevant experiences and the wider world. The theory says in each case that if such connections hold, that theorist knows. Each theorist, confidently sticking to his own story, believes that they hold in his case and not the other. Each might even try to settle the matter by consulting his own experience and his own theory, and find himself content with the discovered result. In that respect, the two positions are equally satisfactory, or unsatisfactory. (Stroud 2004: 171)

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

ON METAEPISTEMOLOGICAL SCEPTICISM



However, this response will strike most externalists as begging the question against externalism. After all, one might suggest that the reliabilist explain the epistemological difference between the ball gazer and the perceiver as follows: the ball gazer does not have reliable beliefs, beliefs in the market for knowledge, because ball-gazing is unreliable, whereas the perceiver does have reliable beliefs, beliefs in the market for knowledge, because perception is a reliable belief-forming basis. Perhaps one will think that the ball gazer and the perceiver are epistemic equals, but to do that would be to reveal one’s internalist intuitions, if ball-gazing is unreliable, while perception is reliable. The second question asks whether or not the ball gazer and the perceiver are in a position to know that their beliefs are reliable, and so in the market for knowledge. In short, the second question is whether the theorists can achieve higher-order knowledge of their respective epistemic sources: ball-gazing and perception. Here the externalist will claim that, because ball-gazing is unreliable, the ball gazer is not in a position to know that her ball-gazing beliefs are reliable, and so in the market for knowledge, whereas the perceiver is in a position to know that her perceptual beliefs are reliable, and so in the market for knowledge. According to this view, if externalism is true, then insofar as the perceiver’s belief that her beliefs formed on the basis of perception are reliable is in fact the product of a reliable beliefforming process, then her higher-order belief will be in the market for knowledge as well. Of course, one might argue that even if the perceiver’s beliefs about the world are in the market for knowledge, the application of their externalist proposal the next level up is problematic, because we do not achieve what Fumerton calls ‘philosophical assurance’ of our belief-forming processes or methods.7 In contrast, Stroud focuses on the fact that (i) both the theorists are externalists, and (ii) that both of their epistemic resources are impoverished in the same way. What he’s interested in is whether one of the theorists is better off than the other, as far as achieving a philosophical understanding of how their (putative) knowledge of the world is possible. What is crucial to his proposal is that neither is better off than the other, because of how impoverished both theorists’ epistemic resources are supposed to be. For in both the perceiver’s case and the crystal-ball gazer’s case, their epistemic basis for forming beliefs about the world is supposed to be limited to what Stroud calls a ‘restricted domain’: mere perceptual experiences as of an external world, and mere ball-gazing images as of an external world. According to Stroud, if neither theorist can perceive that there are external things, then the externalist perceiver is in no better of a position for understanding how her (putative) perceptual knowledge of the world is possible than the externalist ball gazer is for understanding how her (putative) ball-gazing knowledge of the world is 7

See Bergmann (2008) for an exposition and assessment of this response, and Fumerton (2006: 184–6) for a defence.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



DUNCAN PRITCHARD AND CHRISTOPHER RANALLI

possible. This is because what we get in sense-perception bears the same relation to the world we think we know about through perception as the ball gazer’s ball-gazing bears to the world the ball gazer thinks she knows about through ball-gazing. Now this response might seem obscure, because it’s not clear what makes the externalist committed to the thesis that we cannot perceive that there are external things, as opposed to just having perceptual experiences as of there being external things. However, recall that Stroud thinks the rejection of the straightforward answer is dismissed from the start. On Stroud’s view, then, if the externalist takes this response, it’s no better or worse than the straightforward answer. Indeed, he wonders whether we would need an externalist response to the problem, if the straightforward response were available (see, for example, Stroud 2004: 172). But if we reject the straightforward answer because we cannot perceive that there are external things, then according to Stroud (2004: 171), all that ‘we are aware of in perception’ is therefore ‘restricted to features of our perceptual experiences’, such that the ‘external facts we know as a result of those experiences are nothing we ever perceive to be so’. The source of Stroud’s dissatisfaction with externalist responses to the problem of the external world is grounded in this kind of concern. One might query Stroud’s reasons for thinking that the straightforward answer has to be dispensed with from the start. Recall that his reason was that it seems to be a pre-condition of the philosophical problem of our knowledge of the external world. One might question whether the straightforward answer is incompatible with an externalist answer. Perhaps the thesis that one can see that p, where p expresses a proposition about the external world, can be integrated within an externalist account of knowledge, and in particular, be appealed to in order to respond to the problem of the external world (for example, see Williamson 2000). Against this, however, Stroud holds that this kind of response would be puzzling. He tells us that the ‘[straightforward answer] does not look like an answer to a deep and challenging question that we need an “externalist” or any other kind of philosophical theory of knowledge to answer’ (Stroud 2004: 173). The argument here seems to be this. If the straightforward answer were true, then the problem of the external world would not be an interesting, deep, and challenging philosophical problem. But if one thinks that it is an interesting, deep and challenging philosophical problem, it follows that one has rejected or otherwise dismissed the straightforward response.8 In this fashion, the tension between the straightforward answer and the problem of the external world is that the latter requires the former to be false, even if in fact it isn’t false. We can bracket whether one should think that Stroud is right about the straightforward answer having to be presupposed as false for there to be a philosophical 8 Indeed, Fumerton (2014) advances an argument of this kind against those ‘disjunctivist’ views (broadly speaking)—e.g., McDowell (1995), Williamson (2000), and Pritchard (2011, 2012)—which maintain that the nature of one’s epistemic support in cases of veridical perceptual experience can be

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

ON METAEPISTEMOLOGICAL SCEPTICISM



problem of the external world. These are interesting and difficult issues that we won’t attempt to settle here. Instead, our focus is on whether Stroud has made a good case for the view that if the crystal-ball gazer and the perceiver have restricted epistemic sources, then they are epistemically on par as far as understanding how their (putative) knowledge of the world is possible. In order to motivate Stroud’s case, consider the following example. First, imagine the crystal-ball gazer, who believes that a book is there on the basis of seeing an image of a book in the crystal ball. Second, imagine the perceiver, who believes the same proposition, but on the basis of her perceptual experience as of a book being there. Both of these epistemic sources are restricted in the sense that it does not follow from S ball-gazed that p, that S knows that p (or even that p), and it does not follow from S has a perceptual experience as of p, that S knows that p (or even that p). Moreover, what the perceiver and the ball gazer appeal to in forming their beliefs about their target epistemic sources is restricted to the mere deliverances of those epistemic sources alone, and what one could reason from what can be known on the basis of those sources alone. Now the crucial question here is how far each theorist’s track record of their epistemic sources would take them. After all, when the perceiver forms beliefs about how reliable their perceptual experiences as of the external world are, their comparison will be between previous perceptual experiences as of p being true, and their believing that p is true, their perceptual experiences as of q being true, and their believing that q is true, and so on. For example, the perceiver will believe that their perceptual experiences as of the cat being on the mat are reliable perceptual experiences, because on most of the occasions she had that kind of experience, a cat was on the mat—or rather, it was for the perceiver just as if a cat was on the mat. What the perceiver can consult is more perceptual experiences, and whatever they can infer from what can be known a priori or by perceptual experience. Likewise, when the ball gazer forms a track record about the constellations and people’s personalities, she too will be restricted to what can be gazed in the ball, and known on the basis of crystal-ball-gazing alone. For example, when she goes to check whether her crystal-ball-gazing experiences as of the Moon being full are reliable indicators of their friend Smith being sad, the crystal-ball gazer will be consulting the ball and whatever can be inferred that she knows on that basis. Which theorist is better off as far as understanding how their knowledge is possible? On the one hand, it might be that one theorist has more true beliefs than the other. In this sense, one is better off because one has more true beliefs than the other. But when it comes to them knowing that their beliefs are true, how could one of the theorists be better off than the other? As Stroud noted, the relation the radically different from cases of introspectively indistinguishable hallucination. For an overview of how metaphysical and epistemological disjunctivism makes contact with scepticism, see Pritchard and Ranalli (forthcoming).

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



DUNCAN PRITCHARD AND CHRISTOPHER RANALLI

perceiver stands to their perceptual experiences is the same kind of relation that the ball gazer stands to their crystal-ball-gazing experiences. Again, it might be the case that the perceiver’s first-order beliefs about the world are true, and their secondorder beliefs about the epistemic merit of their perceptual experiences are true, but when it comes to knowing that this is so, and so being in a position to understand how their putative knowledge is possible, it’s hard to see how the perceiver could be better off than the crystal-ball gazer, because the perceiver will be consulting the restricted deliverances of their perceptual experiences, just as the crystal-ball gazer will be consulting the restricted deliverances of their crystal ball. Of course, the externalist still might not think there’s a real problem here, because perceptual experiences as of p—even if ‘restricted’ along the lines that Stroud has in mind—are still reliable, whereas crystal-ball-gazing is not. But Stroud’s rejoinder here is that the externalist philosopher—in providing this kind of answer—must have relied upon their perceptual experiences as epistemic support for that claim. In this fashion, the externalist philosopher who accepts that our perceptual experiences of the world are restricted along the lines that Stroud has in mind provides an actual test case for his metaepistemological claim. The thought here is this: what else is the externalist philosopher going to appeal to in order to support their belief that their own and other people’s perceptual experiences are reliable, if not that he has had perceptual experiences as of himself and others making reliable judgements about the world which have been correct in the past, and that we had perceptual experiences as of our and other people’s judgements, assertions, and expressions of belief being accurate representations of the world? So, the problem here is that what the externalist philosopher must appeal to is no better at the higher-order level than it was at the first-order level: either it’s as good or it’s as bad. But the problem with the track-record argument is that the externalist must acknowledge that he does not . . . ever see or otherwise perceive that those human beings and other external things that he is interested in are there. Nor does he ever perceive the reliability of the connections that he believes hold between them. The most he is perceptually aware of or presented with in experience are the qualities or character of his perceptual experiences. (Stroud 2004: 170)

On Stroud’s view, then, the problem with the externalist using a track-record argument in order to provide support for the belief that perception is reliable is that all that that would-be externalist can perceive to be so falls short of the facts in the external world. All we would have are perceptual experiences as of p, rather than perceptions that p. In short, the track record does not improve upon the so-called impoverished epistemic sources, such as perception (so understood) and crystal-ball-gazing. The core disagreement between Stroud and Fumerton, then, is that: (1) we have assurance of the truth of our beliefs about the external world, and we can have assurance of the truth of this belief as well.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

ON METAEPISTEMOLOGICAL SCEPTICISM



(2) if (1), then we could have a philosophically satisfying explanation how that knowledge is possible. Stroud is committed to (1), but he denies that (2), while Fumerton is committed to (2), but denies that (1). While both theses are contentious, it might strike one that the denial of (2) is much stronger than accepting it. In this fashion, Fumerton is on stronger ground than Stroud. On the other hand, both Fumerton and Stroud’s arguments against externalist explanations of how our knowledge of the world is possible seem to turn on their dissatisfaction with rule-circular explanations. After all, what’s so pathetic about using perception in order to provide support for the thesis that perception is a reliable source of knowledge of the world, contra scepticism, is that that kind of explanation is rule-circular. At first glance, Stroud thinks that the straightforward answer, that we can perceive that facts about the external world obtain, and that we can perceive that the straightforward answer is true, is not a good answer to the philosophical problem of the external world because it’s rule-circular. In other words, when Stroud demands the answer or explanation to be ‘general’, his demand is that it not be rule-circular. Of course, as Michael Bergmann (2000, 2005) has argued, this kind of antiexternalist complaint, that one cannot acquire non-inferential justification or knowledge that one’s principles, methods, or rules that are required for that kind of knowledge or justification are true without presupposing or using those principles, methods, or rules, is applicable to Fumerton’s own view. After all, explaining how we know that acquaintance with facts generates non-inferential knowledge or justification on the basis of being acquainted with the fact that acquaintance with facts generates non-inferential knowledge or justification is rule-circular. According to Fumerton (e.g., 2006, }4), however, the acquaintance theorist avoids this objection, the objection that externalists do not avoid, because acquaintance allows, while one’s belief-forming process or method meeting various externalist conditions does not, one to acquire philosophical assurance for the relevant higherorder beliefs. So, Fumerton might accept a metaepistemological constraint like the following: (F) No satisfying, philosophical explanation of how knowledge of kind K is possible can be rule-circular unless our explanation of how we know that our explanation is correct assures our philosophical curiosities. Stroud, on the other hand, appears to be committed to a stronger thesis than (F), where the qualification is dropped. Indeed, this helps corroborate Fumerton’s claim that Stroud sets less modest internalist goals than he does, since, for Stroud, there are no satisfying rule-circular explanations of how our knowledge of the world is possible. However, a proponent of Stroud’s views might respond to this objection as follows. We need to distinguish between (a) the deficiencies in the externalist explanation

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



DUNCAN PRITCHARD AND CHRISTOPHER RANALLI

from (b) the deficiencies in the straightforward explanation (that we can perceive facts about the external, and that we know that we do on the basis of perceiving that that is so). What’s wrong with the straightforward explanation is not that it’s rulecircular, even if it is. It’s that it is not an answer to the problem at all. On the other hand, what’s wrong with the externalist answer, for Stroud anyway, is that from the first-person perspective, the externalist theorist about perceptual knowledge of the world is in just as good or bad an epistemic position as regards understanding how their (putative) knowledge is possible as other kinds of externalist theorists are, even ones who maintain a different view, not about the nature of knowledge or justification, but about how it is acquired. Stroud’s constraint that one’s explanation is philosophically convincing only if it neither implies nor presupposes that one has knowledge of the world has to be satisfied, then, because without it we are not answering the philosophical problem of the external world. We can therefore summarize Stroud’s ‘resolution’ to the problem of the external world as follows: if we can explain perceptual knowledge so that the philosophical problem of the external world cannot emerge, then the problem will dissipate—there will be no problem for which a positive ‘anti-sceptical’, or negative ‘sceptical’, answer can be given. This, of course, should seem surprising. When it comes to giving a satisfying explanation of how our knowledge of the world is possible, what more could one want—to echo Fumerton—than to know that we are directly epistemically related to facts about the external world? For Stroud, this isn’t enough because it violates his generality condition, the condition that our explanation cannot imply or presuppose that we know anything about the external world. This, then, is the heart of the difference between Fumerton and Stroud.9

References Bergmann, Michael. (2000). ‘Externalism and Skepticism’, Philosophical Review, 109: 159–94. Bergmann, Michael. (2005). ‘Defeaters and Higher-Level Requirements’, Philosophical Quarterly, 55: 419–36. Bergmann, Michael. (2008). ‘Externalist Responses to Skepticism’, in John Greco (ed.), Oxford Handbook to Skepticism. Oxford: Oxford University Press, 504–38. BonJour, Laurence. (1985). The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press. Campbell, John. (2009). ‘Consciousness and Reference’, in Ansgar Beckermann, Brian P. McLaughlin, and Sven Walter (eds), The Oxford Handbook of Philosophy of Mind. Oxford: Oxford University Press, 648–62. Fumerton, Richard. (1995). Metaepistemology and Skepticism. Boston, MA: Rowman and Littlefield. 9

Thanks to Michael Bergmann, Brett Coppenger, Richard Fumerton, and the other participants at the Orange Beach Epistemology Workshop, at which an earlier version of this chapter was presented.

