Consciousness and the Prospects of Physicalism 9780199764037, 2010015401, 0199764034

Table of contents :
Cover......Page 1
Contents......Page 8
Acknowledgments......Page 10
Introduction......Page 14
1. The Knowledge Argument and Introspective Inaccuracy......Page 20
2. Phenomenal Concepts and the Explanatory Gap......Page 40
3. Conceivability Arguments and Qualitative Inaccuracy......Page 58
4. Qualitative Inaccuracy and Recent Objections to Conceivability Arguments......Page 77
5. Russellian Monism I......Page 96
6. Russellian Monism II......Page 113
7. Robust Nonreductive Physicalism......Page 134
8. Mental Compositional Properties......Page 159
Conclusion......Page 181
Bibliography......Page 184
C......Page 198
K......Page 199
P......Page 200
S......Page 201
Z......Page 202
G......Page 204
P......Page 205
Z......Page 206

Citation preview

Consciousness and the Prospects of Physicalism

Derk Pereboom

PHILOSOPHY OF MIND SERIES Series Editor: David J. Chalmers, Australian National University SIMULATING MINDS The Philosophy, Psychology, and Neuroscience of Mindreading Alvin I. Goldman SUPERSIZING THE MIND Embodiment, Action, and Cognitive Extension Andy Clark PERCEPTION, HALLUCINATION, AND ILLUSION William Fish PHENOMENAL CONCEPTS AND PHENOMENAL KNOWLEDGE New Essays on Consciousness and Physicalism Torin Alter and Sven Walter THE CHARACTER OF CONSCIOUSNESS David J. Chalmers THE SENSES Classic and Contemporary Philosophical Perspectives Fiona Macpherson ATTENTION IS COGNITIVE UNISON An Essay in Philosophical Psychology Christopher Mole THE CONTENTS OF VISUAL EXPERIENCE Susanna Siegel CONSCIOUSNESS AND THE PROSPECTS OF PHYSICALISM Derk Pereboom

Consciousness and the Prospects of Physicalism

By Derk Pereboom

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

Copyright © 2011 by Oxford University Press, Inc. Published by Oxford University Press, Inc. 198 Madison Avenue, New York, New York 10016 www.oup.com Oxford is a registered trademark of Oxford University Press All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Library of Congress Cataloging-in-Publication Data Pereboom, Derk, 1957– Consciousness and the prospects of physicalism / by Derk Pereboom. p. cm. — (Philosophy of mind series) Includes bibliographical references (p.) and index. ISBN 978-0-19-976403-7 1. Consciousness. 2. Materialism. I. Title. B808.9.P47 2011 126—dc22 2010015401

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

For Nancy, Eleanor, and Marilyn

This page intentionally left blank

CO N T E N TS

Acknowledgments ix Introduction 3 1. The Knowledge Argument and Introspective Inaccuracy 9 2. Phenomenal Concepts and the Explanatory Gap 29 3. Conceivability Arguments and Qualitative Inaccuracy 47 4. Qualitative Inaccuracy and Recent Objections to Conceivability Arguments 66 5. Russellian Monism I 85 6. Russellian Monism II 102 7. Robust Nonreductive Physicalism 123 8. Mental Compositional Properties 148 Conclusion 170 Bibliography 173 Index of Topics 187 Index of Names 193

This page intentionally left blank

A C K N O W L E D G M E N TS

The first two chapters of this book are a revision, with substantial additions, of Derk Pereboom, “Consciousness and Introspective Inaccuracy,” in Metaphysics and the Good: Themes from the Philosophy of Robert M. Adams, ed. L. M. Jorgensen and Samuel Newlands, Oxford: Oxford University Press, 2009, pp. 156–87. The last two chapters are an extensively revised version of Derk Pereboom, “Robust Nonreductive Materialism,” Journal of Philosophy 99 (2002), pp. 499–531, and of some of the material in Derk Pereboom and Hilary Kornblith, “The Metaphysics of Irreducibility,” Philosophical Studies 63 (1991), pp. 125–45, in particular sections I, II, and V. Research on this project was facilitated by sabbatical leaves at the University of Vermont in 2001 and 2005, a generous Visiting Fellowship in the Centre for Consciousness of the Research School of Social Sciences at the Australian National University in 2005, and a study leave at Cornell University in 2009. I am grateful to the participants in two of my seminars at Cornell, in the spring and fall semesters of 2008, in which I presented the contents of the manuscript; to those in Nico Silins’s seminar in the spring semester of 2010, in which the first two chapters were read; and to the students in courses over the years on this material at the University of Vermont. Thanks in addition to audiences at Rutgers University, the University of Vermont, Yale University, the University of Auckland, the Australian National University, the University of Alabama, Syracuse University, the University of Buffalo, the State University of New York at Brockport, the University of California at San Diego, Brown University, and the meetings of the Pacific Division of the American Philosophical Association in 2009. Personal gratitude for valuable discussion is due to Zachary Abrahams, Robert Adams, Lauren Ashwell, Kati Balog, David Barnett, Lynne Baker, Ned Block, Richard Boyd, David Braddon-Mitchell, David Braun, Wylie Breckenridge, Sin yee Chan, Andrew Chignell, Philippe Chuard, Philip Clark, Jonathan Cohen, Earl Conee, Michael Della Rocca, Keith DeRose, Matti Eklund, Carl Ginet, Robert Howell, Frank Jackson, Larry Jorgensen, David Kaplan, Theodore Korzukhin, Arthur Kuflik, Sukjae Lee, Don Loeb, Fiona Macpherson, Stephen Mahaffey, William Mann, Andrew McGonigal, Colin McLear, Alyssa Ney, L. A. Paul, David Robb, Denis Robinson, Carolina Sartorio, Karin Scheiber, Ira Schnall, Tim Schroeder, Sydney Shoemaker, Subrena Smith, Brad Thompson, Andrea Viggiano, Adam Wager, Brian Weatherson, David Widerker, Jessica Wilson, and Stephen Yablo. I wish to thank especially Torin Alter, Karen Bennett, David Chalmers, David Christensen, Louis deRosset, Tyler Doggett, Janice Dowell, Hilary Kornblith, Mark Moyer, Nico Silins, and Daniel Stoljar for reading all or substantial parts of drafts of the manuscript and for excellent commentary that shaped the final version. ix

This page intentionally left blank

Consciousness and the Prospects of Physicalism

This page intentionally left blank

INTRODUCTION

Recent developments in philosophy of mind have presented the strongest challenges ever devised to physicalism as a comprehensive ontological position. The truth of physicalism would plausibly be established by a general argument to the best explanation, and the greatest obstacle to the success of such an argument derives from phenomenal consciousness. The knowledge and conceivability arguments, which articulate this impediment, have received increasingly sharper formulations, and convincing responses have proven difficult to produce. These two arguments crucially presuppose that we introspectively represent phenomenal properties as having specific qualitative natures that are distinct from any features that physical theories represent them as having. If this presupposition is endorsed, as I think it should be, the range of contending physicalist responses is restricted. Each of the two I will develop results from an application of a Kantian theme to the issues at hand. The first draws on the open possibility that our introspective representations fail to represent mental states as they are in themselves. More specifically, introspection represents phenomenal properties as having certain characteristic qualitative natures, and it may be that these properties actually lack such features. I call this open possibility the qualitative inaccuracy hypothesis. The seriousness of this open possibility is enhanced by an analogy with our perceptual representations of secondary qualities. Our vision represents colors as having certain qualitative natures, and it is an open possibility, widely regarded as actual, that colors actually lack them. A stronger thesis involves the claim that due to how we represent both secondary qualities and phenomenal properties, we take them to be primitive or have a strong tendency to do so. That is, we take these properties to be metaphysically simple—as not constituted by or analyzable into multiple properties—and, more significantly, as having qualitative natures wholly revealed in sensory or introspective representations of them. In the color case, it is often agreed that due to how we visually represent color, we take it to be primitive,

3

4 Consciousness and the Prospects of Physicalism

while it is actually not. Similarly, I suggest that many of us are apt to take phenomenal properties as primitive, but it might well be that they are really not. The open possibility of qualitative inaccuracy has implications for the knowledge and conceivability arguments against physicalism. If it might be that representing phenomenal properties introspectively attributes to them qualitative natures that they actually lack, then the force of these arguments may be blunted. This open possibility has an implication for David Chalmers’s zombie argument that has independent interest.1 This argument depends on the assumption that pure phenomenal concepts have a certain kind of structure, an assumption to which this open possibility yields a challenge. I set out the qualitative inaccuracy hypothesis and its consequences for the knowledge argument in chapter 1. Chapter 2 focuses on the implications of this hypothesis for what it is that phenomenal concepts represent, and for the claim that there is an explanatory gap between the phenomenal and the physical. Chapter 3 assesses the upshot of the qualitative inaccuracy hypothesis for conceivability arguments against physicalism. Chapter 4 contends that the serious open possibility that this hypothesis is true significantly strengthens or otherwise embellishes several recent objections to conceivability arguments. The subject of chapters 5 and 6 is the second Kantian theme, that our ignorance of things in themselves consists in part in our lack of knowledge of the fundamental intrinsic properties of things. This idea has been developed by Bertrand Russell and more recently by Chalmers into a framework for a unified account of the mental and the physical. In Chalmers’s version, the currently (but not inevitably) unknown or at least incompletely understood intrinsic properties provide the categorical bases for the known physical dispositional properties and would also yield an account of consciousness. While there are nonphysicalist versions of this view, some are amenable to physicalism. In Chalmers’s terminology, the variants that are potentially physicalismfriendly propose that the fundamentally intrinsic properties are protophenomenal, that is, properties that are not phenomenal but nonetheless explain the instantiations of phenomenal properties. The resulting type of physicalism has an advantage over the kind discussed in the first four chapters, since it can preserve the intuitive claim that phenomenal properties really possess the qualitative natures we introspectively represent them as having. The third theme of this book is antireductionism, which is also Kantian or, perhaps more accurately, neo-Kantian in its early development. It is the topic of chapters 7 and 8. Neo-Kantians argued that the human or mental sciences are autonomous in that they fail to reduce to the natural sciences. Their position is at least implicitly metaphysically irrealist, since our scientific practices are governed by principles that we impose and are in this sense a priori. On one influential version, the natural sciences are governed by a principle of causality, so that the task of those sciences is to discover the 1 David Chalmers, The Conscious Mind (New York: Oxford University Press, 1996); “Consciousness and Its Place in Nature,” in Philosophy of Mind: Classical and Contemporary Readings, ed. David Chalmers (New York: Oxford University Press, 2002), pp. 247–72.