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

ON METAEPISTEMOLOGICAL SCEPTICISM



Fumerton, Richard. (1998). ‘Précis of Metaepistemology and Skepticism’, Philosophy and Phenomenological Research, 58: 905–6. Fumerton, Richard. (2001). ‘Classical Foundationalism’, in M. DePaul (ed.), Resurrecting OldFashioned Foundationalism. Lanham, MD: Rowman and Littlefield, 3–20. Fumerton, Richard. (2006). ‘Epistemic Internalism, Philosophical Assurance and the Skeptical Predicament’, in Thomas M. Crisp, Matthew Davidson, and David Vander Laan (eds), Knowledge and Reality: Essays in Honor of Alvin Plantinga. Dordrecht: Springer, 179–91. Fumerton, Richard. (2014). ‘The Challenge of Refuting Skepticism’, in Matthias Steup, Ernest Sosa, and John Turri (eds), Contemporary Debates in Epistemology (2nd edition). Oxford: Wiley-Blackwell, 120–32. McDowell, John. (1994). Mind and World. Cambridge, MA: Harvard University Press. McDowell, John. (1995). ‘Knowledge and the Internal’, Philosophy and Phenomenological Research, 55: 877–93. Moore, G. E. (1939). ‘Proof of an External World’, Proceedings of the British Academy, 25: 273–300. Pritchard, Duncan H. (2005). Epistemic Luck. Oxford: Oxford University Press. Pritchard, Duncan H. (2011). ‘Epistemological Disjunctivism and the Basis Problem’, Philosophical Issues, 21: 434–55. Pritchard, Duncan H. (2012). Epistemological Disjunctivism. Oxford: Oxford University Press. Pritchard, Duncan H. and Christopher Ranalli. (Forthcoming). ‘Disjunctivism and Scepticism’, in Baron Reed and Diego Machuca (eds), Skepticism: From Antiquity to the Present. London: Bloomsbury. Stroud, Barry. (2000a). Understanding Human Knowledge: Philosophical Essays. Oxford: Oxford University Press. Stroud, Barry. (2000b). ‘Understanding Human Knowledge in General’, in Understanding Human Knowledge: Philosophical Essays. Oxford: Oxford University Press, 99–112. Stroud, Barry. (2000c). ‘Scepticism, “Externalism”, and the Goal of Epistemology’, in Understanding Human Knowledge: Philosophical Essays. Oxford: Oxford University Press, 139–55. Stroud, Barry. (2004). ‘Perceptual Knowledge and Epistemological Satisfaction’, in John Greco (ed.), Sosa and His Critics. Oxford: Blackwell, 165–73. Stroud, Barry. (2009). ‘Scepticism and the Senses’, European Journal of Philosophy, 17: 559–70. Stroud, Barry. (2011). ‘Seeing What Is So’, in J. Roessler, H. Lerman, and N. Eilan (eds), Perception, Causation, and Objectivity. Oxford: Oxford University Press, 92–101. Travis, Charles. (2004). ‘The Silence of the Senses’, Mind, 113: 57–94. Travis, Charles. (2005). ‘A Sense of Occasion’, Philosophical Quarterly, 55: 286–314. Williamson, Timothy. (2000). Knowledge and Its Limits. Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

11 How Our Knowledge Squares with Skeptical Intuitions Despite the Circle Ernest Sosa

I welcome this opportunity to continue a dialog that for decades I have enjoyed with our honoree. It is a great pleasure to be able now to express publicly my great admiration for Richard Fumerton, both the philosopher and the man, and for his work. I am sure that, at least on my side, our present exchange will be once again a learning experience.

1 Preliminaries on Skepticism in General If a skeptic challenges us to prove ex nihilo that we see a hand, or a fire, we should decline. Nor should we take the bait when he offers us as premises just obvious a priori truths and facts about our own current subjective states. Even given such premises, pure reason (deductive or inductive) will not yield nearly enough of what we believe ordinarily. Our knowledge of hands and fires is not explicable in terms of such reasoning from the given. Far from accepting such skeptical challenges, we should decline the presuppositions from which they are launched. We should reject the sheer assumption that only reasoning from the foundational given will provide or explain whatever knowledge we may enjoy. I take more seriously a more serious skeptical challenge. Here’s how I would distinguish it from those of lesser concern.

1.1 Varieties of skepticism Skepticism comes in various forms. Pyrrhonian skepticism is a determined suspension of judgment, through tropes that counter whatever reasons may be offered in favor of belief. The Academic skeptic, by contrast, is more assertive, if only by claiming that we know nothing, either in general, or in some large department of

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

HOW OUR KNOWLEDGE SQUARES WITH SKEPTICAL INTUITIONS



our supposed knowledge: the external world, for example, or other minds, or morality. Any attempt to refute the fully global claim therefore begs the question. A claim might be irrefutable, however, without being true. Consider the global claim that we know nothing at all. To refute this, one must adduce some premise, doing which is implicitly to claim knowledge of its truth, thus begging the question. Dialectically irrefutable he may be, our global skeptic, since the context of dialectic prohibits such question-begging. But he is not thereby shown to be right. Besides, if the very making of a claim commits also to knowledge of what is claimed, then the global skeptic contradicts himself. Such varieties of skepticism are less problematic than a more interesting variety. This more interesting skeptic spots a commitment ostensibly at the heart of common sense, and shows how it entails that we know very little, either in general or in some main department.

1.2 Skepticism, sensitivity, and safety Consider, for example, the idea that a belief constitutes knowledge only if sensitive, only if it satisfies the following condition: had its propositional content been false, it would not have been held by the believer. In order for you to know that you see a hand, your belief must be sensitive, in that had you not seen a hand, you would not have believed that you saw one. If this sensitivity requirement is indeed among our core commitments, the skeptic is in a good position. Belief that one is not radically misled cannot be sensitive. Indeed, skeptical scenarios are framed to secure precisely this result. If you were now a brain in a vat being fed experiences as if you enjoyed normal perception, that would not stop you from believing that you were not radically deceived. Various responses to such skepticism have been developed over many decades, including closure-denying tracking approaches, and contextualist semantic ascent. A rich and subtle dialectic can be found in an extensive and still actively growing literature. No such sensitivity-influenced approach is fully satisfactory, however, though that is not something I can argue in the space available here. There is anyhow, in my view, a better response to the skeptic. It begins by noting that subjunctive conditionals do not validly contrapose,1 which suggests a requirement of safety rather than sensitivity, as follows. Basis-relative safety: A belief cannot constitute knowledge if the believer might too easily have so believed on the same basis while his belief was false. (Alternatively, in order to know one must believe on a certain basis, possibly the null basis, such that one’s so believing on that basis must have a strong enough tendency to be right.) 1 For example, if water flowed out of my kitchen faucet, it would be false that water flowed while the main house valve was closed. But consider the contrapositive of this conditional: namely, that if water flowed out of the faucet while the main valve was closed, then water would not flow. This contrapositive would be obviously false, showing contraposition to be invalid for such conditionals.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



ERNEST SOSA

Safety does not serve the skeptic as does sensitivity. Belief that one is not radically deceived is insensitive, and is even insensitive relative to its deep experiential bases. But a belief can be safe while insensitive. Scenarios of radical deception are outlandish, remote possibilities not liable to occur (not really, not too easily). Therefore, a belief that one is not radically deceived is safe while insensitive: not too easily might one have been radically deceived. Belief that one is not radically deceived would tend to be correct. The possibility of radical deception is so outlandish that one’s belief to the contrary would tend to be correct.

1.3 Why the dream scenario is special Our line of reasoning is effective against radical scenarios, such as the brain in a vat, the evil demon, the Matrix, and so on. Only the dream scenario stands apart. Dreams being so common, the possibility that one dreams is not outlandish. Therefore we cannot defend the safety of our belief that we are awake by adducing how remote is the possibility that we go wrong in so believing. Too easily for comfort might we have been not awake but only dreaming. Here one might reasonably ask: “Which dream scenario?” The scenario that life is but a dream—all of life—is a dream scenario, but it is hardly less outlandish than BIV or evil demon scenarios. That is, accordingly, not the dream scenario of special interest to us. The more relevant dream scenario is the one posed by Descartes when he wonders, as he sits before a fire, whether he is then just dreaming. Our dream scenario is one that arises for any arbitrary case where we consider whether our ostensible perceptual knowledge—of a fire, say, or a hand—is real on that occasion. Such a dream scenario has a distinctive importance by comparison with the familiar radical scenarios. Unlike the others, it is not outlandish. And, for another thing, it threatens our perceptual beliefs directly, not by way of closure.2 The threat involves danger or risk, the danger or risk of believing falsely. This does not depend on awareness by the subject in danger that or how he may be at risk, nor on his evidence concerning such risk, nor even on whatever evidence may be available to him. So I would reject that line of reasoning. As one strides across a minefield one can be in great danger, even with no inkling of that fact, nor any available evidence for so believing. The danger to one’s attaining ordinary knowledge does require a certain orthodox conception of dreams, according to which beliefs and experiences in our dreams are hosted not only in the dream but also in actuality, while we dream. Only thus would our ordinary perceptual beliefs be threatened by the possibility that in a realistic dream we might believe the same on the same experiential basis. 2 Alternatively, one might argue that just as the modal proximity of enough possible fake barn encounters creates a problem for the belief that one perceives a barn, so the modal proximity of enough possible dreams creates a problem for the belief that one perceives an external reality.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

HOW OUR KNOWLEDGE SQUARES WITH SKEPTICAL INTUITIONS



Is that really how we should conceive of our dreams? Are dreams made up of conscious states just like those of waking life except for how they fit their surroundings? The orthodox answer is in the affirmative. Dream states and waking states are thought intrinsically alike, though different in their causes and effects. The orthodox view is deeply flawed, however, or so I have argued in detail, before suggesting a better view. To dream is to imagine, not to hallucinate.3 Ordinary perceptual beliefs can hence retain their status as apt, animal knowledge, despite the nearby possibility that one is asleep and dreaming. Ordinary perceptual beliefs can still attain success through perceptual competence, despite the fragility of that competence and of its required conditions. However fragile a performer’s competence may be, and however fragile may be the conditions appropriate for its exercise, if a performance does succeed through the exercise of that competence in its proper conditions, it is then an apt performance, one creditable to the performer. Knowledge is just a special case of such creditable, apt performance. Perceptual knowledge is unaffected by any fragility either in the knower’s competence or in the conditions appropriate for its exercise. The knower’s belief can thus remain apt even if unsafe through the proximity of the dream possibility.

1.4 Skepticism and circularity Thus, safety-based and sensitivity-based skepticism deserve an answer, and fortunately good answers can be delivered. By contrast, any epistemology that demands a ‘legitimating account’ of our knowledge in general deserves only to be rejected as incoherent. The supposed ‘legitimating’ account would need to lay out justifying reasons for every bit of our knowledge, and to do so without circularity or infinite regress. This is soon seen to be obviously beyond human limitations, indeed beyond the almighty himself. So, the demand seems ‘incoherent.’ In what follows I will argue that any epistemology that demands such ‘legitimacy’ is thus rendered illegitimate through a kind of incoherence.