5 Introduction

general causal laws that govern the phenomena in its purview. By contrast, the point of the human sciences is to discern individual events that involve human activity and to explain them in terms of particular reasons and motivations, which are not amenable to causal lawlike regularities. Because of such methodological differences, the human sciences are importantly distinct from the natural sciences and do not reduce to them. The logical positivists argued that this scheme is mistaken, that the right way to conceive of the human sciences is as natural sciences, and that ultimately all of the sciences reduce to physics. This is the doctrine of the unity of science. An alternative sort of nonreductivism was proposed in the 1960s and 1970s by Richard Boyd, Hilary Putnam, and Jerry Fodor.2 They agreed with the logical positivists that at least part of the point of the human sciences is to find general causal laws in their domains and that all the phenomena in these domains are wholly physical. But they denied that these concessions warranted the reduction of all sciences to physics or the reduction of the human sciences to the natural sciences. Thus the view they proposed is a nonreductive physicalism. In my preferred version of this position, the core reason for nonreductivism is not methodological or pragmatic but metaphysical. Natural kinds in psychology are not identical to natural kinds in physics because psychological causal powers are not identical to microphysical causal powers. The fact that psychological kinds are multiply realizable at the level of microphysical kinds yields the important clue as to why this is so. The version of nonreductivism I will defend departs from other nonreductivisms in that it rejects the token identity of psychological and microphysical entities of any sort—including causal powers. The deepest relation between the psychological and the microphysical is constitution, where this relation is not to be explicated by the notion of identity. At the same time, this metaphysical view has a methodological implication: psychological theories, laws, and explanations cannot be supplanted, even on the approach of ideal science, by theories, laws, and explanations in more basic sciences, and in this sense psychology is autonomous. Although Kant employed the first two ideas in the service of ontological agnosticism and not of physicalism, they nonetheless can be used to defend a physicalist position. But what is physicalism, exactly? One reasonable requirement is that all facts about the universe be necessitated by facts about a base of paradigmatically physical entities, plausibly the microphysical entities. As we shall see in chapter 7, the relation must be tighter than just this, credibly spelled out in terms of realization or constitution. But are the paradigmatically physical entities in the base those that feature in our current physics? This seems inadequate, since it is highly probable that currently unconceived entities will be discovered by future physics and that they will be counted 2 Hilary Putnam, “The Nature of Mental States,” in his Philosophical Papers, vol. 2 (Cambridge, MA: Harvard University Press, 1975), pp. 429–40, first published as “Psychological Predicates,” in Art, Mind, and Religion, ed. W. H. Capitan and D. D. Merill (Pittsburgh, PA: Pittsburgh University Press, 1967), pp. 37–48; Jerry Fodor, “Special Sciences,” Synthèse 28 (1974), pp. 97–115; Richard Boyd, “Materialism without Reductionism: What Physicalism Does Not Entail,” in Readings in the Philosophy of Psychology, ed. Ned Block (Cambridge, MA: Harvard University Press, 1980), pp. 67–106.

6 Consciousness and the Prospects of Physicalism

as physical.3 One attractive alternative is that ‘physical’ is a prototype-similarity concept, where the prototypes are the sorts of entities we find in current physics (or are wholly constituted or realized by them). Whether the entities of future physics will count as physical depends on their similarity to these prototypes.4 I’m at least somewhat partial to this prototype-similarity view. But I agree with Jessica Wilson that a key issue in this debate is whether among the fundamental entities of the universe some are mental and that physicalism in the main sense at issue in philosophy is incompatible with the existence of fundamentally mental entities, even if they come to be accepted by future physics. As she argues, “Physicalists have not handed over all authority to physics to determine, a posteriori, what is physical . . . a physics-based account of the physical should not be understood as the view that any and all entities treated by physics—current, future, or ideal—are physical.” Rather, “an entity is physical just in case it is (approximately accurately) treated by current or future (at the end of inquiry, ideal) physics, and is not fundamentally mental.”5 Wilson’s type of approach is fairly liberal. Physicalism can, for instance, allow positing consciousness to explain the collapse of the wave function if consciousness itself has a physical account or, supposing physicalism to be making a methodological recommendation, if consciousness is reasonably presumed to have a physical account. Also, physicalism can posit new entities just to account for consciousness, so long as they have a complete physical account or are reasonably presumed to have one. Wilson’s approach does have the consequence that it’s a priori that there are no fundamental physical entities that are mental, and Janice Dowell, for example, finds this counterintuitive. But it’s a result I accept; if a theory in future physics were to posit fundamental entities that are mental, then physicalism would be false on that theory.6

3 Tim Crane and Hugh Mellor, “There Is No Question of Physicalism,” Mind 99 (1990), pp. 185–206. 4 David Chalmers, “Consciousness and Its Place in Nature,” pp. 261–63; Robin Brown and James Ladyman, “Physicalism, Supervenience, and the Fundamental Level,” Philosophical Quarterly 59 (2009), pp. 20–38, at p. 32. 5 Jessica Wilson, “Supervenience-Based Formulations of Physicalism,” Noûs 39 (2005), pp. 426–59, at p. 428; for a general overview of these issues, see Daniel Stoljar, “Physicalism,” Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, at http://plato.stanford.edu/entries/physicalism/. Robin Brown and James Ladyman, in “Physicalism, Supervenience, and the Fundamental Level,” agree with Wilson’s general line, but on their formulation, physicalism claims that physics will not posit new entities solely for the purpose of accounting for mental phenomena, and it will not posit entities with essentially mental characteristics. I dissent, for the reason that such new entities might be paradigmatically physical, and entities with essentially mental characteristics might in turn have a straightforwardly physical account. I agree with Brown and Ladyman that positing a fundamental physical level should not be regarded as a requirement of physicalism. 6 Janice Dowell, in “The Physical: Empirical Not Metaphysical,” Philosophical Studies 131 (2006), pp. 25–60, provides an acute defense of the opposing view. One might allow that ‘physicalism’ has a sense that accords with Dowell’s view, but a pressing and divisive issue in philosophy of mind is whether there are fundamental mental entities, and an important philosophical sense tracks this divide.

7 Introduction

Given that physicalism does not allow fundamentally mental entities, its opponents will include idealism, traditional substance and property dualism, and their panpsychist variants. There is a gray area: for example, if the necessitation of the mental by the microphysical requires emergent laws, is physicalism true? It is common to answer in the negative, but what if mental entities are still wholly constituted of microphysical entities? Perhaps it isn’t clear whether our concept ‘physical’ would apply to the mental in this case, but this is a theory that is at least in a important respect opposed to the spirit of physicalism. This issue is raised in chapter 7. My intention isn’t to work out the details of a positive argument for physicalism but rather to assess the prospects for physicalism in the face of the strongest challenges to it and to set out the versions I think are most likely to be true. As I will make clear especially in chapters 4, 5, and 6, whether these challenges can be met depends on issues that remain undecided—for instance, whether the qualitative inaccuracy hypothesis is true and what the currently unknown fundamentally intrinsic properties turn out to be. One ultimate philosophical issue at stake is whether we human beings are capable of knowing what the fundamental properties of reality are. Descartes contended that our rational capacities allow us to have such knowledge of thinking as the essence of mind and extension as the essence of matter.7 Leibniz and Berkeley were equally optimistic but held that the fundamental properties of all things are mental. Kant was a skeptic on this issue, maintaining that our cognitive faculties are insufficient for us to know which fundamental properties are instantiated.8 Among contemporary philosophers, few are as optimistic as Descartes. Colin McGinn and David Lewis, while partial to physicalism, nevertheless side with the skepticism of Kant but for different reasons, as we shall see.9 A significantly more moderate skeptical position is endorsed by Thomas Nagel and Chalmers.10 While in our current situation we do not grasp what the fundamental properties are, the limitations of our faculties may not permanently bar us from such knowledge. Nagel speculates that a fundamental objective conception that unites the phenomenal and the physical, although currently unrealized, may nonetheless be attainable; Chalmers suggests that we might eventually conceptualize and confirm protophenomenal essences of both the phenomenal and the physical. In the last analysis, McGinn may be right to claim that our ignorance of fundamental properties

7 René Descartes, Principles of Philosophy, Part 1, 54, in The Philosophical Writings of Descartes, vol. 1, tr. John Cottingham, Robert Stoothoff, and Dugald Murdoch (Cambridge: Cambridge University Press, 1984) vol. 1, p. 211 (AT VIIIA 25–26). 8 Immanuel Kant, Critique of Pure Reason, tr. Paul Guyer and Allen Wood (Cambridge: Cambridge University Press, 1987), A277-78/B333-34. 9 Colin McGinn, “What Constitutes the Mind-Body Problem,” in his Consciousness and Its Objects (Oxford: Oxford University Press, 2004); David Lewis, “Ramseyan Humility,” in Conceptual Analysis and Philosophical Naturalism, ed. David Braddon-Mitchell and Robert Nola (Cambridge, MA: MIT Press, 2009), pp. 203-22. 10 Thomas Nagel, The View from Nowhere (New York: Oxford University Press, 1986), pp. 51–53; David Chalmers, “Consciousness and Its Place in Nature.”

8 Consciousness and the Prospects of Physicalism

is permanent, but my sense on balance is to side with the view of Nagel and Chalmers that the arguments for irrevocable ignorance can reasonably be resisted. In the coming chapters I will explore the proposal that all of the fundamental properties of the universe—the totality of the actual concrete contingent entities—are physical. My contention is that there are several physicalist options that are serious possibilities, and although we do not now know that any one of them is true, knowledge of this general sort is not ruled out.

1 THE KNOWLEDGE ARGUMENT AND INTROSPECTIVE INACCURACY

Kant maintained that introspective representations—those of inner sense—are caused by the mental states they represent and are wholly distinct from them. Introspective representations thus mediate the subject’s awareness of those mental states, rendering this awareness in a sense indirect. As a consequence, the subject may represent a mental state as being a certain way, even though it is not really that way, or at least not as it is in itself.1 In this chapter, I propose that the possibility of this sort of 1 Immanuel Kant, Critique of Pure Reason, tr. Paul Guyer and Allen Wood (Cambridge: Cambridge University Press, 1987), B152–54. Leibniz’s views on perception provide another model for the idea that our introspective representations of our phenomenal states are inaccurate, as Robert Adams suggested to me. For instance, Leibniz claims: It does not cease to be true that at bottom confused thoughts are nothing other than a multitude of thoughts which are in themselves like the distinct, but which are so small that each separately does not excite our attention and cause itself to be distinguished. We can even say that there is at once a virtually infinite number of them contained in our sensations. (G. W. Leibniz, Die philosophischen Schriften, ed. C. I. Gerhard, 7 vols. [Hildesheim, Germany: Olms, 1965], vol. 4, pp. 574–75; cf. Discourse on Metaphysics 33, G 4, 458–59) Contained in our sensations are a virtually infinite number of thoughts, so “small” that they are not consciously distinguished. G. H. R. Parkinson points out that in the late 1670s and beyond, Leibniz held that it is impossible for us to reach genuinely primitive concepts in our analysis of sensations; G. H. R. Parkinson, Leibniz, Logical Papers (Oxford: Clarendon, 1966), pp. xxvii–xxviii, 51–52; for a discussion of this point, see my “Kant’s Amphiboly,” Archiv für Geschichte der Philosophie 73 (1991), pp. 50–70. But as Nico Silins has remarked, then our sensations would be a certain way even though they are not introspectively represented by us as being that way, while the stronger, distinctively Kantian claim I am singling out is that we introspectively represent sensations (for example) to be a certain way, even though they are not that way, at least as they are in themselves. In Silins’s helpful terminology, Leibniz’s claim is that introspective representation is merely silent about certain features of sensations, while Kant’s idea is that it is in one respect mistaken about them. We will revisit the Leibnizian thesis in chapter 6. 9

10 Consciousness and the Prospects of Physicalism

inaccuracy generates a significant challenge to Frank Jackson’s knowledge argument against physicalism.

THE QUALITATIVE ACCURACY INTUITION In Jackson’s story, Mary has lived her entire life in a room that displays only various shades of black, white, and gray.2 She acquires information about the physical nature of the human being, and the world outside, by means of a black-and-white television monitor. By watching television programs, Mary eventually comes to have knowledge of all of the physical information there is about the nature of the human being and, we might imagine, all of the physical information there is about the actual world. This physical knowledge might be conceived as microphysical-level knowledge, or knowledge of every entity that is either microphysical or else wholly microphysically constituted, or perhaps as knowledge of every entity that is uncontroversially physical. Following Jackson and Chalmers, I opt for the first, microphysical-level, alternative, and I will assume it in this discussion.3 But even if she has all of this knowledge, the argument continues, there is much she does not thereby know and cannot thereby come to know about human experience. She does not know and cannot come to know, for example, what it is like visually to experience a ripe red tomato, and in particular, she lacks knowledge of what it is like to see red. When Mary leaves the room and sees a red tomato, she will come to know for the first time—she will learn—what it is like to see red. She will come to have knowledge for the first time of a particular phenomenal property or of a mental state that has this property—a phenomenal state.4 The conclusion is that there are facts about phenomenal states that are not physical facts, and thus phenomenal states are not completely physical. The intuition underlying the knowledge argument is that if someone who has complete microphysical knowledge of the actual world cannot thereby come to know some fact about a phenomenal state, then the fact cannot be physical, and the phenomenal state cannot be entirely physical. The more general views of Jackson and 2 Frank Jackson, “Epiphenomenal Qualia,” Philosophical Quarterly 32 (1980), pp. 127–36, and “What Mary Didn’t Know,” Journal of Philosophy 83 (1986), pp. 291–95; cf. Thomas Nagel, “What Is It Like to Be a Bat?” Philosophical Review 83 (1974), pp. 435–50; Martine Nida-Rümelin, “Qualia: The Knowledge Argument,” in The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, available at http://plato.stanford.edu/entries/qualia-knowledge. 3 David Chalmers and Frank Jackson, “Conceptual Analysis and Reductive Explanation,” Philosophical Review 110 (2001), pp. 315–61, and, for example, David Chalmers, “The Two-Dimensional Argument against Materialism,” in The Character of Consciousness (New York: Oxford University Press, forthcoming). 4 David Chalmers characterizes phenomenal properties as those that “type mental states by what it is like to have them”; see “The Content and Epistemology of Phenomenal Belief,” in Consciousness: New Philosophical Perspectives, ed. Q. Smith and A. Jokic (Oxford: Oxford University Press, 2003). The “what it is like to have them” locution should perhaps be taken as a means of signaling to an audience what to look for as instances of phenomenal properties, which can then serve as paradigms, and not so much as a thorough descriptive characterization of this type of property.

11 The Knowledge Argument and Introspective Inaccuracy

Chalmers on these issues recommend a more precise specification: if someone who has complete microphysical knowledge of the actual world and flawless powers of reasoning cannot derive some fact about a phenomenal state from what she knows, and she has the minimum information required to ensure adequate possession of the phenomenal concept involved in knowledge of that fact, then the fact cannot be physical, and the phenomenal state cannot be completely physical.5 I agree with Jackson and Chalmers that the sense of derivability at issue here is best construed as a priori.6 Exactly why a perfect reasoner’s inability to derive a phenomenal truth a priori from her physical knowledge would secure the falsity of physicalism is not an entirely straightforward matter, and we will return to it when we discuss the conceivability argument in chapter 3. Now consider the “old fact/new guise” response to the knowledge argument— which I do not endorse, although the reply I will develop can be understood as a successor to it.7 According to this kind of response, Mary, when she is still in the room,

5 This more precise version of the intuition derives from David Chalmers and Frank Jackson, “Conceptual Analysis and Reductive Explanation,” and David Chalmers, “The Two-Dimensional Argument against Materialism.” Chalmers and Jackson argue, convincingly to my mind, that the higher-level conceptual knowledge required for these sorts of a priori derivations need not amount to a conceptual analysis. More minimal conceptual knowledge is typically sufficient. One reply to the knowledge argument is that the reason pre-emergence Mary lacks knowledge of phenomenal states is just that she is missing the phenomenal concepts and the associated knowledge. In response, Daniel Stoljar strengthens the argument by specifying that pre-emergence Mary possesses all the phenomenal concepts, while she nevertheless lacks knowledge of how correct applications of phenomenal concepts are correlated with physical states. In “Physicalism and Phenomenal Concepts,” Mind and Language 20 (2005), pp. 469–94, Stoljar tells a plausible story as to how Mary might come to fit this description: after acquiring the phenomenal concepts, Mary suffers selective amnesia. David Chalmers makes a similar point in “The Two-Dimensional Argument against Materialism.” The resulting argument has a somewhat different focus. What I say in chapter 2 in reply to Robert Adams’s antimaterialist argument is also a response to this version of the knowledge argument; see in particular note 32 of that chapter. 6 David Chalmers and Frank Jackson, “Conceptual Analysis and Reductive Explanation.” As they explain in this article, the claim that the nonfundamental truths are derivable a priori, and not merely a posteriori, from a base of fundamental truths can be defended by arguing that all of the fundamental information relevant to deriving the nonfundamental truths can be included in the base, whereupon no other information, let alone empirical information, will be required to execute the derivation itself. Consequently, the derivation itself will be a priori. The a priori derivability issue will be revisited in chapter 3, especially note 4. 7 Proponents of the old fact–new guise response include Terence Horgan, “Jackson on Physical Information and Qualia,” Philosophical Quarterly 32 (1984), pp. 147–52; Paul M. Churchland, “Reduction, Qualia and the Direct Introspection of Brain States,” Journal of Philosophy 82 (1985), pp. 8–28; Robert Van Gulick, “Physicalism and the Subjectivity of the Mental,” Philosophical Topics 13 (1985), pp. 51–70; Michael Tye, “The Subjective Qualities of Experience,” Mind 95 (1986), pp. 1–17; Brian Loar, “Phenomenal States,” in Philosophical Perspectives 4: Action Theory and Philosophy of Mind, ed. James Tomberlin (Atascadero, CA: Ridgeview, 1990), pp. 81–108; William G. Lycan, “What Is the ‘Subjectivity’ of the Mental?” Philosophical Perspectives 4 (1990), pp. 109–30.