2 A More Specific Disagreement on Philosophical Skepticism Two main strands of thought have concerned Fumerton in our earlier dialectic on philosophical skepticism. I will focus here on what I take to be the more fundamental of these two. Once we see what I take to be a good way to handle that one, this should have clear implications for how to handle the other one, and for how the two responses should be interwoven for a satisfying epistemology. The lesser strand concerns the problem of the external world, and how that relates to a particular sort of externalism in epistemology, naturalist externalism, which 3

Sosa 2007: ch. 1 defends this in detail.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



ERNEST SOSA

takes various forms in the epistemology of recent years, including causal, tracking, process, and virtue varieties. This is the worry that I will put aside here, in the hope that the position to be defended concerning the other worry will have significant bearing on this one as well. Here I will not consider this lesser worry further in its own right. The more important worry concerns a much more general epistemological and skeptical problematic, found in Pyrrhonism, and also important in Descartes’ epistemology. This will be my main focus in what follows. Fumerton has emphasized how my virtue reliabilism is a form of externalism similar to standard reliabilism, which should make it a proper target for standard internalist objections. He writes: When I stress the similarities between Sosa’s virtue epistemology and standard reliabilism, I do so presupposing what one might call a minimalist version of reliabilism. While there typically is a genuine process that takes input and churns out output beliefs, it has always seemed to me that a reliabilist should (and usually does) allow for a ‘process’ that is nothing more than an ‘input’ that immediately takes us to an ‘output’—that’s probably the best way to think of something like introspection. And Sosa certainly wants to allow that one can exhibit intellectual virtue if one forms a true belief as a result of an ability where the manifestation of the ability is nothing more than adroitly believing as a result of contemplating a given proposition. (Fumerton 2011: 114)

With this minimalist reliabilism I must agree. Indeed Descartes himself clearly accepts it as well. Consider, for example, the following passage near the beginning of his Third Meditation: I am certain that I am a thinking thing. But do I not therefore also know what is required for me to be certain of anything? Surely in this first instance of knowledge, there is nothing but a certain clear and distinct perception of what I affirm. Yet this would hardly be enough to render me certain of the truth of a thing, if it could ever happen that something that I perceived so clearly and distinctly were false. And thus I now seem able to posit as a general rule that everything I very clearly and distinctly perceive is true. (CSM 2: 24)4

Note the sequence here. Descartes begins with the attribution to himself of certainty that he is a thinking being. And he then reasons as follows: 1. I am certain that I am a thinking being. 2. Nothing gives me this certainty but the clear and distinct perception that I am a thinking being. 3. But such clear and distinct perception could provide such certainty only if nothing could ever be that clear and distinct while false. 4. Hence, nothing could ever be that clear and distinct while false.

4

References to Descartes’ work are to volume and page number of Cottingham et al. 1986–91. References are abbreviated CSM.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

HOW OUR KNOWLEDGE SQUARES WITH SKEPTICAL INTUITIONS



Here premise 3 strikes me as an unequivocal commitment to a strong reliabilism. Immediately thereafter Descartes still wonders how he can be sure that his acceptance of 4 is really true. And this sets off the theological inquiries that lead him to his positive perspective on clear and distinct perception, i.e., on the direct clear and distinct perception constituted by intuition, and on the indirect clear and distinct perception provided by deduction from the intuitive. Not only is Descartes a reliabilist, however. As I argue elsewhere, he is clearly a virtue epistemologist among whose main epistemological concepts is that of the apt belief or judgment. And he is, moreover, a bi-level virtue epistemologist, with an animal level of cognitio, and a reflective level of scientia.5 Descartes does of course tackle the external-world epistemological problematic. Clearly that is not his deepest skeptical and epistemological concern, however, since the skeptical doubts that he takes up with utmost seriousness extend far beyond external world concerns about fires and dreaming, to cover even the simplest truths of arithmetic and geometry. In what follows it will be helpful to focus on this deeper problematic, which concerns not just sense perception, but even intuition and deduction, the faculties of main interest to Descartes in his quest for certainty. So, let us join Descartes in his radical internalist virtue reliabilism of intuition and deduction. After he has reflected in the early Meditations and reviewed various radical scenarios, by the beginning of his Third Meditation he is prepared to selfattribute certainty that he is a thinking being, which he does in the passage quoted above. And he notes that, as far as he can see, what gives him this certainty is simply clarity and distinctness. Descartes believes in a faculty of intuition that delivers certainty through direct clear and distinct perception of basic truths, independently of any inference. But he is willing to put even this faculty in question, as part of his general question as to what might assure him that nothing could possibly reach a certain level of clarity and distinctness—the level reached by his knowledge that he is a thinking being—without being true. And now what is to stop him from reasoning as follows? Here in this first instance is something I intuit to be true and is true. Here in this second instance is something that I intuit to be true and is true. ... The best explanation for all this success is that intuition is a supremely reliable source of truth. (In each instance here the truth claim derives simply from intuition.)

Descartes did after all believe in inference to the best explanation.6 And he might thus have adopted this easy inference where each of the premises is delivered (at least in essential part) through an exercise of intuition, reaching thereby the conclusion that 5

This I do in Sosa 2013 and 2014. See his Principles of Philosophy, Principle 205. (This would of course require going beyond intuition and deduction strictly conceived; but there are other ways in which Descartes seems willing to be thus flexible, as when at the end of the Meditations he reasons that it would be laughable to be fooled by dreams, 6

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



ERNEST SOSA

intuition is reliable. Thus he would have been able to arrive at easy knowledge of the reliability of his intuition, of his clear and distinct direct perception of truth. That is one way in which the problem of easy bootstrapping might have affected even a Cartesian infallibilist internalist, and not just a fallibilist naturalist externalist. There is moreover a further issue of bootstrapping that affects Descartes. As we have seen, he seems to reason as follows: (a) If I (now) think, then (necessarily) I exist. (b) I (now) think that I exist. Therefore, (c) My (present) thought that I exist cannot possibly be false (and I cannot possibly be deceived in so thinking).7 If so, why then could he not have reasoned as follows, already in Meditation One? (a’) 3 + 2 = 5 (necessarily so). (b’) I (now) think that 3 + 2 = 5. Therefore, (c’) My (present) thought that 3 + 2 = 5 cannot possibly be false (and I cannot possibly be deceived in so thinking). Yet this primed reasoning, a’-b’-c’, seems viciously circular, unlike the earlier a-b-c. The primed argument amounts to ‘blatant bootstrapping,’ which adds no significant epistemic support for its conclusion. By contrast, Descartes’ cogito reasoning (as in the unprimed a-b-c) does not seem similarly suspect. This may help explain why Descartes favors cogito thoughts in his Meditation Two. Cogito thoughts he can defend against skeptical doubts.8 He can defend them by appeal to the general which evidently differ from waking life in respect of a kind of diachronic coherence that we can presumably track with the aid of our memory.) 7 Against my suggestion here, someone might argue as follows: “Descartes says in the Second Replies that his knowledge that he exists is not derived from the general proposition that whatever thinks exists. Rather, the Meditator recognizes [the cogito] as something self-evident by a simple intuition of the mind. The general proposition is itself derived from the cogito, rather than the other way around” (see CSM 2: 100). If the cogito is thus self-evident through sheer intuition, moreover, then it is not derived inferentially from the conditional claim that if one thinks then one exists. This is all compatible with my suggestion, however, since the Meditator might (a) recognize the cogito (as self-evident), and (b) recognize even that it is self-evident, by directly intuiting each of these (each of: the cogito itself, and also its being self-evident), while yet (c) he can also reason his way to a meta-belief that it is infallible, where this latter also bears on its epistemic status, by further defending it reflectively. In effect, the Meditator’s cognitio that he exists could be just intuitive (and might attain the highest level of certainty as cognitio), even though his scientia that he exists must go beyond that, by relying also on the sort of reflective defense provided by the reasoning early in the Meditations. 8 Against this someone might argue as follows: “But how could the first premise, premise (a), be defended against skeptical doubts? And, if it can’t be so defended, how could the conclusion be defended? On the other hand, if the conclusion can be defended against skeptical doubts but the premise cannot,

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

HOW OUR KNOWLEDGE SQUARES WITH SKEPTICAL INTUITIONS



consideration that they cannot be believed incorrectly: their very content guarantees that when affirmed they must be true (that they are also indubitable is something he stresses in the Second Replies (CSM 2: 104)). This we can see without reasoning from the truth (or the necessary truth) of what is affirmed, which distinguishes Descartes’ cogito reasoning from blatant bootstrapping like our primed reasoning above.9 One might thus (with Descartes) entertain the question whether one’s armchair faculties are reliable enough to deliver certainty. The well-known antinomies and paradoxes of reason might moreover give rise to the worry that armchair reason itself is subject to cognitive illusion. So, one might hope to attain some perspective on one’s own faculties of reason—such as intuition and deduction—some perspective that might assure us that they are indeed reliable enough to deserve our trust. That is, I believe, Descartes’ situation as the Meditations unfold. How, now, might one go about seeking such assurance. Descartes does not hesitate to put his best and most basic faculties to work in the development (through rational theology) of such a reassuring perspective. My bi-level approach similarly allows the use of our basic foundational faculties in the elaboration of a second-order assuring perspective. So, we can use the animal knowledge that we attain through the exercise of such faculties; we can use such animal knowledge in the (proper, coherential) elaboration of the endorsing perspective. This Fumerton rejects as follows: Either the coherence of a belief with other beliefs is by itself intellectually satisfying, or it is intellectually satisfying only if accompanied by the prior animal-level justification. Based on Sosa’s discussion of Stroud in Chapter 8 of Reflective Knowledge II, I think Sosa wants to suggest that it is the latter. But here the skeptic is bewildered. The skeptic imagines someone who has the very same beliefs that we have about appearances but who couples those beliefs with . . . skeptical hypotheses . . . The skeptic wants to know why the beliefs we have are any more likely to be true than beliefs embedded in one of these alternative belief systems . . . Any answer that appeals to the purely external virtues of animal-level justification enjoyed by one set of beliefs will strike the skeptic as blatantly question begging and unable to give us the relevant intellectually satisfying justification. (Fumerton 2011: 122)

As far as I can see, this accusation of petitio would be no less in order against the Cartesian approach than against my own. And that is, I believe, because these are doesn’t that give us reason to think that the certainty of the conclusion is not actually grounded in the premise or the argument?” As argued earlier, however, there can be a defense of the cogito belief by appeal to given premises, even if the epistemic status that it derives from this defense is not that of a first-order conclusion based on first-order inference from those premises. 9 Compare the following reasoning: (a1) I exist, (b1) I (now) think that I exist; therefore, (c1) my (present) thought that I exist is true. This reasoning too seems viciously circular as does the primed reasoning in the text. It too seems a case of blatant bootstrapping. But the initial unprimed reasoning in the text seems better, as does the following unmodalized correlate of it: (a2) Whatever thinks exists, (b2) I (now) think that I exist; therefore, (c2) my (present) thought that I exist is true. Neither the a-b-c reasoning, nor its correlate a2-b2-c2, appeals to a premise that simply asserts as true (or as necessarily true) the very content of the thought said to be true (or necessarily true) in the conclusion.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



ERNEST SOSA

really at bottom the same approach, the same bi-level approach. It’s just that Descartes restricts himself to armchair faculties whereas I do not. But the structural issues are, I believe, exactly the same. If the circularity-wielding opponents of the Cartesian virtue-epistemological approach are wrong, then the petitio-wielding opponents of my animal/reflective virtue-epistemological approach are wrong for the same reasons. Fumerton holds out for an intellectual reassurance that will not be subject to the easy-knowledge—circularity, petitio—sorts of objections. One might thus hold out for what I above called a “legitimating account of our knowledge in general.”10 The relevant account, which I reject as ‘incoherent,’ demands a ‘legitimating account’ of our knowledge in general. This is an account that would lay out justifying reasons for every bit of our knowledge, and would do so without circularity or infinite regress. Since it quickly becomes obvious that this is beyond human limitations, and indeed beyond the almighty himself, the demand is one that I find obviously ‘incoherent.’ This means not just that it is unfulfillable, but that it is unfulfillable with metaphysical necessity. A further incoherence involves the knowledge that the aim is obviously impossible together with the demand or desire for its realization nonetheless. These two attitudes do not cohere properly. One cannot coherently desire that p while knowing that it is obviously impossible that p. I need not insist on the terminology of ‘coherence,’ however, since the targeted combination is in any case bad. It is surely bad to want something that not even God could attain, especially when the impossibility becomes obvious. There is now a strong reason to try to escape that bad situation. But there’s not much to be done about the impossibility. One must instead try to get rid of the desire. Conceivably the desire in question, or even its combination with the impossibility, might have excellent consequences. If so, this might then trump our reason for getting rid of it: the reason, namely, that this desire will be inevitably frustrated, obviously so. Since we can foresee no such excellent consequences, however, nothing like that speaks in favor of retaining any desire for a legitimating account of our knowledge, and it still seems clear that we should try to get rid of that desire (if we have it) and put our time to better use. To abandon the search for such a general legitimating account, is not to give up altogether on the project of developing a general epistemology. We must be clear, however, on the sort of epistemology desired. It cannot be a methodology, a set of rules or principles by following which we would attain any knowledge that is attainable. According to such a ‘methodology,’ it would be by following such instructions that we would always gain the status of justification required for knowledge. That seems little better, however, than the search for a legitimating account, and indeed the two ill-conceived objectives are closely related. Following a rule would 10

That’s the label I use in my controversy with Stroud in Sosa 2011: ch. 8, a chapter referred to by Fumerton in the present connection.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

HOW OUR KNOWLEDGE SQUARES WITH SKEPTICAL INTUITIONS



require that we ascertain the obtaining of the conditions laid down in the antecedent of the rule. But this would require that we be justified in believing that those conditions obtain. So there must have been some non-posterior application of a rule. And once again it seems clear that there is no way to follow any such set of rules without vicious circularity or regress so as to attain all of our justified belief. If we still seek a general epistemology, then, it cannot be a general methodology of that sort. But it might still be a general explanatory account of our knowledge, an account that would be able to explain, for any bit of knowledge, how it is that the constitutive belief comes to have the epistemic normative status required for knowledge. If we call that normative status “(epistemic) justification,” the objective would be a general explanatory account of such justification. Once we are clear on this, consider now the circularity that such a successful account would inevitably involve. By a successful account I mean an account through which we epistemologists would succeed in understanding human knowledge in general. We would presumably have a set of basic competences some subset of which we would use in coming to know our general account of how we attain our knowledge. This general account would then have to give us an understanding of such competences and their reliability. So, the competences by means of which we would have come to know our account would be among the competences that our account invokes, and so they would be among the competences that our account must claim to be reliable. I see no vice in this sort of circularity. Compare the circularity that is involved in coming to understand how our faculty of vision gives us reliable access to our surroundings. In arriving at the relevant theory about the specifics of our faculty of vision we will presumably be using our eyes to gather relevant data, based on which data we come to know about the optic nerve, the structure of our eyes, etc., so as to have a specific theory that explains how it is that vision gives us reliable access to the shapes and colors of objects around us. In reliably arriving at that theory we will have exercised the very faculty whose reliability is explained by the theory. But there is no problem of vicious circularity that I can discern. A quite general theory of knowledge would be like this theory of visual knowledge in the relevant respects. We would exercise various epistemic faculties in arriving at a theory concerning our nature and surroundings, based on which we could see those very faculties to be reliable. The theory would presumably go into more or less detail in explaining the constitution of these faculties and of how they fit us to gain access to relevant domains of facts. Here we would just have a more general example of the same phenomenon of circularity that we found in the specific case of visual knowledge. And here again I can discern no vice in the circularity that is clearly inevitable for such a general theory by means of which we gain the relevant general understanding. Moreover, we surely are not only armchair thinkers, but also knowers of the world around us. And our knowledge of that world is as dependent on our sensory perception as is our knowledge of the abstract realm of reason dependent on our

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



ERNEST SOSA

direct intuition. I see no good reason deriving from any sort of vicious circularity applicable to our epistemological reliance on perception that would not apply equally to Descartes’ epistemological reliance on intuition. If we are denied a philosophically desirable assurance by that sort of vicious circularity, then so would Descartes be denied such assurance by a precisely parallel circularity. Moreover, this would have been so even if every point of theology that he relied on had been perfectly true, and even if he had correctly intuited every truth that he thought himself correctly to intuit as he meditated along. If, in such circumstances, Descartes still would have fallen short of proper philosophical assurance, I have trouble understanding what this desideratum amounts to, or why we should really care to attain it.

3 Conclusion In conclusion I would draw a distinction among competences between those that are and those that are not fundamental. Thus our visual ability to tell that the bells are tolling can gain support from our auditory ability to tell that they are tolling, and vice versa. Even if we could gain any support at all from circular, blatant bootstrapping for our ability to tell the contents of our fuel tank by just reading our fuel gauge, which is quite doubtful, we are far better off if we can test the gauge independently, by direct perception of the contents of the tank, or through some other means. There are practices or faculties, by contrast, not checkable independently. Which are these? They would be practices or faculties that are ‘fundamental.’ Is sense perception one such? Not clearly. Again, hearing is testable through vision, and vice versa. Plausibly, then, sense perception, in each of its modes, can perhaps after all be checked independently. In another way, however, it cannot. This may be seen as follows. Assume that we have a set of perceptual faculties each of which is checkable through one or more of the others, although none is checkable in any other way. In that case, perception as a whole is not checkable independently of perception (of some perception or other). We can hardly object to the use of perception on that basis, however, nor can we plausibly hold on that basis that there’s no way for us to properly judge, concerning any variety of perception, whether or not it is reliable. After all, there’s bound to be some total set of faculties or practices none of which is checkable except through the use of one or more of the others. This total set, call it ‘fundamental human cognition,’ is then not checkable independently of ‘fundamental human cognition.’ But we can hardly object on that basis to the use of any fundamental human cognition, nor can we plausibly conclude that no fundamental human cognitive practice or faculty could ever be properly judged reliable. It would of course be an accomplishment to reduce perceptual cognition to more fundamental cognition, say to armchair cognition, including introspection and pure reason. However, as Descartes saw so clearly, armchair cognition is not properly endorsable except through the operation of (theological) armchair cognition. So, it is

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

HOW OUR KNOWLEDGE SQUARES WITH SKEPTICAL INTUITIONS



not reducible to something more fundamental. The objection to a mode of cognition then cannot possibly be that it is not endorsable except through its own operation. What then would be the proper response to (a) our perceived inability to endorse even armchair cognition that way, along with (b) our perceived inability to endorse empirical, scientific cognition through the operation of armchair cognition. One possible response would be to reject empirical, scientific cognition as groundless. But once we see that there is no hope of further grounding for armchair cognition itself, once we suffer an anti-Cartesian disillusionment, it seems just arbitrary to hold on to armchair cognition while rejecting empirical, scientific cognition wholesale. It is crucial, however, to be clear on why we see no further grounding for armchair cognition itself. It is not because the Cartesian Circle is inevitably circular, viciously so. It is only because the theological reasoning on which Descartes rests his case is unacceptable, and we see little hope for any acceptable substitute that will restrict itself to the armchair. And then, given this, we might still reasonably hold out for an endorsing perspective that we can access through a more inclusive fundamental human cognition, and that will enable us to endorse inclusive human cognition, both armchair and empirical. This perspective of course is not accessible through a single flash of insight. If it is to a sufficient extent accessible at all, it will be so only through the patient study of human cognition that joins together the efforts of the special sciences and, so we can hope and expect, the insights of a reasonable epistemology.11

References Cottingham, John, Robert Stoothoff, and Dugald Murdoch (eds). (1986–91). The Philosophical Writings of Descartes (3 vols.). Cambridge: Cambridge University Press. Fumerton, Richard. (2011). “Reflective Knowledge and Intellectual Assurance,” International Journal for the Study of Skepticism, 1: 113–23. Sosa, Ernest. (2007). A Virtue Epistemology: Apt Belief and Reflective Knowledge. (Vol. 1). Oxford: Oxford University Press. Sosa, Ernest. (2011). Reflective Knowledge: Apt Belief and Reflective Knowledge. (Vol. 2). Oxford: Oxford University Press. Sosa, Ernest. (2013). “Descartes and Virtue Epistemology,” in Kelly Clark and Michael Rea (eds), Reason, Metaphysics, and Mind: New Essays on the Philosophy of Alvin Plantinga. Oxford: Oxford University Press, 107–22. Sosa, Ernest. (2014). “Descartes’s Epistemology,” in Dylan Dodd and Elia Zardini (eds), Scepticism and Perceptual Justification. Oxford: Oxford University Press, 13–30. 11 Here I have drawn very selectively from several dispersed publications for a unified response to skepticism that goes well beyond my previous main publications, such as, Reflective Knowledge (Oxford University Press, 2011) and the concluding chapters of A Virtue Epistemology (Oxford University Press, 2007), and Knowing Full Well (Princeton University Press, 2011). I have drawn very selectively, and with permission, from my contributions to symposia on my work in Critica (2010), International Journal for the Study of Scepticism (2011), Philosophical Papers (2011), Synthese (2012), and Philosophical Studies (2011).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

Afterword

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

12 The Prospects for Traditional Internalism Richard Fumerton

1 Introduction It may be a bit presumptuous to describe any particular view as traditional internalism.1 There are different ways of defining the internalism/externalism debate and even within the broad category of internalism, there are importantly different variations on the relevant themes. Many of us practice ‘charitable’ interpretation of historical figures by reading into those figures our own views. I do, in fact, think that most of the radical empiricists in the modern era and their twentieth-century intellectual descendants explicitly or implicitly assumed a version of foundationalism that leaned heavily on the idea that foundational knowledge and justified belief critically requires a direct awareness/confrontation/acquaintance with the world, an awareness that should not be identified with belief but which can help confer justification on a belief. It is this version of foundationalism that I am going to call traditional internalism. The view is probably best thought of as a version of internalstate internalism, but as we develop the view we shall have occasion to note the senses in which that label might be misleading. In what follows I’ll begin by discussing noninferential justification. I’ll then move on to a discussion of inferential justification, concluding with some reflections on the prospects traditional internalism has for meeting the challenge of radical skepticism.

2 Noninferential Justification and Acquaintance Let me first lay out the central features of what I still take to be the most plausible version of foundationalism, one that relies critically on the concept of direct 1 I want to thank all of the contributors to both this volume and the May 2014 Orange Beach Epistemology Workshop that was sponsored by the University of South Alabama. These philosophers are all not only outstanding philosophers, but also good friends, and I always profited enormously from both the formal and informal discussion we have had over the years. I would particularly like to thank Ted Poston for the workshop and Michael Bergmann and Brett Coppenger for their time-consuming work on the volume.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



RICHARD FUMERTON

acquaintance. My view continues to evolve. In no small part, the authors who contributed the excellent chapters in this volume have led me to see that I need to clarify some aspects of my views and to revise others. I have suggested on many occasions the following sufficient condition for noninferential justification: S has noninferential justification for believing P if S is directly acquainted with the fact that P, the thought that P and a relation of correspondence holding between the thought that P and the fact that P. I don’t say here that direct acquaintance with the truth maker for a thought is necessary for noninferential justification as I want to leave open the question of whether one can sometimes acquire at least some justification for believing P when one is directly acquainted with a fact that is not the truth maker for P (and is, thus, as Tucker (Chapter 2) points out, one of the truth makers for not-P—I’ll say more about this below). The above is an account of what it is to have what is sometimes called propositional justification, where that is contrasted with doxastic justification. Doxastic justification is a property of a belief. While I won’t argue the point here, it seems to me that one has doxastic justification for believing P when one believes P and the belief that P is based on the propositional justification available to one. And while it is highly controversial, it still seems to me that basing is best understood in terms of what causally sustains (in the right way)2 the belief. I have argued in a number of places that the epistemologist, qua philosopher, should be primarily interested in propositional justification—in the question of when one’s epistemic situation would justify one in believing a given proposition (whether one forms the belief or not). If basing is to be understood in terms of causing, and questions about what causes what are empirical questions, I suspect most epistemologists aren’t particularly competent to evaluate the relevant empirical hypothesis about the causal origins of their beliefs. Furthermore, most analyses of causation render facts external to the subject relevant to whether or not some internal state causes a belief. One should be extremely wary, therefore, about endorsing an internalist thesis with respect to the doxastic justification of a belief. To accept anything like the above acquaintance theory of noninferential justification one needs to buy into a number of highly controversial ontological commitments. The first, of course, is that there is a relation of acquaintance and that one can stand in that relation to a fact (a thing’s or things’ exemplifying a property or properties). While it would be nice to provide an analysis of acquaintance, it won’t come from me. On my view, the idea of acquaintance is one of those indefinitely many simple ideas out of which more complex ideas are formed. We can try to ‘point’

2

One sweeps under the carpet many difficult problems when one protects a causal analysis this way.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

THE PROSPECTS FOR TRADITIONAL INTERNALISM



to it. We think about a pain of which we were aware, how when we were temporarily distracted we didn’t notice the pain, and how we then became aware again of the pain after the distraction ended. But there are alternative descriptions of this familiar phenomenon that will compete with this attempt to ‘ostend’ the relation of direct acquaintance.3 As I understand acquaintance, it is not an intentional state, at least if we understand intentional states as those that can be ‘directed’ at objects that ‘do not exist’ (as we can fear ghosts, hope for an inheritance that never comes, seem to remember having put one’s keys on the table when one didn’t, and so on). Acquaintance is a real relation that requires relata. I cannot be acquainted with a pain that does not exist (though perhaps I might be acquainted with some state similar to pain and believe falsely that I am in pain). For this reason I resist strongly Steup’s (no doubt friendly) suggestion (Chapter 3) that I construe acquaintance as a mode of ‘seeming.’ Whatever else ‘seemings’ are, they are intentional states in the sense sketched above— something can seem to me to be the case that isn’t. The concept of a fact as a truth maker may be just as controversial as the concept of acquaintance.4 Many philosophers take talk of facts simply to be talk of truths. And certainly in ordinary discourse there isn’t a whole lot of difference between saying that it is true that P and saying that it is a fact that P.5 My view presupposes a very specific version of a correspondence theory of truth (see Fumerton 2002) according to which facts are structured complexes that are independent of their being represented, but have the capacity to make true thoughts that represent them. While Tucker (Chapter 2) has convinced me that I should modify somewhat my views about noninferentially justified false belief, I do think that realizing that one can be acquainted with a fact without representing it all, or while representing it with a thought that doesn’t correspond strongly to it, makes slightly less puzzling my suggestion that being acquainted with a ‘not-P’ fact can still justify me to some extent in believing P. If acquaintance with a fact were knowledge of a truth, the previous suggestion would make no sense. But I can be acquainted with a fact without having settled in my mind what fact it is, indeed without representing it at all.6 So my view is committed to a relation of acquaintance and to facts as truth makers. It is also committed to correspondence as a relation between a thought and fact that is the thought’s being true. On this view, there are no truths without thoughts, facts, and a relation of correspondence holding between thoughts. The correspondence 3 For one thing one could argue that if we didn’t notice the pain at all the pain ceased. Alternatively, one could try to understand all this talk of noticing in terms of occurrent or dispositional belief. 4 See Beebee and Dodd (2005) for a relatively recent collection of papers exploring various issues that arise concerning truth makers. 5 Where the latter might be a particularly emphatic way of asserting truth. Similar facts about ordinary language no doubt contribute to so-called redundancy theories of truth. Asserting that it is true that P itself can seem to be just an emphatic way of asserting that P. 6 I suspect that many animals are acquainted with facts and lack the means altogether to represent those facts.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