12 Consciousness and the Prospects of Physicalism

can indeed come to know every fact about phenomenal states, while what she is missing are only ways of introspectively representing those states or, as I will put it, introspective modes of presentation of those states.8 Suppose that while she is in the room, Mary has not only exhaustive microphysical physical knowledge but also knowledge of whatever can be derived from that physical base. Then, when she leaves the room and sees the red tomato, she comes to represent a phenomenal state, about which she already knew everything, by an introspective mode of presentation, with which she had never represented that phenomenal state while she was in the room. In this way, the appearance of Mary’s coming to know a new fact can be explained without granting that she actually acquires new knowledge. This sort of reply might be illustrated by various analogies. According to William Lycan, the difference between the introspective and the physical representations is akin to the difference between my use of ‘I’ and your use of ‘you’ to represent me in the representation of some fact about me.9 For example, consider: (1) ‘I weigh 195 pounds’ (asserted by me) (2) ‘You weigh 195 pounds’ (asserted by you). You cannot represent the fact that I weigh 195 pounds by ‘I weigh 195 pounds,’ whereas I can represent this fact by means of that sentence. But suppose that you have knowledge of this fact and represent it by ‘You weigh 195 pounds.’ Then there is no fact of which I have knowledge but you don’t; the only fact to be known here is that DP weighs 195 pounds, and we both know it.10 Although some find analogies of this sort sufficient to dislodge the knowledge argument, its proponents remain unconvinced. To advance the debate, we need to explore why the argument has this residual force. Are there features of Mary’s epistemic situation disanalogous with Lycan’s example that might explain this force? Phenomenal states have characteristic phenomenal properties, and it is intuitive for some of us that:

8 I use the Fregean term ‘mode of presentation’ as a convenient nominalization, without intending the full Fregean theory. The claims made in this chapter can generally be made in more neutral terms or in terms of other theories of cognition and language. For discussions of these issues, see David Chalmers, “Perception and the Fall from Eden,” in Perceptual Experience, ed. Tamar Szabó Gendler and John Hawthorne (Oxford: Oxford University Press, 2006), pp. 49–125; and Brad Thompson, “Senses for Senses,” Australasian Journal of Philosophy 87 (2009), pp. 99–117. 9 William G. Lycan, “What Is the ‘Subjectivity’ of the Mental?” 10 A thoroughly developed reply along these lines is provided by John Perry, Knowledge, Possibility, and Consciousness (Cambridge: MIT Press, 2001); see also John Hawthorne, “Advice for Physicalists,” Philosophical Studies 109 (2002), pp. 53–74. For someone who is not convinced by these sorts of accounts, see David Chalmers, “Imagination, Indexicality and Intensions,” Philosophy and Phenomenological Research 68 (2004), pp. 182–90. Robert Stalnaker’s Our Knowledge of the External World (Oxford: Oxford University Press, 2008) features an in-depth but inconclusive discussion of this type of position.

13 The Knowledge Argument and Introspective Inaccuracy

(i) Both the physical and introspective modes of presentation represent a phenomenal property as having a specific qualitative nature, and the qualitative nature that the introspective mode of presentation represents the phenomenal property as having is not included in the qualitative nature the physical mode of presentation represents it as having.11 It is also intuitive—again, for some of us—that: (ii) The introspective mode of presentation accurately represents the qualitative nature of the phenomenal property. That is, the introspective mode of presentation represents the phenomenal property as having a specific qualitative nature, and the attribution of this nature to the phenomenal property is correct. There is no uncontroversial way to characterize qualitative natures that introspective modes of presentation represent phenomenal properties as having. One option, inspired by John Locke, is to characterize such a nature by way of resemblance to modes of presentation. Thus, in our example, we might say that Mary’s introspective representation of her new color sensation presents that sensation in a what-it-is-liketo-sense-red way, and it is intuitive that a qualitative nature that resembles this what-itis-like mode of presentation is correctly attributed to the sensation’s phenomenal property.12 Or in deference to concerns about the cogency of such resemblance characterizations, one might say simply that the qualitative nature of the phenomenal property is as the introspective mode of presentation represents it to be. Given these claims about what is intuitive, an advocate of the knowledge argument can account for its residual force in the following way: when Mary leaves the room and sees the tomato, she comes to have the belief: (A) Seeing red has R, where the concept ‘R’ in this belief is the phenomenal concept of ours that directly refers to phenomenal redness, phenomenal property R. The qualitative nature of 11 Joseph Levine, in Purple Haze (New York: Oxford University Press, 2001), accounts for the existence of the explanatory gap partly by the fact that “modes of presentation whereby we make cognitive contact with qualia are substantive and determinate” (p. 8) and that “there is real content to our idea of a quale” (p. 84). What I say here aims to explicate these kinds of intuitions; cf. Alex Byrne, “Review of Purple Haze,” Philosophical Review 111 (2002), pp. 594–97. 12 John Locke, An Essay Concerning Human Understanding (Oxford: Oxford University Press, 1975), II, viii. Accepting a resemblance claim of this sort does not amount to endorsing a discredited resemblance theory of representation, as is sometimes suggested. In accepting that an introspective mode of presentation resembles a phenomenal property, one is not also endorsing a resemblance account of how it is that the mode of presentation represents the phenomenal property. By analogy, one does not need to endorse a resemblance account of photographic representation to accept the claim that photographs can resemble what they represent.

14 Consciousness and the Prospects of Physicalism

phenomenal redness is accurately represented introspectively by way of the what-it-islike-to-sense-red introspective mode of presentation. But on the physicalist hypothesis, every truth about the qualitative nature that an introspective mode of presentation accurately represents a phenomenal property as having would need to be derivable from a proposition detailing only features that physical modes of presentation represent the world as having. However, (A) is not derivable from such a proposition. Thus not every truth about the qualitative nature that an introspective mode of presentation accurately represents a phenomenal property as having is so derivable. So the physicalist hypothesis is false.13 One might challenge this version of the knowledge argument at various points. In particular, one might take issue with one or both of the claims about what is intuitive just listed. The one I will dispute is (ii), the claim about the accuracy of introspective representation. I will leave (i) as common ground and assume that (i) is in fact true. On (ii), in my view introspective modes of presentation represent phenomenal properties as having certain qualitative natures, and it is an epistemic possibility of a certain sort that these properties do not in fact have these qualitative natures, and that introspective representation is in this sense inaccurate. This is not to say, let me note, that our phenomenal concepts misrepresent phenomenal properties, even if they have their source in introspection that in a certain respect misrepresents them (more on this later; I will advocate a dual-content view, according to which phenomenal concepts both misrepresent and correctly represent phenomenal properties). Of the many notions of epistemic possibility, the sense I here have in mind is: possible given what we human beings now rationally believe. The relevant “we” in this case are perhaps those who have thought carefully about these philosophical issues. For this sense of epistemic possibility, I will use the term ‘open possibility.’

IS QUALITATIVE INACCURACY A SERIOUS OPEN POSSIBILITY ? My contention, then, is that, given the supposition of (i), it is an open possibility that introspective representation is inaccurate in the respect that it represents phenomenal properties as having qualitative natures they do not in fact have; that is, it is an open possibility that the qualitative inaccuracy hypothesis is true. For example, upon seeing the red tomato, Mary introspectively represents the qualitative nature of phenomenal redness in the what-it-is-like-to-sense-red way, and it is an open possibility that her representing it in this way attributes to it a qualitative nature that it actually lacks.

13 On an alternative version of the old fact–new guise response, phenomenal modes of presentation should not be taken to represent phenomenal properties as having a qualitative nature at all. Rather, they are just devices for securing reference to phenomenal properties, analogous to demonstratives. There would then be no good reason to think that the physical and phenomenal modes of presentation of phenomenal properties are not coreferential. I think that this sort of response to the knowledge argument is weakened by the plausibility of the claim that phenomenal modes of presentation represent phenomenal properties as having a qualitative nature.

15 The Knowledge Argument and Introspective Inaccuracy

The notion that there might be such a discrepancy between the real nature of phenomenal properties and the qualitative natures we introspectively represent them as having is consistent with certain claims about the correctness of introspective representation. For example, even if introspective representation inaccurately represents phenomenal properties in the sense just outlined, still it may be that a belief that I am in a phenomenal state characterized by a certain phenomenal property, a belief that is formed on the basis of an introspective representation (perhaps a belief that does not feature a specific term for the phenomenal state), is generated by a mechanism that is very reliable. So in general, there might be no discrepancy between which phenomenal states I introspectively represent myself as being in and those I am actually in— introspective representation might sort phenomenal states and properties quite accurately—while at the same time phenomenal properties lack the qualitative natures we introspectively represent them as having. I do not mean to endorse or deny the claim that whenever I am in a phenomenal state characterized by a certain phenomenal property I am in a position to know that I am in this state. Timothy Williamson has criticized this claim, which would be an instance of luminosity: for every case α, if in α C obtains, then in α one is in a position to know that C obtains.14 Williamson’s argument against luminosity for phenomenal states is controversial,15 and others, such as John Hawthorne, have suggested more restricted and perhaps more plausible variants of luminosity.16 What I do want to contend is that luminosity as applied to which phenomenal state one is in is consistent with the position that, in general, there is a discrepancy between the real qualitative nature of a phenomenal property and the qualitative nature we introspect it as having. On this view, a type of representation might successfully secure a referent by having instances that are caused by this referent, and yet misrepresent this referent by representing it as having a property that it actually lacks. Locke’s conception of sensory secondary quality representation provides an analogy. He maintains that these representations do indeed secure their referents causally, while they nevertheless misrepresent external objects in a certain respect, or, more cautiously, they are apt to do so:

14 Timothy Williamson, Knowledge and its Limits (Oxford: Oxford University Press, 2000), p. 95. 15 See, for example, Earl Conee, “The Comforts of Home,” Philosophy and Phenomenological Research 70 (2005), pp. 444–51. Conee criticizes Williamson’s argument against luminosity but proposes another. Williamson responds to Hawthorne in “Replies to Commentators,” Philosophy and Phenomenological Research 70 (2005), pp. 468–91. 16 John Hawthorne, “Knowledge and Evidence,” Philosophy and Phenomenological Research 70 (2005), pp. 452–58. Hawthorne argues that in the kinds of cases that Williamson uses to challenge luminosity, the ignorance is due to vagueness. The variant condition he suggests, coziness, precludes ignorance due to vagueness: in every case α in which, determinately, C obtains, one is in a position to know that C obtains (p. 453). Williamson responds to Hawthorne in “Replies to Commentators,” Philosophy and Phenomenological Research 70 (2005), pp. 468–91.