RICHARD FUMERTON

relation is internal, however, and will necessarily hold given the existence of its relata. I don’t think one can give a naturalistic analysis of correspondence—it is yet another fundamental building block out of which philosophers can construct important analyses. To characterize a property (or relation) as unanalyzable is not to suggest that one can’t say anything interesting about it. If one can make sense of all this talk of correspondence, it seems to me that one will also be able to make sense of correspondence coming in degrees. I have the thought of some paradigmatic way of being red and I am directly acquainted with the exemplification of a color property that exactly corresponds to that thought.7 I then become aware of something a bit closer to orange on the color continuum but that still corresponds relatively well to my thought of red. As we get closer and closer to the orange part of the continuum, the correspondence gets weaker and weaker and the justification I have for believing that the object is red itself gets weaker and weaker. It is this sort of example that leads me to think that one can make sense of someone’s having some noninferential justification for believing that the object is red even when it isn’t. As what I’m aware of just barely crosses the border into the orange territory, doesn’t it at least seem right to suggest that I have more justification for thinking that it is red than that it is light blue? I haven’t always put the above point this cautiously, however. I have on occasion just said that one can have a noninferentially justified but false belief. And one might naturally infer that I am committed to the view that one can have noninferential justification that makes epistemically likely for me the truth of a false belief. As Tucker (Chapter 2) points out, if I really am acquainted with a fact that is, by hypothesis, a truth maker for the thought that the object is not red, while I am also acquainted with the (not very strong) correspondence between the thought that the object is not red and that fact, then surely I have a bit more justification for believing that it is not-red than that it is red. That now seems right to me, although the matter might be more complicated if the person in question doesn’t possess the conceptual resources to form the negative thought. It is also harder to figure out just what to say about how a correspondence theorist should understand a negative thought’s corresponding to the world, an issue that raises the question of whether there are negative facts. Is the negative thought that an object is not red the kind of thought that corresponds to its being blue, its being yellow, its being brown, etc., or is there simply a negative fact, its being not-red, that is the relevant truth maker and sometimes the object of awareness? These are important and difficult issues lurking in the background. I do want to emphasize in the context of this discussion how misleading it might be to talk as if there are sharp boundaries in examples like that above between the truth of P and the truth of not-P. On my view, thoughts are the

7

Let’s suppose for simplicity that we are talking about a sense datum that exemplifies phenomenal redness. I don’t think one can be acquainted at all with redness as a property of a physical object.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

THE PROSPECTS FOR TRADITIONAL INTERNALISM



primary8 bearers of truth value, though a thought’s being true is a complex relational state of affairs. I strongly suspect that there are at least subtle differences between your thought of red and my thought of red, and further that my thought of red is almost certainly subtly unstable through time, even, perhaps, from moment to moment. And at a time, as the correspondence between a thought and a fact is very weak, it does seem to me that one acquainted with the correspondence is often making more of a decision than a discovery with respect to whether or not the relevant thought is true. I have claimed that one has noninferential justification that P when one is acquainted with a relation of correspondence between one’s thought that P and the fact that makes it true. Markie (Chapter 1) thinks that on this sort of view it is obvious that an acquaintance theory should allow that one has (propositional) evidence that P when one is acquainted with the fact that P whether or not one forms the thought that P or notices a relation of correspondence between the thought that P and the fact that P. Indeed, he is highly suspicious that there is such a state as being acquainted with correspondence. He may be right, but on the view I am defending, acquaintance with one’s pain is not, by itself, evidence of any truth—it is not a source of epistemic justification. I think that Sellars was right and that if one is going to talk of justification at all one needs to bring into the picture truth and truth only enters the philosophical landscape when thought or representation does. We want to get at truth with justification and one can’t get at truth directly without being related to truth bearers, truth makers, and truth making. To some extent this may be just terminological. A similar controversy arises when considering inferential justification. Before pointing out the guilty party, the fictional detective Hercules Poirot was fond of pointing out to his audience that they already had all of the evidence needed to solve the mystery. I assume he meant that they already knew various propositions, which propositions did make probable that, for example, the butler committed the murder. But it is important to distinguish two quite different situations—one in which they knew the relevant propositions and had at least the capacity to grasp on reflection the probabilistic implications of that evidence, and one in which they lacked the great detective’s power of reasoning. In the latter case, I don’t want to say that they had evidence or propositional justification for believing that, for example, the butler is the guilty party. In exactly the same sense I don’t want to say that I have unused evidence or justification for believing every necessary truth (including those that mathematicians win Nobel prizes for discovering). In the case of inference, I don’t have inferential access to truth until I ‘see’ the evidential

8 So on one classical model, we will also allow, for example, sentences, statements, and beliefs (among others) to be true or false. But a sentence or statement is true because it conventionally expresses a thought. A belief can be true because it is a species of thought. Everything with a truth value has that truth value only because a thought has the truth value.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



RICHARD FUMERTON

connections. In the case of noninferential justification, I don’t have the relevant access to truth until I am aware of the truth-making relation.

2.1 Acquaintance and mental states The account of noninferential justification offered above is in principle neutral as to what kinds of beliefs can be noninferentially justified. Markie (Chapter 1) suggests that I believe that one can only be directly acquainted with mental states. I’m actually tempted by that position, but only if one is clear that mental states include facts that involve relations between mental states (like the relation of correspondence between the thought that I am in pain and my pain), and only if the truth makers for necessary truths are relations between thoughts (the primary bearers of truth value). But I’m keenly aware of how controversial this position is. I’m not sure whether or not there are universals, but if there are, the Russelian view that one can be acquainted with them and relations that hold between them is surely attractive. One also needs an argument for rejecting the view that one can be directly acquainted with facts about the physical world—such facts as that there is a physical object with a certain shape before me now. I think there are such arguments. They typically require one to reject at least radical disjunctivism about the nature of justification as they rely heavily on the intuition that in classic skeptical scenarios, one has the same sort of justification for one’s beliefs that one has in the non-skeptical counterpart. So if all of my experiences were caused directly by the demon Descartes asked us to imagine, it wouldn’t alter one bit the justification (if any) I have for believing what I do about the world around me. Since the justification in the ‘bad’ case can’t be direct acquaintance with the relevant truth makers (they don’t exist), that same sort of justification in the ‘good’ case can’t be construed as involving that sort of direct acquaintance either.

2.2 Acquaintance and seemings Obviously, one doesn’t have to be an ‘acquaintance’ foundationalist to embrace the kind of internalism that rejects disjunctivism. Phenomenal conservatists like Huemer and Brogaard identify noninferential justification with the occurrence of certain intentional states (see Huemer (Chapter 7) and Brogaard (Chapter 4)). Schellenberg (Chapter 5) has a more complicated view, but she certainly wants to allow that in the skeptical scenarios one still possesses at least some evidence for the beliefs produced by non-veridical experience. And given her view it is hard to resist the conclusion that in the final analysis it is the alleged intentional character of sense experience (hallucinatory or not) that is supposed to give rise to epistemic justification (in the case of Schellenberg, the existence of epistemic justification in the ‘bad’ cases is still supposed to be parasitic upon the way in which sense experience ‘succeeds’ in picking out objects in the good cases, but because she is not willing to argue that there ever needs to have been actual veridical experience in order for this sort of justification to exist, one wonders if the view isn’t really closer to phenomenal conservatism than it might initially seem to be).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

THE PROSPECTS FOR TRADITIONAL INTERNALISM



Back when sense-datum theories were popular, epistemologists often worried about the way in which sense data formed a ‘veil’ between what we are directly aware of and the physical world. Frankly, I’m ultimately sympathetic with the idea that we have no access to the intrinsic character of physical objects other than what we can legitimately infer (if anything) from the appearances that we think are caused by those objects. But on an acquaintance theory we are supposed to be able to take at least some comfort from the fact that we have direct access to some aspects of reality—at the very least the character of our phenomenal experiences. On virtually all forms of conservatism (traditional or phenomenal),9 there is an even more impenetrable veil that always hangs between us and everything. Our only access to anything—property, state, or fact—is through an intentional state that (correctly or incorrectly) represents it. And that claim covers those intentional states as well. Our only access to first-order intentional states is through second-order intentional states that (correctly or incorrectly) represent them. Our only access to second-order intentional states is through third-order intentional states that (correctly or incorrectly) represent them. And so on ad infinitum. All of reality endlessly recedes as we try to contact it directly through consciousness. The truth makers lie forever beyond our epistemic grasp.

2.3 Speckled hens and acquaintance In characterizing noninferential justification above I offered only a sufficient condition for one’s having noninferential justification for believing some proposition. Nor did I claim that when one has a sensation one automatically gains noninferential justification for believing every truth that describes that sensation. No one has thought, for example, that as one has a sensation one gains epistemic justification for believing that the sensation occurs at the same time that Vladimir Putin is sneezing. But even if we were to restrict our attention to the non-relational properties of sensations, or those relations that present sensations stand in to each other, it still seems far too strong to claim that one possesses justification for believing all of the truths made true by those features of our experience. Markie (Chapter 1) and Sosa (Chapter 11) have argued that this admission poses insuperable problems for an acquaintance theory of noninferential justification. Consider the well-known problem of the appearance presented by a speckled hen. The appearance is as of a surface that is many speckled. Let us also suppose that there is some determinate number of speckles, say 47, characterizing the visual sense experience.10 Unless one has ‘Rain 9 The traditional conservative takes the mere fact that someone believes P to provide that person with prima facie justification for believing P. The phenomenal conservative argues that there is an intentional state distinct from belief—its seeming to one that P—that provides the prima facie justification. 10 It is not easy to discuss this controversy without appearing to beg critical metaphysical controversies concerning the nature of sensations. We are not talking about the number of speckles on the surface of the hen, at least on the assumption that we are not directly acquainted with that surface. But when we move to the characterization of subjective sensation there are disagreements among the sense data theorists, the

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



RICHARD FUMERTON

Man-like’ abilities,11 one will not have epistemic justification for believing that one’s visual experience has those 47 speckles, even if one does have epistemic justification for believing that one is appeared to ‘many speckled-ly.’ Here and elsewhere (1995, 2009, 2011), I have admitted that a defense of an acquaintance theory inevitably involves us in difficult metaphysical issues. I also emphasized that even if we were acquainted with some determinate property exemplified in a sensation, that wouldn’t by itself have any epistemic significance for whether one has justification for believing a proposition made true by that feature. The justification is constituted only by that awareness coupled with awareness of some degree of correspondence between thought and fact. To figure out what best to say about the speckled hen appearance (and indefinitely many other puzzles like it) we will need to figure out whether there are both determinate and determinable properties simultaneously exemplified (e.g. being cherry red, being dark red, being red, and being colored). We should also at least evaluate the question of whether a determinable property (e.g. being red) can be exemplified without a determinate property being exemplified. And if my view is right, we critically need to figure out whether one can be directly aware of the exemplification of a determinable without being aware of any of its determinate properties. And however promiscuous we are in postulating awareness of properties, none of that gives us noninferential justification unless we have located in awareness thoughts and their correspondence with those properties. Within this byzantine range of metaphysical positions there are a number of ways for an acquaintance theory to explain how the view leaves open the question of whether having the many-speckled experience will give anyone noninferential justification. By way of conciliation, Markie suggests that we could find a robust epistemic role for acquaintance to play by construing that role as one that needs to be understood in terms of corresponding abilities or capacities. There is a sense in which I have no problem with that suggestion as long as we construe the capacities (contra Markie) as abilities to hold before one’s mind the constituents of truth making that constitute noninferential justification.

2.4 Noninferential justification and assurance I haven’t said much by way of arguing for the account of noninferential justification described above. Nor have I given any arguments for accepting the foundationalist’s view about the structure of justification. In addition, I have only hinted at my reasons

appearing theorists, and the intentionalists with respect to how to talk about sensation. The adverbial theorists prefers cumbersome technical locutions describing ways of being appeared to—e.g. being appeared to 47 speckledly. The sense datum theorist thinks that the visual field literally contains those 47 spots. The intentionalist thinks of sensation in terms of what they represent—that 47-speckled surface of the hen. Although I am sympathetic to the adverbial theorist’s position, for ease of exposition I’ll sometimes use language more appropriate to a sense datum theory. 11

The reference, of course, is to the central character of the movie Rain Man, a victim of autism who had uncanny abilities to ‘pick up’ on facts about the number of things presented in experience.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

THE PROSPECTS FOR TRADITIONAL INTERNALISM



for rejecting externalist accounts of justification (both noninferential and inferential). The traditional internalist is convinced that externalist accounts of justification fail to satisfy the intellectual curiosity that drives philosophers to worry about epistemic justification in the first place. One can get a feel for the worry by thinking about the crudest possible form of externalism. Suppose we are told that we have justification for a given belief just in case the belief is true—otherwise it is unjustified. There might be something good about having a true belief and something bad about having a false belief, but if I am wondering whether there is justification for me to believe that there is a God, that there is life on other planets, that there is a world whose features are largely what I take them to be, it is hardly reassuring to be told that all is well epistemically if my beliefs are true—things are going badly if my beliefs are false! Of course, paradigmatic externalist accounts of justification are far more sophisticated. We are told that we have noninferential justification for believing that P if our belief that P is caused by some process that typically results in true belief (where the ‘input’ doesn’t involve the having of other justified beliefs). Or we are told that we have noninferential justification for believing that P when our belief that P is caused by its truth maker, or when our belief that P ‘tracks’ the truth of P, but not via other justified beliefs. But the internalist typically feels just as frustrated by these accounts of justification as by the crudest ‘truth-justification’ theory. We can easily imagine just about any belief ’s being reliably produced, being caused by its truth maker, tracking truth, and so on, without these features of belief being the slightest bit relevant to the question we’ve asked ourselves about justification. I’ve sometimes summarized the above concern in the following way. Qua philosophers we are interested in a concept of justification that brings with it assurance of truth. To be sure, we mustn’t confuse this concern with the issue of when we have second-order justification that we have a justified belief. The conditions for having a justified belief are distinct from the conditions for having a justified belief that we have a justified belief. Contra Steup (Chapter 3), I categorically reject all access requirements on having a justified belief (an issue to which I will return when we talk more about inferential justification). The relevant issue is whether one can develop an account of justification that brings with it assurance of truth. But how are we to understand all this talk of assurance? Here I confess that there is a real sense in which I am substituting rhetoric for argument.12 We surely can’t understand assurance psychologically, as for example subjective certainty or the absence of doubt. For all I know, one can manufacture drugs that bring about complete confidence and the absence of doubt with respect to something one believes. And I certainly wouldn’t argue that the way to acquire philosophically relevant justification is to ingest such drugs. The internalist’s plea that we develop a concept of epistemic justification that brings with it epistemic assurance may 12