16 Consciousness and the Prospects of Physicalism

Ideas of primary qualities are resemblances; of secondary, not. From which I think it easy to draw the observation that the ideas of primary qualities are resemblances of them and their patterns do really exist in the bodies themselves, but the ideas produced in us by these secondary qualities have no resemblance of them at all. There is nothing like our ideas existing in the bodies themselves.17 On one interpretation (or variant) of this view, our ordinary tactile ideas of temperature represent the ambient air, the icicles above the door, or the coffee one is drinking as having certain features, while those features are incorrectly attributed to those things. On a warm day, we have a particular sort of tactile temperature representation of the ambient air, which represents the air as having a certain feature—put in Lockean terms, as having a quality that resembles the sensory temperature idea. However, if Locke is right, that quality is incorrectly attributed to the air. Many others have endorsed a position of this sort. William Alston expresses a view of this type when he says: “When I look at a shirt and take it to be red, when I feel a fabric and recognize it as very smooth, when I hear a bell ringing and recognize it as giving out a typical bell-like sound, I attribute to the perceived objects qualities that they do not, in strictness, bear.”18 A stronger claim is that our sensory experience incorrectly represents secondary qualities as primitive. In the sense at issue, a primitive property is (a) one whose entire qualitative nature or essence is revealed in our sensory (or perhaps quasi-sensory, as in the case of introspection) representation of it and thus is not identical to a property with a qualitative nature distinct from what is revealed by the sensory representation, and (b) one that is metaphysically simple and thus not constituted by, in the sense of metaphysically analyzable into, a plurality of other properties. Correlatively, properties can be represented as primitive. For the redness of a sunset to be represented as primitive requires that it be represented as having that familiar simple qualitative nature revealed in visual experience of red things under normal conditions and as not identical with any property, such as being spectral reflectance profile S or being molecular basis M of spectral reflectance profile S, whose qualitative nature is not revealed in that sensory experience.19 To my mind, it is plausible that either our sensory experience represents secondary qualities as primitive or, alternatively, as a result of how our sensory experience represents secondary qualities, we have a strong tendency to believe that they are primitive (a tendency that might be overcome by, for example, scientific 17 John Locke, An Essay Concerning Human Understanding, II, viii. For a sympathetic exposition of Locke’s position on this issue, see Michael Jacovides, “Locke’s Resemblance Theses,” Philosophical Review 108 (1999), pp. 461–96. 18 William Alston, “Mystical and Perceptual Awareness of God,” in The Blackwell Guide to Philosophy of Religion, ed. William E. Mann (Oxford: Blackwell, 2004), pp. 198–219, at p. 211. See also, for example, Paul Boghossian and David Velleman, “Colour as a Secondary Quality,” Mind 98 (1989), pp. 81–103; Barry Maund, “The Illusion Theory of Colour: An Anti-Realist Theory,” Dialectica 60 (2006), pp. 245–68. 19 Alex Byrne and David Hilbert, “Color Primitivism,” Erkenntnis 66 (2007), pp. 73–105; David Chalmers, “Perception and the Fall from Eden.”

17 The Knowledge Argument and Introspective Inaccuracy

knowledge). On this last option, let’s say that our sensory experience is primitive-belief occasioning. There is a reason to think that this second option is more credible than the first. One might question whether the specific content at issue could be represented by sensory experience. In particular, the thought that sensory experience could represent the qualitative nature of a secondary quality as exhausted by what the experience reveals seems less plausible to me than the corresponding claim about belief. By contrast with sensory experiences of secondary qualities like color, some sensory experiences of properties clearly do not represent them as primitive. When we by means of sensory experience represent a table as being made of wood, we do not represent it as having a primitive property. Evidence for this is that representing the table in this way has no tendency to occasion a belief that its being made of wood is not identical with its being made of something with a molecular structure not revealed in the sensory experience. The notion of a primitive property also applies to the phenomenal. My sense is that most of us either introspectively represent phenomenal properties as primitive or else our introspective representations of phenomenal properties are primitive-belief occasioning. But it is also an open possibility that phenomenal properties are in fact not primitive. Accordingly, the primitivist inaccuracy hypothesis is that we take phenomenal properties to be primitive in either of these two ways, while it is an open possibility they in fact are not. Note, however, that this claim is stronger than the qualitative inaccuracy hypothesis, which specifies only that when we introspect a phenomenal property, we represent it as having a specific qualitative nature that it might actually lack. It is stronger in the respect that it adds the supposition that when we introspectively represent a phenomenal property, we represent its complete qualitative essence or, alternatively, we are disposed by introspection to believe that its complete qualitative essence is introspectively represented, and also that we represent it as metaphysically simple or are disposed to believe that it is.20 The primitivist inaccuracy hypothesis also has a role in what follows, not essentially in responding to the knowledge and zombie arguments but in answering conceivability arguments such as those of René Descartes and Saul Kripke, which trade on the intuition that phenomenal or thought properties possibly exist independently of any underlying physical property (see chapter 4).21 My suggestion is that our mistaking phenomenal

20 This stronger supposition is arguably featured in what Mark Johnston and David Lewis call revelation. Johnston: “The intrinsic nature of canary yellow is completely revealed in a standard experience as of a canary yellow thing”; Mark Johnston, “How to Speak of the Colors,” Philosophical Studies 68 (1992), pp. 221–63, at p. 223. Lewis: “when I have an experience with quale Q, the knowledge I thereby gain reveals the essence of Q: a property of Q such that, necessarily, Q has it and nothing else does”; David Lewis, “Should a Materialist Believe in Qualia?” Australasian Journal of Philosophy 73 (1995), pp. 140–44, reprinted in Lewis’s Papers in Metaphysics and Epistemology (Cambridge: Cambridge University Press, 1999), pp. 325–31, at p. 328. 21 René Descartes, Meditations on First Philosophy, in The Philosophical Writings of Descartes, tr. John Cottingham, Robert Stoothoff, and Dugald Murdoch, vol. 2 (Cambridge: Cambridge University Press, 1984), p. 54 (AT VII 78); Saul Kripke, Naming and Necessity (Cambridge, MA: Harvard University Press, 1980), pp. 144–55.

18 Consciousness and the Prospects of Physicalism

properties to be primitive would explain this intuition away and thus yields a challenge to arguments of this kind. However, even the weaker contention about secondary qualities, that our sensory representations of secondary qualities are qualitatively inaccurate in the respect outlined, is controversial. Some would deny that there is any sense in which our ordinary visual color representations generally misrepresent, for the reason that what a type of representation represents is determined solely by the typical cause of its instances. Claims of this last sort have often been disputed by way of devices such as inverted spectrum thought experiments, in which what is represented in the external world is held fixed, while the phenomenal content of the representation varies. Familiarly, there is widespread disagreement about the force of the attendant argument. Nevertheless, I will make use of the secondary quality analogy, assuming the position that, for example, our ordinary visual color sensations represent physical objects as having qualitative features that are incorrectly attributed to them. This is not to say that on this type of view our secondary quality concepts simply misrepresent those qualities, even though they have their source in sensory representations that in a sense misrepresent them—more on this later as well. A more localized example is that, as Descartes pointed out, from a certain distance we visually represent square towers as round, while the property of roundness is incorrectly attributed to the tower.22 Another is that many (but not all) of us visually represent the lengths of the Müller-Lyer pair of lines as different, while they are in fact the same. It is the open possibility of an analogous disparity between how phenomenal properties are represented introspectively and their real nature that would generate the physicalist response. In the proposed open possibility, the specified kind of inaccuracy is universal—it is a feature of all human introspective representation of phenomenal properties, and it is in a significant respect extensive—phenomenal properties altogether lack certain qualitative natures they are represented as having. In these respects, the inaccuracy at issue differs from the sort featured in the visual representation of the lengths of pairs of lines, or of shapes of objects from a significant distance.23 It might be argued that the universality and extensiveness of the proposed inaccuracy provide reason to believe that the open possibility under consideration is unlikely to be actual. However, on the Locke-inspired view of our sensory representations of secondary qualities, which is not implausible, the inaccuracy of these representations is similarly universal and extensive. This analogy provides at least some reason to believe that the proposed open possibility about phenomenal property representation is not unlikely to be actual. Moreover, as I’ve pointed out, this open possibility is compatible with introspective representation reliably generating true beliefs about which phenomenal state the subject is in. So this possibility can preserve the

22 René Descartes, Meditations on First Philosophy, in The Philosophical Writings of Descartes, vol. 2, p. 53 (AT VII 76). 23 Louis deRosset made this point in conversation.

19 The Knowledge Argument and Introspective Inaccuracy

accuracy of many introspectively-based beliefs about phenomenal states, and the extensiveness of introspective misrepresentation of phenomenal states that it involves is for this reason actually fairly limited. In the case of our visual color representations, it is their specific causal nature that plausibly allows for a disparity of this kind. By the standard causal theory of such representations, they are typically caused by external objects (perhaps by their propertyinstances) and are distinct from them, and they mediate the subject’s representing of the object. This mediation by these representations can result in a disparity between the features objects appear to have and those they really have, and so representation may in certain respects be inaccurate. For instance, they might represent the color of an object as having a specific qualitative nature, while the attribution of this nature to the property is incorrect. It is an open possibility that our introspective representation of phenomenal properties is similarly causal, whereupon a guarantee of the accuracy of how introspection represents these properties would be precluded.24 Noncausal theories of introspective representation are also contenders. One might, with Franz Brentano, endorse a self-presentation view and argue that a token sensation of green is on the one hand a sensation of green, while that very sensation is also an experience of itself.25 Alternatively expressed, besides representing to the subject the property of being green, this sensation also simply presents itself to her without the mediation of a (further) representation of it. So in one kind of case—when a sensation is an experience of itself—representation of something occurs without causal

24 Here I would advocate what Shoemaker calls “the broad perceptual model of introspection, according to which introspection is in some key respects disanalogous to visual representation, but still causal”; see “Self-Knowledge and ‘Inner Sense,’” Philosophy and Phenomenological Research 54 (1994), pp. 249–314, reprinted in Sydney Shoemaker, The First-Person Perspective and Other Essays (Cambridge: Cambridge University Press, 1996), pp. 201–68); my pagination is from the latter source. Shoemaker’s main argument against this position is that it would have to admit the possibility of selfblind subjects, who have mental states but do not have introspective access to them: “To be self-blind with respect to a certain kind of mental fact or phenomenon, a creature must have the ability to conceive of those facts and phenomena. . . . And it is only introspective access to those phenomena that the creature is supposed to lack” (p. 226). Alex Byrne, in “Introspection,” Philosophical Topics 33 (2005), pp. 79–104, at pp. 89–92, develops a set of objections to Shoemaker’s self-blindness argument, which in my view carry appreciable weight. My suggestion is that the apparent impossibility of selfblindness results from the reliability of a causal mechanism by which introspective representations of first-order mental states are produced by those states, and that this apparent impossibility does not force acceptance of Shoemaker’s own view, according to which “there is a conceptual, constitutive connection between the existence of certain sorts of mental entities and their introspective accessibility” (p. 225). See also David Armstrong, “The Nature of Mental States,” in Readings in the Philosophy of Psychology, vol. 1, ed. Ned Block (Cambridge, MA: Harvard University Press, 1980), pp. 191–99, and the following discussion. 25 Franz Brentano, Psychology from an Empirical Standpoint, tr. A. C. Rancurello, D. B. Terrell, and L. L. McAlister (London: Routledge and Kegan Paul, 1973), pp. 153–54. Uriah Kriegel develops this position in Subjective Consciousness: A Self-Representational Theory (Oxford: Oxford University Press, 2009).

20 Consciousness and the Prospects of Physicalism

mediation. Representation is instead reflexive and noncausal.26 Perhaps a self-presentation view meshes with certain of our ordinary intuitions about our consciousness of sensation. It is also an open possibility. Or with Chalmers and David Papineau, one might advocate a constitution view for (pure) phenomenal concepts. Chalmers says “one might say very loosely that the referent of the concept is somehow present inside the concept’s sense, in a way much stronger than in the usual cases of ‘direct reference’ . . . in the phenomenal case, the epistemic content itself seems to be constituted by the referent.”27 In Papineau’s conception, “the use of a phenomenal concept to refer to some experience will standardly involve the thinker actually having the experience itself, or a faint copy of it. Perhaps . . . we should think of this instantiation of the experience as literally part of the term the thinker uses to refer to that experience.”28 Papineau goes on to develop a quotational model according to which phenomenal concepts involve a frame of the form ‘the experience: —,’ where the blank is filled in with a particular phenomenal experience or a faint copy of it. In an imaginative use of a phenomenal concept, it refers to any experience that appropriately resembles this experience or its faint copy.29 Here again, a phenomenal property could be represented without causal mediation. But note that the self-presentation view of phenomenal representations does not obviously preclude the qualitative inaccuracy hypothesis. Self-presenting sentences can misrepresent in some respect (while being accurate in another). ‘This German sentence has six words,’ for instance, represents itself as having a feature it lacks—as being a German sentence (while it accurately represents itself as having six words).30 It may be, then, that nothing we understand rules out the possibility that self-presenting phenomenal states are qualitatively inaccurate in the sense I’ve specified. Perhaps a

26 Christopher Hill and Brian McLaughlin explain this position as follows: Sensory states are self-presenting states: we experience them, but we do not have sensory experiences of them. We experience them by being in them. Sensory concepts are recognitional concepts: deploying such concepts, we can introspectively recognize when we are in sensory states simply by focusing our attention directly on them. Matters are of course quite different in the case of perceptual and theoretical concepts. An agent’s access to the phenomena that he or she perceives is always indirect: it always occurs via an experience of the perceived phenomena that is not identical with the perceived phenomena, but rather caused by it. (Christopher Hill and Brian McLaughlin, “There Are Fewer Things in Reality Than Are Dreamt of in Chalmers’s Philosophy,” Philosophy and Phenomenological Research 59 [1999], pp. 445–54, at p. 448) 27 David Chalmers, “The Content and Epistemology of Phenomenal Belief,” pp. 13–14. 28 David Papineau, Thinking about Consciousness (Oxford: Oxford University Press, 2002), p. 105. 29 David Papineau, Thinking about Consciousness, pp. 116–21. 30 Mark Moyer and Brian Weatherson each made this point about self-presenting sentences and suggested that the possibility of misrepresenting self-presentations would strengthen the argument. Louis deRosset provided the example of a self-presenting sentence that is accurate in one respect and inaccurate in another.

21 The Knowledge Argument and Introspective Inaccuracy

constitution view also does not preclude such qualitative inaccuracy—this would depend on the details of the theory. Still, I think that the stronger case for my position can be made by analogy with secondary-quality representation, and here it is reasonable to believe that qualitative inaccuracy is partly due to causal mediation and not to the sort of problem that arises in the case of misrepresenting self-presenting sentences. At the same time, if the requisite kind of qualitative inaccuracy for introspective representations of phenomenal properties would be an open possibility even if phenomenal states were self-presenting, then my case would be stronger.

A DISANALOGY WITH EXTERNAL PERCEPTUAL REPRESENTATION It may seem highly intuitive that phenomenal properties are introspectively represented in an intimate way that guarantees that their qualitative nature is represented accurately. The qualitative nature of the color of a physical object might not be accurately represented by our ordinary sensory representations, but how could the qualitative nature of pleasure, or the qualitative nature of one’s visual sensation of red, not be as they are introspectively represented? Daniel Dennett expresses skepticism about the coherence of this sort of introspective misrepresentation, affirmation of which he traces to “the image of the Cartesian Theater”: The Cartesian Theater may be a comforting image because it preserves the reality/appearance distinction at the heart of human subjectivity, but as well as being scientifically unmotivated, this is metaphysically dubious, because it creates the bizarre category of the objectively subjective—the way things actually, objectively seem to you even if they don’t seem to seem that way to you! (Smullyan (1981)) Some thinkers have their faces set so hard against “verificationism” and “operationalism” that they want to deny it even in the one arena where it makes manifest good sense: the realm of subjectivity.31 Although the epistemic possibility of our introspectively misrepresenting the qualitative natures of phenomenal properties might be at odds with strong intuitions, still its being an open possibility is forced on us by the prospect that introspection might be causal on analogy with visual color representation. Furthermore, the reason we allow qualitative misrepresentation for properties of external objects like colors but tend to resist its possibility for introspective phenomenal-property representation might be due to differences between these kinds of representations that do not, in the last analysis, count decisively against qualitative inaccuracy in the introspective case. Let me explain. 31 Daniel Dennett, Consciousness Explained (Boston: Little, Brown, 1991), p. 132. The work by R. M. Smullyan that Dennett cites is “An Epistemological Nightmare,” in The Mind’s I: Fantasies and Reflections on Self and Soul, ed. D. R. Hofstadter and Daniel Dennett (New York: Basic Books, 1981), pp. 415–27.