Coppenger’s (2014) paper at the Orange Beach Epistemology Workshop is extremely helpful sorting through these issues.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



RICHARD FUMERTON

amount to nothing more than the plea that we develop a plausible account of epistemic justification that is clearly relevant to our epistemic curiosity. But once we admit that, we probably are begging the question against our externalist epistemic counterparts who will, no doubt, argue that they have successfully and plausibly analyzed the very epistemic concept with which we are concerned. There is a slightly different way of approaching our question, however. From an epistemic (philosophical) perspective, do we want to do better than having mere true belief? Almost everyone agrees that we do. From that same epistemic perspective, do we want more than having a belief that is caused by its truth maker? It is surely not hard to convince ourselves that there is more to having justification than that. And the empirical question of whether a belief is reliably produced or tracks its truth maker surely isn’t any more relevant to our epistemic position with respect to the truth of some proposition, at least if that epistemic position is driven by curiosity from the first-person perspective. But when we are entertaining the thought that we are in pain, as we are directly acquainted with the very state that makes true the thought and we are acquainted with the truth-making relation, what more could we want? From the epistemic perspective, how could we do any better than that? That’s the thought that is supposed to make attractive the acquaintance theorist’s conception of noninferential justification.

3 Inferential Justification In the spirit of Descartes we started with an attempt to get clear about noninferential justification.13 I didn’t give an argument for the conclusion that there is such a thing as noninferential or foundational justification. Regress arguments are still surely persuasive for the conditional conclusion that if there are any justified beliefs then there are noninferentially justified beliefs. But I have argued that there are two importantly different regresses about which one might be worried—an epistemic regress and a conceptual regress. On the assumption that one can only justifiably infer a proposition from another proposition if one has justification for believing the latter, it is hard to see how a finite being could ever complete the infinitely long chain of reasoning that would be required to have doxastic justification. But if one focuses on propositional justification, the issue is a bit more complicated. Arguably, finite beings not only can have, but do have an infinite number of justified beliefs.14 13 Descartes is an interesting figure in the history of the debate between internalists and externalists. Sosa (Chapter 11) claims that he has strong reliabilist leanings. And of course this is, in a sense, right. He wanted justification so strong as to preclude the possibility of error. Beliefs produced in a way that met his standards would be 100 per cent reliable. But he also clearly wanted an ‘internal’ guarantee of truth. He wanted something that would satisfy both internalists and hyper-extreme reliabilists. He didn’t get it and it is not clear to me what he would have settled for if he realized that he had to choose. 14 Though it is not clear that they would have the structure necessary to provide the relevant inferential justification. The most obvious examples of infinitely many justified beliefs involve deductions from justified belief in general propositions.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

THE PROSPECTS FOR TRADITIONAL INTERNALISM



In any event, as I have frequently argued (e.g. 1995, 2014), it might be more promising to emphasize the need to avoid conceptual regress. One can only understand inferential justification in terms of the availability of inference from justified belief. Because that is so, we need to define inferential justification recursively, and the recursive definition requires a base clause that defines a kind of justification that is not inferential. Epistemic controversies over the structure of justification are, of course, much more complex than the above cryptic remarks suggest. Coherence as a criterion for justification has never seemed to me plausible, if only for the reason that there are devastating counterexamples to the idea that coherence without access to coherence is sufficient for justification. But to be honest, I’m not sure that it is the availability of arguments for foundationalism that make so plausible the idea that there is noninferential justification. To the radical empiricists it seemed just obvious that we didn’t need to infer truths about the subjective character of our experience. And if you ask yourself what justifies you in believing that you are in pain as you are aware of that very experience, the idea that you would turn to evidence from which you infer the proposition is downright bizarre. The acquaintance theory is supposed to capture the idea that direct confrontation with a truth maker obviates the need for anything remotely resembling inference from other propositions believed. But even if an acquaintance theory captures a plausible concept of noninferential justification, it also seems obvious that if we are to get justification for most of our ‘commonsense’ beliefs we will need to move beyond our foundations through inference. Furthermore, if one restricts one’s foundation to ‘demon-proof ’ justification (more about this later), the road to commonsense will obviously be long and complicated. When talking about inferential justification it is useful to begin with Huemer’s (Chapter 7) important distinctions between ways in which one’s justification for a belief might depend on the having of another justified belief. With Huemer (Chapter 7) and Dougherty (Chapter 6) I’m inclined to agree that paradigmatically conscious inference from all of the propositions evidentially crucial to one’s justification is rather rare. And I have insisted from time to time that when I talk about inferential justification, I’m talking about an intellectually demanding philosophical ideal to which we might aspire, but will often fall short. I also think that it is possible to develop derivative ‘degenerate’ concepts of inferential justification that will allow us to make sense of ordinary ascriptions of justified belief. In that respect my approach bears at least a superficial similarity to Sosa’s. But while Sosa builds a more intellectually satisfying sort of justification by adding to conditions for a less demanding sort of justification, I start with an ideal and allow for less demanding justification that results from stripping elements from ideal justification. The distinction isn’t merely verbal, for our accounts give radically different verdicts with respect to even the less demanding justification when it comes to questions of what we are justified in believing in the skeptical scenarios (see Fumerton 2004).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



RICHARD FUMERTON

For ideal inferential justification I have defended the following principle: (PIJ) One has justification for believing P on the basis of E only if one is justified in believing E and one is justified in believing that E makes probable P. I have also argued that the only way one will avoid skepticism with respect to this kind of justification is if one can know a priori when one’s total evidence makes probable a given proposition—otherwise regress looms (see, for example, Fumerton 1995: ch. 7, 2006, 2014). This conclusion might seem devastating to the prospects of avoiding a fairly radical skepticism as it may seem obvious that claims about truths making probable other truths are both contingent and empirical. The color of litmus paper might indicate the acidity of a solution in this world, but it surely doesn’t in other possible worlds. That, of course, seems right, but, strange though it sounds, I would argue that propositions describing the color of litmus paper aren’t evidence from which one can infer the relevant truths about the solution containing it.15 Such reasoning is enthymematic and includes contingent premises describing correlations between the behavior of litmus paper and the nature of solutions. When the critical premises are made explicit, it is not nearly as obvious that the support relation between the premises and the conclusion is a relation that holds only contingently. If one can make sense of this Keynesian notion of making probable as an internal relation holding between propositions, one might be able to block regress and satisfy the admittedly demanding conditions I propose for ideal, intellectually satisfying inferential justification. Huemer agrees with me that the mere existence of inferential connections between evidence and a conclusion isn’t enough to secure justified belief in the conclusion when one has justification for believing the propositions that comprise one’s evidence. As one might expect given his other views, however, he suggests that we add to the conditions for inferential justification of this sort the requirement that it seems to one that the premises make probable the conclusion (though he also warns, correctly, that one must not think of the proposition that describes that attitude as an additional premise supporting the conclusion). Recently, under the influence of a former student, Sam Taylor (2013), I’m much more sympathetic to a third alternative that requires for inferential justification only awareness or acquaintance with the evidential connection (see Fumerton, Forthcoming). The different approaches to understanding this sort of ideal inferential justification might seem to reflect only the antecedent disagreement about whether it is acquaintance or ‘seeming’ that is the linchpin of justification. But they also have different implications. Because acquaintance is not an intentional state, one can’t be acquainted with a probabilistic connection unless it exists. Because ‘seeming’ is an intentional state, it can seem to one that E makes probable P even when it doesn’t.

15

For seeing this more clearly, I am indebted to Huemer’s (2002).

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

THE PROSPECTS FOR TRADITIONAL INTERNALISM



And one might suppose that this favors Huemer’s view for he gives us an initially plausible example of the unfortunate mathematician who gains a justified belief by completing a proof from justified premises when it seems to him (after careful reflection) that the premises entail the conclusion. Even if the mathematician made a mistake, the argument goes, we are surely inclined to grant that he had an epistemically justified belief. The issue, though, is complicated. If I want to accept Huemer’s conclusion about the case, I probably won’t be able to borrow a leaf from what I said about noninferentially justified false belief—I don’t know what relation would be available as one that is ‘close’ to entailment. But as Huemer would anticipate, we might turn to an entirely different sort of justification available to the mathematician. Mathematicians have, after all, performed many such proofs and seem to remember that when premises seem to them to entail conclusion they have almost always independently proved correct to have trusted the seeming. Our intuition that the unfortunate mathematician has a justified belief might be traceable to the realization that there was a back-up justification for the belief. In the end, I think that the position one should take on this (and other similar examples) really should depend on whether one is sympathetic to phenomenal conservatism. I’ve argued against the view elsewhere (2007), but there is no denying that of the hard-core internalist epistemologies, the phenomenal conservatists (and their close cousins, the more traditional epistemic conservatists) have the easier road to travel in responding to skepticism.

4 Prospects for Traditional Internalism 4.1 Traditional internalism and skepticism I have said on many occasions that one shouldn’t let one’s metaepistemology be driven by the desire to refute skepticism. The skeptic’s worry that external physical reality (if it exists at all) might not be much like what we take it to be is a distinctively philosophical worry, albeit one exacerbated by ongoing scientific hypotheses that seem incompatible with much of commonsense.16 On the other hand, I am sensitive to Bergmann’s (2006) and Sosa’s (Chapter 11) worry that we shouldn’t perversely define our way into a foregone skeptical conclusion. Pritchard and Ranalli (Chapter 10) compare my views with Stroud’s. Stroud’s view is harder to interpret 16 It may be that we are so used to the posits of theoretical physics that we no longer find puzzling Eddington’s discussion of two tables—the one that appearance leads us to think of, and the one that has little to do with anything dictated by commonsense. But Eddington’s discussion reinforces in a particularly vivid way Berkeley’s constant reminder that we can’t trust appearance as a reliable guide to how things are, if only for the reason that appearance varies so radically under different internal and external conditions. To be sure, the less we build in to the hypothesis that physical objects exist, the more we are content to think of them as those entities whatever they are responsible for various patterns of experience, the more we can reconcile ‘commonsense’ with the vastly different appearances presented by the physical world about which we think.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



RICHARD FUMERTON

now than it was when he seemed to argue for a skeptical conclusion in 1984. In his early discussion of skepticism (and I still see this thread in his later work), Stroud seemed to insist that to thwart skepticism in a non-circular way, we need some way to certify the legitimacy of all of our ways of forming beliefs, and we need to do so without presupposing the legitimacy of any of them. As Bergmann (2006) and Sosa (Chapter 11) argue, that condition for refuting skepticism can’t possibly be met. Even God wouldn’t be able to satisfy this condition on intellectually satisfying epistemic justification! By contrast, rightly or wrongly, I have argued that when one satisfies my conditions on noninferential justification, and ideal inferential justification, one has all the justification one could have or want by way of satisfying intellectual curiosity. But this claim is conditional, and I concede that it is not easy to satisfy the conditions for intellectually satisfying inferential justification. Goldberg (Chapter 8) tries to convince us that the kind of internalism I defend puts unrealistic and implausible demands on justification. He correctly observes that my kind of internalism (I bear a suspicious resemblance to the character Ricardo in his chapter) wants justification that is ‘demon-proof.’ Descartes was satisfied that he found at least one secure plank in the foundations of knowledge when he realized that even the most powerful demon imaginable couldn’t deprive him of his justification for believing that he exists. Indeed, my justification for believing that I exist is so strong that it precludes even the conceivability of error. But Descartes’ hypothetical demon can surely deprive us of true belief (if not justified belief) in propositions about our past, our future, the external world, other minds, and, of course, all of the hypotheses of contemporary science. Goldberg is understandably worried about the cost of insisting that the only real epistemic justification is ‘demon-proof ’ justification. Like Goldman, he thinks that this sort of internalism will succumb inevitably to a fairly radical skepticism. Goldberg considers three hypothetical situations. In the first, for a 25-minute period, the demon deprives a woman of all of her veridical memories about the past, replacing them with memory illusions. She is, however, like you, looking out the window and, like you, concludes that the sun has risen. The details of a story like this matter—it wasn’t completely clear to me whether the demon victim shares your apparent memories of background information that might be relevant to having justification for believing that the sun is there, but, in any event, that’s how I’m interpreting the situation. Our demon victim is in a pretty bad situation. She has, by hypothesis lost lots of knowledge about the past and the future. Perhaps I am too much of a confirmed internalist, but I’m afraid that thinking about the scenario doesn’t do much to weaken my conviction that her epistemic situation is nevertheless just like yours when it comes to the question of whether she has a justified belief that the sun is rising. The epistemic rationality of her belief (like yours) is a function of the evidence available to her at the time the belief is formed. Of course, if she lacks the relevant apparent memories, she is not your internal duplicate and all bets are off with respect to what she has justification for believing.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