22 Consciousness and the Prospects of Physicalism

In our sensory representation of colors of external objects, we fairly easily and not infrequently become aware of a difference between the real nature of the property represented and how it happens to be represented on some occasion. The car appears to have a different color under the sodium vapor lights than it does in natural light, but it is clear that nothing about the car itself has changed, and so we come to believe that there is a discrepancy between the car’s real color and the way it is visually represented under the unusual lighting conditions. But for phenomenal properties, awareness of analogous discrepancies does not readily occur. Perhaps such awareness sometimes arises, but the difficulty of adducing examples indicates that it is at best a rare phenomenon. This might, of course, count as good evidence that introspection does not and even cannot misrepresent the qualitative nature of a phenomenal property. Yet at the same time, if there were such misrepresentation, the infrequency of its detection would help explain why we would resist its possibility. It would arguably count in favor of the position I’m defending if our becoming aware of a discrepancy between the real qualitative nature of a phenomenal property and how it is introspectively represented sometimes but nonetheless very rarely occurs. Then we would have reason for thinking that such introspective misrepresentation is possible, and we would also have an explanation for our resistance to this possibility. I don’t believe that there are uncontroversial examples of our becoming aware of such discrepancies, but here are two candidates. First, Christopher Hill cites the following case, presented by Rogers Albritton in a seminar: The case involves a college student who is being initiated into a fraternity. He is shown a razor, and is then blindfolded and told that the razor will be drawn across his throat. When he feels a sensation he cries out: he believes for a split second that he is in pain. However, after contemplating the sensation for a moment, he comes to feel that it is actually an experience of some other kind. It is, he decides, a sensation of cold. And this belief is confirmed when, a bit later, the blindfold is removed and he is shown that is throat is in contact with an icicle rather than a razor.32 There are a number of ways to analyze this example, but one possibility is that in his introspective awareness, the fraternity pledge at first misrepresents the qualitative features of the sensation of cold he actually has as qualitative features of pain, and later it becomes clear to him that he has misrepresented them. This is a controversial analysis but not an implausible one.33

32 Christopher Hill, Sensations: A Defense of Type Materialism (Cambridge: Cambridge University Press, 1991), pp. 128–29. 33 Hill takes this example to provide evidence that we make errors of judgment in our introspection-based beliefs about sensation, where errors of judgment “are usually due either to some form of inattention or to the influence of expectation upon judgment.” He differentiates between errors of judgment and errors of ignorance, which occur “when beliefs are based on appearances that fail to do justice to the entities to which the beliefs refer.” Hill claims “we are perforce innocent of committing

23 The Knowledge Argument and Introspective Inaccuracy

A second example involves my daughter, on the occasion of her requiring a Novocain shot at the dentist’s. Rather than simply showing her the needle in advance and then giving her the injection, the dentist hid the needle from her and told her that he would be dropping bits of cold water into her mouth. She didn’t flinch. When I asked her afterward whether the experience was unpleasant, she said that she didn’t like the drops of water much, but they didn’t hurt. In this case, it may be that the dentist’s suggestion, together with his hiding the needle, kept her from introspectively representing the qualitative features of the pain state she was actually in as qualitative features of pain; instead, she misrepresented those features as qualitative features of a sensation of cold. This is also a controversial analysis but again not implausible. So far, we have some reason to believe that we are infrequently but nevertheless sometimes can be aware of a discrepancy between the real qualitative nature of a phenomenal property and how it is introspectively represented. Is there a further factor that would help explain why we are resistant to the possibility of such qualitative inaccuracy consistent with the supposition that it does in fact exist? In the case of external sensory representation, we have readily available ways of checking the entity represented that are independent of the representation under scrutiny, while for introspection such ways of checking are at best very limited. One might have a closer look at Descartes’s tower to test whether one’s visual representation of its shape as round was accurate, or measure the Müller-Lyer lines to determine whether one’s visual representation of them as having different lengths was correct. Analogously decisive ways of checking introspected phenomenal properties are not available to us. The icicle and Novocain cases exemplify the best we can do. This limitation yields an explanation as to why we are at most only infrequently aware of discrepancies between the real qualitative natures of phenomenal properties and how they are introspectively represented, which to my mind provides a fairly plausible account of our resistance to the possibility of qualitative inaccuracy consistent with its actually existing. Moreover, given that awareness of a discrepancy between the real nature of a phenomenal property and the qualitative nature we introspectively represent it as having seldom, if ever, arises, and given the scarcity of means of checking the accuracy of such representations, there would be little if any noticeable difference between having an introspective experience in which we represented phenomenal properties causally and as possessing qualitative features they actually lacked, and having one in which phenomenal properties were self-presenting without such misrepresentation. Thus what we do and do not notice in having introspective experience, all by itself,

errors of ignorance in forming beliefs about our own sensations”; see Sensations, pp. 127–28. The open possibility I am envisioning would have us making errors of ignorance in our introspection-based beliefs about phenomenal properties, since such beliefs would be based on appearances that fail to do justice to the real qualitative nature of those properties.

24 Consciousness and the Prospects of Physicalism

will not adjudicate whether we misrepresent the qualitative natures of phenomenal properties.34

A RESPONSE TO THE KNOWLEDGE ARGUMENT While Mary is in the room, she does not represent phenomenal states in the characteristic introspective way, and she does not appear to have the information required to represent the complete real natures of these phenomenal states by deriving them from what she knows. But it is a serious open possibility that by virtue of her physical knowledge she can nevertheless accurately represent the complete real natures of these states. Phenomenal properties of these states, in particular, might not have the qualitative natures they are introspectively represented as having. Instead, the natures of these properties might accurately be represented by way of Mary’s physical knowledge. If this possibility is actual, then from her physical knowledge, she can derive every truth about the real natures of phenomenal states.35 How exactly does this story yield a response to the knowledge argument? Let’s focus on which true beliefs, and not which knowledge, Mary can have before and after she leaves the room. What is germane to the knowledge argument when it comes to the states she can have while she is in the room is just that they are true beliefs, and thus we can set aside the complex concerns that are specific to knowledge. The key issue is

34 Stephen Wykstra proposes the following plausible condition of reasonable epistemic access: “On the basis of cognized situation s, human H is entitled to claim ‘It appears that p’ only if it is reasonable to believe that, given her cognitive faculties and the use she has made of them, if p were not the case, s would likely be different than it is in some way discernible by her”; see “The Humean Obstacle to Evidential Arguments from Suffering; On Avoiding the Evils of ‘Appearance,’” International Journal for Philosophy of Religion 16 (1984), pp. 73–93, reprinted in The Problem of Evil, ed. M. M. Adams and R. M. Adams (Oxford: Oxford University Press, 1980), pp. 138–60, at p. 152). By Wykstra’s criterion, we would not be justified in claiming that it appears that introspective representation of phenomenal properties is noncausal and qualitatively accurate. 35 David Chalmers, “Phenomenal Concepts and the Explanatory Gap,” in Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism, ed. Torin Alter and Sven Walter (New York: Oxford University Press, 2006), pp. 167–94, points out that on the sort of view advocated by Loar, Levine, and others—the phenomenal concept strategy—it is maintained that zombies are ideally, positively, primarily conceivable, while our having phenomenal concepts has a physical explanation. These are two key features of what Chalmers calls Type-B materialism, a widely held view. He argues that this position is unstable. I suspect that he is right about this and that in the last analysis, physicalism requires denying the ideal, positive, primary conceivability of zombies, for then it would avoid the tension between affirming this sort of zombie-conceivability, which has the consequence that phenomenal properties are in the crucial sense not physically explainable, and claiming that our having phenomenal concepts is physically explainable. Another key feature of the phenomenal concept strategy is its claim that the entailment of any phenomenal truth by the complete physical truth is a posteriori, and Daniel Stoljar (in “Physicalism and Phenomenal Concepts”) argues that it cannot adequately explain how this can be so. The response I’ve developed here does not require that this conditional is a posteriori, and thus it avoids the issue Stoljar highlights. At the end of chapter 4, I suggest a way to strengthen the phenomenal concepts strategy.

25 The Knowledge Argument and Introspective Inaccuracy

whether upon leaving the room and seeing the red tomato Mary acquires a true belief that is new in the sense that she did not have it while she was in the room, and it is not derivable from the true beliefs she had then. We’ll initially suppose that our open possibility is actually realized, but subsequently we’ll discharge this supposition and instead think of the open possibility as a hypothesis about how things might turn out. We’ll then ask the crucial question: do we theorists now have a reason to believe that Mary has not acquired a new true belief? So on the supposition that the open possibility is in fact realized, how should we describe what happens when Mary leaves the room and sees the red tomato? We imagine her now having a belief of the form: (A) Seeing red has R. Consider first the initially plausible proposal (i) that the concept ‘R’ in this belief refers to a property with the qualitative nature accurately represented by the introspective what-it-is-like-to-sense-red mode of presentation. On our open possibility, phenomenal redness has no such qualitative nature, so this belief will be false. Thus she does not acquire a new true belief. Next, consider the perhaps initially less plausible proposal (ii) that ‘R’ refers to a property with a physical qualitative nature that appears to Mary in the what-it-is-like-to-sense-red way but is misrepresented by this introspective mode of presentation. Under this interpretation, we can suppose that this belief is true, but since while she was in the room, she already believed the truth expressed by it, or was able to derive it from the true beliefs she already had, she also does not acquire a new true belief. Now consider the open possibility just as a hypothesis about how things might turn out to be. Does this give us, as theorists, a reason to believe that Mary hasn’t acquired a new true belief? In my estimation, the open possibility is serious enough to provide us with such a reason. In fact, my sense is that this possibility is sufficiently serious to preclude rational conviction that Mary does acquire a new true belief, and herein lies the challenge to the knowledge argument. But I have not yet presented my entire argument for this claim; that task will be extended to the next several chapters. How high would our rational credence be that she acquires a new true belief? I’ll leave this to the reader to decide, but I have more to offer by way of an argument that it is not especially high. But note that this is consistent with this rational credence nevertheless being quite substantial. As I conceive it, the seriousness of the open possibility does not provide us with good reason to believe that the antiphysicalist consideration raised by the knowledge argument has no force, but it does yield a significant reason to believe that this consideration falls short of establishing that physicalism is false.36 Given the qualitative inaccuracy hypothesis, which option is to be preferred: that ‘R’ refers to property with the qualitative nature accurately represented by the

36 Thanks to Nico Silins for discussion of these issues.

26 Consciousness and the Prospects of Physicalism

introspective what-it-is-like-to-sense-red mode of presentation, on which Mary’s belief of form (A) is false, or that ‘R’ refers to a property with a physical qualitative nature that appears to her in the what-it-is-like-to-sense-red way but is misrepresented by this introspective mode of presentation, whereupon the belief is true but not new? The answer I propose is: both. I will develop this answer in the next chapter. This account is inspired by and exactly analogous to Chalmers’s dual-content view for perception, on which our perceptual representation of color, for example, has a perfect, primitive content, which is inaccurate in our world, and an ordinary content, which is in a sense nonideal but actually accurate.37 Chalmers’s view is eliminativist about primitive color but not about color. Similarly, the resulting view about phenomenal representation is eliminativist about primitive phenomenal properties and those that are accurately introspectively represented, but not about phenomenal properties themselves.