THE PROSPECTS FOR TRADITIONAL INTERNALISM



In Goldberg’s second situation our demon is messing with a man’s ability to appreciate the legitimacy of inferences (both deductive and non-deductive). He is your internal doppelganger with respect to sense experience but differs from you in that where you think an inference is strong, he thinks that it is weak. Now again, given the description of the case, the demon victim and you are not internal duplicates. But I think the idea is supposed to be that from the subject’s perspective there would be no way for him to discover that he was in the demon-victim context as opposed to one in which he was correctly appreciating the evidential force of his evidence. Put another way, the internalist’s insight is supposed to derive from our strong intuition that if there is a ‘bad’ case indistinguishable from a ‘good,’ where there would be no way of telling from the first-person perspective that you are in the good case rather than the bad, your justification is only as good as it is in the bad case. But the idea is that we have discovered a really ‘bad’ case, one which surely precludes justification for the demon victim, but also surely leaves intact the justification available to the person whose reasoning ability is, by hypothesis functionally properly. The difficulty here is real. It reminds one of Descartes momentarily concluding that a demon could make us believe falsely even elementary propositions of mathematics. The point of raising the possibility was to cast doubt on our justification for believing even the simplest of mathematical truths. In the end, I suppose, I will bite the bullet and deny that the demon-victim and the person genuinely acquainted with the truth makers for simple truths of reason (and I include among those, truths describing certain simple evidential connections) have the same sort of propositional justification. At the same time, I will try to deflect somewhat the rhetorical force of Goldberg’s argument. So on my view, our justification for believing propositions describing simple relations of entailment, for example, is that we are directly acquainted with the truth makers for such truths. That justification is as good as it gets and one couldn’t substitute for it something that is clearly inferior but indistinguishable from it. But a demon probably could make it the case that it seems to you that P entails Q when it doesn’t. Now on my view its seeming to me that P isn’t the same thing as being acquainted with the truth-making relation holding between the thought that P and the fact that P. Further, I don’t believe that its seeming to me that P by itself is a reason to believe P (whatever P is—even if P is a proposition asserting evidential connections between two propositions). On the other hand, strong instinctual belief against a background of other justified beliefs might be a reliable sign of true belief—though here there is a very real danger of one’s ego having too much influence on one’s belief. So the person who has strong instinctive belief that there is a relation of making probable where I, by contrast, have direct acquaintance with a genuine evidential connection might have inferential justification for believing the same proposition that I am noninferentially justified in believing. But the source of our justified belief will not be the same, and my justification will be better than his.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



RICHARD FUMERTON

I would say something similar about Goldberg’s third hypothetical situation. Here my demon-manipulated counterpart has the demon messing with the application of concepts. Consider the example of pain (Goldberg talks instead about visual experience). You feel the very same (kind) of pain as I do—you have the same pain experience. I apply my concept of pain to the experience, but the demon has so messed with your apparent memories of past experiences and your judgments about relevant similarity so as to produce in you the judgment that it is not pain you are experiencing. Now as Goldberg admits, the situation is not one in which you and I are internal duplicates—your apparent memories, judgments of similarity to past experiences, and current judgment that you are in pain are different from mine. But more importantly, I think, it is no longer clear in the example that you and I have the same concept of pain—that your thought of pain* is the same as my thought of pain. Here there are very complicated questions concerning the nature of thought and judgment that must be carefully considered as one tries to develop an account of noninferential justification. The ballgame is pretty much over for the traditional foundationalist if one concedes that in applying a concept one is making a judgment about relevant similarity between what is before consciousness and other past experiences—most foundationalists reject the idea that one can have noninferential justification for believing propositions about the past (more about that below). But the obvious solution is to reject the ‘comparison’ model of making judgment. I might have no recollection of any past pain and still be capable of forming the thought that I am in pain while I notice the correspondence between that thought and the experience of which I am directly aware. You will recall that on my view one’s noninferential justification for believing P is constituted by a direct awareness of correspondence between a thought and its truth maker. As I understand Goldberg’s third example the demon victim’s epistemic situation and mine (assuming that I am not so victimized) do differ with respect to the proposition for which I have noninferential justification. A demon can’t take away my justification for believing the proposition that I entertain and replace it with something that is indistinguishable from the first-person perspective, though that demon may be able to get me to believe something not supported by the justification that is available to me. In such a situation the demon might induce in me a doxastically unjustified belief, but as long as I stand in the relevant relations of acquaintance to thoughts and their correspondence to truth makers, there is still available to me the best justification possible to believe that I am in pain.

4.2 Will we find the future of traditional internalism by thinking about the past? I’ve been trying to develop a form of internalism that allows for a kind of intellectually satisfying noninferential justification. Provided that one can develop a plausible Keynesian account of epistemic probability, the account allows for at least the

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

THE PROSPECTS FOR TRADITIONAL INTERNALISM



possibility of fairly extensive inferential justification. But as many of the contributors to this volume have argued, the threat of skepticism looms heavily over all intellectually demanding internalist accounts of what is required for epistemic justification. As Poston (Chapter 9) points out, nowhere is that threat more apparent than when we consider our justification for believing propositions about the past. To answer fully questions about our epistemic access to the past, we will no doubt need to address extremely difficult issues in the metaphysics of time and the metaphysics of events occurring in time. So I keep talking about awareness of this and that, and awareness is something that seems to take place not only at a time, but also, perhaps importantly, through an expanse of time. Prospects for avoiding skepticism are really dim if the acts of awareness that give rise to noninferential justification are literally momentary, lost in less than the blink of an eye to a past that can only be recovered through inferential justification. But perhaps a temporally extended act of awareness can keep before it a temporally extended experience, say a pain, and a temporally extended correspondence between a thought and experience leaving us with temporally extended justification for a belief. Perhaps. But it will be hard to answer difficult questions about just how long this ‘specious present’ is without the answer sounding completely ad hoc. Russell (1921: 159) famously suggested that the hypothesis that everything came into existence five minutes ago, including us with all of our apparent memories, is perfectly intelligible. If we employ classical internalist criteria for determining what can count as a noninferentially justified belief, and we agree with Russell, it looks as if we would have access to a past five minutes older than our present experience only through inference. After all, if our history is as we think it is, what reason could we have for believing that the sun rose yesterday that our five-minute-old doppelganger wouldn’t have as well. Of course, in the factive sense of ‘memory,’ we can say that we remember that past event while our counterpart does not. But we have rejected the epistemic relevance of that sort of distinction elsewhere. If we turn to disjunctivism here, we have lost the epistemic high ground when it comes to thought experiments involving demon-induced hallucination. If my doppelganger and I have the same sort of justification for believing facts about a past older than five minutes, then since my doppelganger’s justification doesn’t consist in direct acquaintance with the relevant past facts (they don’t exist), then neither does my justification. I don’t want to overstate the epistemic problem (it is bad enough). Poston worries that I won’t even be able to classify my current experience—he suggests that judgment involves implicit comparison with what is present to prior experiences. As I indicated earlier, I just don’t think that is true, but to argue for that conclusion involves presenting and defending a theory of representation and how it corresponds (or fails to correspond) to the world (see Fumerton 2002). In any event, on my view, I can still always enjoy justification for beliefs about the current contents of my mind and, probably, for relatively simple necessary truths (assuming that I can be presently acquainted with their truth makers). But that’s small comfort for the philosopher

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi



RICHARD FUMERTON

interested in seeking justification for most of what falls under commonsense. The past holds the key to justification for most beliefs about the unobserved and indirectly observed present and the future. Enumerative inductive reasoning requires that we have justified belief in premises describing past correlations. On almost any plausible view of causal connection, causal reasoning also involves, critically, being able to correlate data from the past. Hume (1888, Part IV: 212)17 thought that all inferential reasoning involved causal reasoning (either reasoning from cause to effect, or effect to cause). Even complicated logical or mathematical reasoning seems to involve relying on memory to ‘hold on’ to premises through which we reasoned. If the reasoning takes any time at all, and knowledge of the past is inferential, then we lose our noninferential justification for the conclusion of mathematical and logical reasoning in the passing of the specious present. So it is an understatement to suggest that it would be nice to figure out how we can get intellectually satisfying justification for beliefs about a past that extends beyond the specious present. There really aren’t many options for a classical foundationalist. Of course, if there were a Keynesian relation of making probable holding between propositions, one might convince oneself that one’s seeming to remember that P makes it likely that P. To be sure, this sounds very much like phenomenal conservatism, the view that prima justification is bestowed on a belief by the fact that what is believed seems to be true. But one could borrow a page from Brogaard (Chapter 4) and argue that it is only certain seemings that create for one prima facie noninferential justification. Furthermore, one might argue, apparent memory is one of them. The argument seems painfully ad hoc, however. A moment can contain a flood of memories and one might hope to employ Bayesian reasoning or reasoning to the best explanation for the reliability of apparent memory by relying on the premise that the remarkable coherence among one’s apparent memories would be extremely unlikely were it not the case that apparent memory is at least usually veridical. But one better get this reasoning done quickly lest one lose the critical premises to the past. Earlier, I suggested that one should probably allow for derivative (degenerate) concepts of inferential justification. The ideal of inferential justification constituted by a fully conscious inference from noninferentially justified premises, where one is directly aware of the evidential connection between all of one’s premises and one’s conclusion, is just that, an ideal—an ideal seldom realized (see again, Huemer (Chapter 7) and Dougherty (Chapter 6)). One might allow that if one’s belief that P is caused by some fact E1 that is a truth maker for a proposition E1, where E1 is a critical element in a set of true propositions that makes probable P, one has some sort of justification for believing P. One might add to that the suggestion that the justification will be even stronger if the other propositions from which the References to David Hume’s A Treatise of Human Nature are to page number of the “Oxford Philosophical Texts,” edited by L. A. Selby-Bigge. 17

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

THE PROSPECTS FOR TRADITIONAL INTERNALISM



justification derives are, at least in principle, accessible (even if only a few at a time). But even if one allows for the possibility of such degenerate justification, one would need to be suitably cautious in one’s claims about whether or not one ever has such justification. Arguably, to have ideal justification for believing that one has degenerate justification one would need access to the ideal justification of which it falls short. But it wouldn’t surprise me if it turns out that our epistemic predicament is precisely that. We are constantly in search of ideal inferential justification of a sort that would vindicate our (often unsupported) conviction that we are epistemically on the right track. It is the plight of the epistemologist that there is no guarantee that we won’t be frustrated in achieving that goal.

References Beebee, Helen and Julian Dodd (eds). (2005). Truthmakers: The Contemporary Debate. New York: Oxford University Press. Bergmann, Michael. (2006). Justification without Awareness. Oxford: Oxford University Press. Coppenger, Brett. (2014). “Assurance Internalism,” Orange Beach Epistemology Workshop. Fumerton, Richard. (1995). Metaepistemology and Skepticism. Lanham: Rowman and Littlefield Press. Fumerton, Richard. (2002). Realism and the Correspondence Theory of Truth. Lanham, MD: Rowman and Littlefield. Fumerton, Richard. (2004). “Achieving Epistemic Ascent,” in John Greco (ed.), Ernest Sosa and His Critics. Malden, MA: Blackwell Publishing, 72–85. Fumerton, Richard. (2006). “Epistemic Internalism, Philosophical Assurance, and the Skeptical Predicament,” in Thomas M. Crisp, Matthew Davidson, and David Vander Laan (eds), Knowledge and Reality: Essays in Honor of Alvin Plantinga. Dordrecht: Springer Press, 179–91. Fumerton, Richard. (2007). “Epistemic Conservatism: Theft or Honest Toil?,” in Tamar Szabo Gendler and John Hawthorne (eds), Oxford Studies in Epistemology (Vol. 2). Oxford: Oxford University Press, 64–87. Fumerton, Richard. (2009). “Markie, Speckles, and Classical Foundationalism,” Philosophy and Phenomenological Research, 79: 207–11. Fumerton, Richard. (2011). “Reflective Knowledge and Intellectual Assurance,” International Journal for the Study of Skepticism, 1: 113–23. Fumerton, Richard. (2014). “Infinitism,” in John Turri and Peter Klein (eds), Ad Infinitum: New Essays on Epistemological Infinitism. Oxford: Oxford University Press, 75–86. Fumerton, Richard. (Forthcoming). “What the Internalist Should Say to the Tortoise,” Synthese, Special 10th Anniversary Conference Edition. Huemer, Mike. (2002). “Fumerton’s Principle of Inferential Justification,” Journal of Philosophical Research, 27: 329–40. Hume, David. (1888). A Treatise of Human Nature, in L. A. Selby-Bigge (ed.), A Treatise of Human Nature. Oxford: Oxford University Press. Russell, Bertrand. (1921). Analysis of Mind. London: Allen and Unwin. Taylor, Samuel. (2013). “The Problem of Easy Justification: An Investigation of Evidence, Justification, and Reliability,” Ph.D. thesis, Iowa City: University of Iowa.

OUP CORRECTED PROOF – FINAL, 9/1/2016, SPi

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

Index acquaintance 5–6, 8–12, 26–41, 43–59, 62–3, 72–5, 78–80, 89, 155–6, 159, 183, 186–8, 190–7, 199–200, 205–6, 208–10, 213–14, 221, 239–41, 243–6, 248–50, 253–5 Alston, William 85 appearances 86, 89, 96, 102, 149, 151–9, 172, 184, 196, 231, 245, 251 n. 16 Armstrong, David 153 n. 8 assurance 1, 4–5, 7–8, 27, 30–1, 36, 39–41, 44 n. 2, 74, 165, 187, 205–10, 213–14, 217, 220–1, 231–2, 234, 246–8 awareness 6, 25–31, 33, 38, 40–1, 43, 45, 52–3, 57, 72, 114, 142, 148, 154, 157, 184, 197, 208, 226, 239, 242, 246, 250, 254–5 Ayer, A. J. 39 n. 15, 120 n. 20 basicality 65, 69, 70–1, 78 doxastic 69, 71 evidential 69, 70–1, 78 Bergmann, Michael 8, 28 n. 7, 29 n. 11, 33 n. 12, 41, 63 n. 2, 67 n. 11, 88 n. 1, 134, 136, 140 n. 10, 205, 217 n. 7, 221, 251–2 Block, Ned 114 n. 16 BonJour, Laurence 5 n. 4, 6, 25, 29, 33, 36–8, 41, 62–8, 193, 198 n. 11, 208 n. 5 bootstrapping 99–102, 230–1, 234 Brogaard, Berit 9, 77 n. 25, 96, 101, 244, 256 Brown, Jessica 80 n. 28, 165 n. 2, 173 n. 13 Brueckner, Anthony 165 n. 2 Burge, Tyler 112 n. 14 Byrne, Alex 90–1, 93, 114 n. 16 Carroll, Lewis 9–10, 133 n. 3, 142, 146–7, 150, 152–3, 159 Cartesian Circle 235 Chalmers, David 114 n. 16, 194–6 Chisholm, Roderick 8, 106 n. 2, 108 n. 7, 140–1 circularity 11–12, 206, 211, 227, 232–4 epistemic 11–12 rule 12, 213, 221–2, 232–3 Cohen, Stuart 11 n. 8, 67 n. 11, 68, 71 n. 18, 77 n. 25, 140 Comesaña, Juan 77 n. 25 Conee, Earl 3, 25 n. 2, 67, 108 n. 8, 109 conservatism 2–4, 49–50, 56–8, 65 n. 7, 66–7, 75–7, 101–2, 131–2, 159, 184, 197–200, 244–5, 251, 256 epistemic 184, 197–200

phenomenal 3–4, 49–50, 56–8, 66–7, 75–7, 101–2, 131–2, 159, 244–5, 251, 256 Coppenger, Brett 247 n. 12 Davidson, Donald 118 defeater 3, 48, 67–8, 69 n. 15, 75–8, 96, 99–102, 109, 120, 130, 134, 142, 149–53, 155, 184–5, 210 rebutting 96, 99, 151 undercutting 96, 99, 151 DeRose, Keith 108 n. 8 Descartes, Rene 2, 4–5, 7, 10, 109, 132, 135 n. 6, 226–32, 234–5, 244, 248, 252–3 direct acquaintance, see acquaintance direct awareness, see awareness disjunctivism 85, 89–94, 219 n. 8, 244–5 dogmatism 9, 65 n. 7, 67, 77 n. 25, 85, 88, 96–103 doppelganger 165–78, 253, 255 Dougherty, Trent 3 n. 3, 10, 136, 249, 256 Dretske, Fred 107 n. 5, 114 n. 16 epistemic circularity, see circularity, epistemic epistemic externalism, see externalism epistemic internalism, see internalism evidence 1, 3–4, 7, 9–11, 49, 65–80, 86, 95–7, 105–12, 118–21, 128–9, 132–42, 147–8, 157, 167–70, 177–9, 184, 191, 193, 196, 198–9, 226, 243–4, 249–50, 252–3 background 193 factive 105 introspective 105–6, 121 perceptual 86, 105–10 phenomenal 105–12, 118–21 evidentialism 2–4, 67 Evil Demon Problem 11, 38, 79–80, 87, 97, 132, 137, 165–82, 226, 244, 249, 252–5 experience, see perceptual experience externalism 1–3, 29, 40–1, 62–3, 71 n. 18, 87–90, 94, 129, 132, 134, 184–5, 205–7, 210–11, 213–17, 227–8, 239, 247, see also reliabilism Fales, Evan 5 n. 4, 43, 44 n. 1, 58 n. 15 Feldman, Richard 3, 61, 64 n. 3, 67, 69 n. 14, 108 n. 8, 109, 135 n. 8, 141 n. 14, 169 n. 8 fleeting present 183–200 Fodor, Jerry 153 n. 8

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi



INDEX

foundationalism 9, 25–31, 34–41, 43, 44 n. 1, 54–9, 61–80, 96, 239–40, 249 Cartesian 61 classical 25 n. 1, 27–8, 43, 44 n. 1, 54–9, 62, 72–5 confrontational 25–31, 34–41 Fumerton, Richard 2 n. 2, 4–12, 25–32, 34, 35 n. 14, 36–8, 39 n. 15, 40–1, 43–54, 56–7, 59, 62–3, 71–5, 78 n. 26, 79–80, 85–6, 88–9, 95, 97, 127–8, 130–1, 138, 140, 141 n. 14, 145 n. 3, 146, 148, 152–6, 166, 176 n. 14, 183–93, 197–9, 205–11, 213–14, 217, 218 n. 8, 220–2, 227–8, 231–2, 241, 249–50, 255 Goldberg, Sanford 11, 166, 252–4 Goldman, Alvin 87, 97, 109, 252 Goodman, Nelson 118 n. 18 Greco, John 112 n. 14, 121 n. 24, 134 hallucination 10, 73, 79, 87, 89–94, 105, 107, 109–10, 112–19, 121, 171, 219 n. 8, 255 Harman, Gilbert 88, 114, 121 n. 22 Hasan, Ali 43–4, 49 n. 6, 54–9 Huemer, Michael 3, 10, 56 n. 13, 58 n. 15, 65 n. 7, 66–7, 70 n. 16, 77 n. 25, 100, 127, 130–5, 137–8, 140, 147, 149, 151 n. 5, 153 n. 8, 244, 249–51, 256 Hume, David 10–11, 32–3, 39 n. 15, 188, 256 illusion 91–4, 105, 108, 109 n. 10, 114, 168, 183, 192, 231, 252 infallibility 62, 73, 78 n. 27, 107 n. 4 inference 35, 55 n. 12, 98, 113 n. 15, 127, 133, 136, 138–42, 144–59, 171–3, 180–1, 190, 193, 213, 229–30, 231 n. 8, 243, 249, 253, 255–6 deductive 62, 133, 149, 170 n. 9, 171–3, 180, 224, 253 non-deductive (inductive) 77 n. 24, 87, 133, 141–2, 149, 171–2, 180, 185–6, 193, 224, 253, 256 internalism 1–12, 41 n. 16, 62, 67 n. 11, 70–4, 121 n. 23, 129–30, 137–8, 141 n. 13, 166–82, 184, 190–1, 199–200, 206, 211, 239, 244, 251–5 Cartesian 1, 230 epistemic 1, 3 n. 3, 121 n. 23, 175, 184, 199 holistic 71, 75 inferential 71, 75, 141 n. 13, 190 monistic 70–2 traditional 1–12, 239, 251, 254 internalist reliabilism, see reliabilism, internalist intuition 2, 3, 11, 35–6, 58 n. 15, 78 n. 26, 89–90, 97, 127, 129, 132, 134, 135 n. 6, 147–9, 153–5, 177, 182, 217, 229–31, 234, 244, 251, 253

Jackson, Frank 106 n. 3, 108 n. 7, 194 justification 1, 3–12, 25–41, 43–59, 61–80, 86–90, 93–103, 109 n. 11, 121 n. 22, 127–42, 144–59, 165–82, 184–200, 205–6, 208–11, 221–2, 231–3, 239–57 demon-proof 11, 165–82, 249, 252 fallible 43–59, 181 infallible 43–8, 53–4, 61–2, 73–5, 78, 230 n. 7 inferential 6–7, 9–12, 25, 27 n. 6, 35, 53, 55, 71–2, 75, 79, 86–7, 95, 97, 103, 127–34, 138, 140–2, 145–59, 172, 187, 191–3, 247–53, 255–7 introspective 43, 54, 57 meta- 9, 63–75, 78 n. 26 non-inferential 5–9, 12, 25, 27–41, 43–8, 50–6, 58–9, 65, 71–3, 75, 86–7, 89–90, 97, 103, 142, 151, 185, 187–90, 192–4, 196, 205–6, 208–10, 221, 239–49, 252, 255–6 perspectival 134–5 knowledge 1, 4, 10, 11 n. 8, 12, 28, 34 n. 13, 50, 55, 61–2, 68 n. 13, 70, 71 n. 18, 77 n. 25, 79, 85, 88–9, 92–3, 97, 99, 127, 129–30, 134–9, 140 n. 11, 141–2, 151 n. 6, 157–8, 165, 167, 182, 184–5, 188–9, 194, 196–9, 205, 208 n. 5, 211–22, 224–34, 239, 241, 252, 256 Lewis, C. I. 185 Lewis, David 111–12, 113 n. 15 Lycan, William 56 n. 13, 114 n. 16 Markie, Peter 9, 34, 38, 41 n. 16, 56, 77 n. 25, 99–101, 243–6 McDowell, John 85, 89–91, 97, 108, 208 n. 5, 218 n. 8 McGrath, Matthew 108 n. 8, 198 n. 12 McGrew, Timothy 5 n. 4 mentalism 3 Millikan, Ruth 112 n. 14, 117 Moore, G. E. 88, 211 naive realism 85, 87, 89–91, 94–6, 99, 103 Neta, Ram 120 n. 20, 121 n. 24 non-inferential identification ability 34–41 Nozick, Robert 87 Peacocke, Cristopher 114 n. 16 perception 11–12, 85, 90–4, 96–7, 103, 105–21, 129, 138–9, 165, 206–7, 212–22, 225, 228–30, 233–4 representational view of 85, 96, 103 perceptual belief 9, 100, 166, 180, 217, 226–7 perceptual capacities 105, 111–21 perceptual experience 9, 68, 80, 105–7, 120, 132, 134, 152, 165, 207, 209–10, 216–20 phenomenal dogmatism 9, 85, 96–7, 99–103

OUP CORRECTED PROOF – FINAL, 11/1/2016, SPi

INDEX

philosophical assurance, see assurance Plantinga, Alvin 97, 106, 151 Pollock, John 62, 96, 109 Poston, Ted 8, 11, 44 n. 2–3, 186 n. 4, 187 n. 5, 193 n. 8–9, 194 n. 10, 198, 199 n. 14, 255 Principle of Inferential Justification (PIJ) 7, 10–11, 35 n. 14, 71, 86–7, 94–5, 97–8, 128, 130–4, 138, 140–1, 148, 193, 250 Pritchard, Duncan 12, 205, 218 n. 8, 251 Pryor, James 65 n. 7, 67, 69 n. 15, 70 n. 16, 76 n. 22, 77 n. 25, 79–80, 108 n. 8, 109, 120 n. 20, 121 n. 23 Pyrrhonism 228 Quine, W. V. 198 Ranalli, Chris 12, 219 n. 8, 251 reasons 96, 135, 137–40, 170–2, 224, 227, 232 regress problems 9–10, 25, 27, 63 n. 2, 67 n. 11, 73, 79, 95, 98, 148, 227, 232–3, 248–50 reliabilism 33 n. 12, 49–50, 63, 67–8, 71 n. 18, 77, 87–90, 94, 97, 210, 228–9 internalist 67–8, 77 representational view of perception, see perception, representational view of Rhoda, Alan 141 n. 13, 144 n. 1, 145 n. 3, 148 n. 4, 149 Russell, Bertrand 27–8, 47, 50 n. 7, 95–6, 109, 130, 147, 168–9, 188, 255 scepticism, see skepticism Schellenberg, Susanna 9, 105 n. 1, 106, 244



seemings 3–4, 9, 54 n. 10, 55–9, 66–7, 76, 78–9, 96, 99, 101–2, 107 n. 5, 108–9, 135 n. 8, 136, 241, 244, 256 Siegel, Susanna 56, 77 n. 25, 101–2 skepticism 1, 4, 7, 9–12, 79, 85–7, 94–9, 103, 148, 153–5, 159, 165–6, 183–200, 205–22, 224–9, 235 n. 11, 239, 250–2, 255 external world 10–12, 73, 79–80, 85–7, 95–9, 103, 175, 184, 198, 205–22, 225–9, 252 memory 11, 96–9, 183–200, 252, 255–6 metaepistemological 205–7, 211–13, 215 Sosa, Ernest 12, 26 n. 3, 32, 34–6, 38, 108 n. 7, 112 n. 14, 165, 227 n. 3, 228, 229 n. 5, 231, 232 n. 10, 245, 248 n. 13, 249, 251–2 Speckled Hen Problem 8–9, 34–40, 245–6 Steup, Matthias 9, 67 n. 11, 76, 241, 246 Stroud, Barry 12, 205–6, 211–22, 231, 232 n. 10, 251–2 traditional internalism, see internalism Tucker, Chris 3 n. 3, 9, 56 n. 13, 109, 128 n. 1, 131 n. 2, 133 n. 4, 134, 136 n. 9, 145 n. 3, 152 n. 7, 240–2 Tye, Michael 114 n. 16, 121 n. 24 Van Cleve, James 65 n. 7, 198 Vogel, Jonathan 71 n. 18 Williamson, Timothy 92–3, 106, 107 n. 4, 108, 218 Wittgenstein, Ludwig 173, 200 Wright, Crispin 108 n. 8 Zagzebski, Linda 112 n. 14