MERELY SHIFTING THE PROBLEM? One may now ask whether the problem for a physicalist explanation of consciousness has merely been shifted from accounting for phenomenal states and their properties to accounting for their introspective phenomenal modes of presentation. Supposing that the way phenomenal states are represented introspectively might be inaccurate in the way specified, and that Mary can derive every truth about the real nature of phenomenal states from the physical base, the pressing issue is now to assess whether these introspective phenomenal modes of presentation, or states featuring these modes of presentation, could have a physical account.38 Chalmers develops this point as an objection to the old fact–new guise strategy. He contends that even if what Mary gains when she leaves the room is only knowledge of an old fact under a different mode of presentation—then there must be some truly novel fact that she gains knowledge of. In particular, she must come to know a new fact involving that mode of presentation. Given that she already knew all the physical facts, it follows that materialism is false. The physical facts are in no sense exhaustive.39 Torin Alter raises a similar objection to an earlier exposition of my account, and he makes a specific proposal for which fact about the mode of presentation Mary learns: How color sensations appear from the first-person perspective is itself a fact about them. Therefore, if when Mary is released she learns how they appear from the first-person perspective, then she learns a new fact about them. This is

37 David Chalmers, “Perception and the Fall from Eden.” 38 I consider this objection in Derk Pereboom, “Bats, Brain Scientists, and the Limitations of Introspection,” Philosophy and Phenomenological Research 54 (1994), pp. 315–29, at pp. 323–26. 39 David Chalmers, The Conscious Mind, p. 142.

27 The Knowledge Argument and Introspective Inaccuracy

true regardless of whether this appearance accurately reflects the way they really are.40 On the qualitative inaccuracy hypothesis, Mary’s color sensation does not have a property whose qualitative nature is accurately represented by her introspective phenomenal mode of presentation—call this MPR. So Mary does not learn that the sensation has a property of this particular sort. But it would seem that she does learn something about how MPR presents this sensation. Since MPR presents the sensation phenomenally, she would appear to learn something about a phenomenal property of MPR—specifically, something about its essential property of presenting red sensations in the what-it-is-like-to-sense-red phenomenal way. Could Mary derive the corresponding phenomenal truth about MPR from her microphysical base? It’s initially intuitive that this wouldn’t be possible for her. In response, there is no less reason to think that the qualitative inaccuracy hypothesis holds for introspective representations of phenomenal modes of presentation than it does for introspective representations of first-order phenomenal states. It is then also an open possibility that Mary introspectively represents MPR’s essential phenomenal property as having a qualitative nature it really lacks. This would be due to her representing MPR and its essential property by a higher-order introspective phenomenal mode of presentation that generates a qualitatively inaccurate representation. But then, despite how MPR is introspectively represented, it might be that while she is still in the room Mary can derive every truth about its real nature from her physical base. So even though pre-emergence Mary has never introspectively represented MPR, it is an open possibility that she can derive every truth about it. The same point might be made for any further iteration of introspective representations of introspective phenomenal modes of presentation. All information about the real nature of any phenomenal entity would then be derivable from the information Mary has prior to leaving the room. I’ve not infrequently heard voiced the concern that this view engenders an unwelcome infinite regress of some kind. On one version of this objection, this account has it that when I represent an introspective representation, I do so by way of a mode of presentation, whereupon I would represent this mode of presentation by a further introspective mode of presentation, and I would represent that further mode of presentation by a yet further introspective mode of presentation, ad infinitum. So when I represent an introspective representation, I would actually have an infinite series of introspective representations, which is absurd. The story just told about Mary must thus be one in which when she introspectively represents her sensation of red, she also introspectively represents her introspective mode of presentation MPR of that sensation, while no actual infinite regress of introspective representations is generated. First of all, a mode of presentation can function 40 Torin Alter, “Mary’s New Perspective,” Australasian Journal of Philosophy 73 (1995), pp. 582– 84. I present the earlier account in Derk Pereboom, “Bats, Brain Scientists, and the Limitations of Introspection.”

28 Consciousness and the Prospects of Physicalism

as the way a subject represents an introspective representation without that subject also representing the mode of presentation itself. She might, in addition, represent this mode of presentation, but this would be a distinct representation that is not necessitated. If she did represent the mode of presentation, it could be by way of a higher order introspective mode of presentation.41 However, it would again not be necessitated that she also represents this higher order mode of presentation. Furthermore, it’s plausible that when someone introspectively represents a sensation of red by way of MPR, she will normally, although not necessarily, also represent MPR by a higher order mode of presentation, but only in unusual cases would she introspectively represent that higher order mode of presentation. This pattern would be explained by the presence of a causal mechanism by which introspective representations of first-order mental states are reliably produced by those states, and by which in cases of introspective representations of sensations phenomenal modes of presentation are also reliably introspectively represented, while it would not typically produce further introspective representations at higher levels of iteration. On a view of this kind, which I take to be a serious contender, no actual infinite regress of introspective representations is generated.42 The success of the knowledge argument depends on phenomenal properties not lacking the qualitative natures we introspectively represent them as having. Alter and Chalmers are right to argue that the standard old fact–new guise response to the knowledge argument transfers the physicalism-challenging consideration from a phenomenal state or property to its introspective phenomenal mode of presentation. But the qualitative inaccuracy move can be reiterated for such introspective modes of presentation. This response does not require, incredibly, that when I introspect phenomenal redness, I actually represent the introspective phenomenal mode of presentation MPR of that property, and an introspective phenomenal mode of presentation for MPR, on to infinity. The position that results is no longer genuinely in the old fact– new guise category, for in the open possibility, pre-emergence Mary can know everything there is to know about MPR, and thus in the relevant sense this guise is not new. Still, prior to emerging from the room, Mary never represented a phenomenal state or property by means of this mode of presentation, and thus one might say that her deployment of MPR is new. On this account, I am not introspectively acquainted with the qualitative nature of any phenomenal property or of any phenomenal mode of presentation, as it is in itself. While this consequence may constitute an initially surprising limitation of our representational abilities, upon reflection we might agree with Kant that it is not implausible that our powers are restricted in this way.

41 Nico Silins, in conversation, suggests that it is implausible that beyond some fairly low level of iteration our mental states are introspectively represented by way of phenomenal modes of presentation. At some level, I form only a belief, without distinctive phenomenology, that I am representing a mental state. Such a belief, since it is nonphenomenal, would not yield further ammunition for the knowledge argument. 42 See note 24 for a contrast with Shoemaker’s position on this issue.

2 PHENOMENAL CONCEPTS AND THE EXPLANATORY GAP

An important objection to the response to the knowledge argument just developed is that it misconstrues the nature of our paradigmatic phenomenal concepts. In this chapter, I begin by defending an answer to this objection. In the process, I propose an account of phenomenal concepts inspired by Hilary Putnam, Frank Jackson, and David Chalmers and a view about the content of phenomenal property representation analogous to the dual-content theory of secondary quality representation advanced by Chalmers. In addition, I contend that the previous chapter’s response can withstand an objection that Sydney Shoemaker directs against projectivist accounts of sensory secondary quality representation. Finally, I argue that all of this provides the physicalist with an effective reply to those, like Joseph Levine and Robert Adams, who suggest that there is an explanatory gap between the physical and the phenomenal that we do not understand how to close.1

PHENOMENAL CONCEPTS AND CONCEPTUAL ANALYSIS Against the response of the previous chapter one might contend that analysis of our phenomenal concepts reveals that they apply correctly to properties with a qualitative nature accurately represented by introspection, and that it is ruled out conceptually that they correctly apply to properties with a qualitative nature not accurately represented in this way. Chalmers suggests an idea of this sort when he specifies that the referent of a pure phenomenal concept is present inside the concept’s sense and that its content is constituted by the referent.2 Some of his physicalist opponents concur. 1 Joseph Levine, “Materialism and Qualia: The Explanatory Gap,” Pacific Philosophical Quarterly 64 (1983), pp. 354–61, and Purple Haze; Robert Adams, “Flavors, Colors and God,” in Adams, The Virtue of Faith (Oxford: Oxford University Press, 1987), pp. 243–62. 2 David Chalmers, “The Content and Epistemology of Phenomenal Belief,” pp. 13–14. 29

30 Consciousness and the Prospects of Physicalism

Brian Loar, for example, argues that phenomenal concepts express the very properties they pick out. In his framework, a concept expresses its reference-fixer. He is thus contending that reference-fixers of phenomenal concepts, which I call the introspective phenomenal modes of presentation of phenomenal properties, are just the properties these concepts pick out. Moreover, he claims that: Phenomenal concepts pick out certain properties directly. They do not pick out those properties via a contingent mode of presentation, in the manner say of visual recognitional concepts, which connect one to some external kind by way of a visual experience. It could then seem, I suppose, that phenomenal concepts conceive their referents as they are in themselves.3 Loar is plausibly interpreted as endorsing the claim that introspective phenomenal modes of presentation accurately represent qualitative natures of the properties they pick out, since these modes of presentation just are the properties they pick out. A stronger view is that we take phenomenal properties as primitive, as metaphysically simple properties whose entire qualitative essence is revealed by introspective representation, and that our phenomenal concepts directly pick out such primitive properties. Colin McGinn provides a clear exposition of a position of this sort. He argues that we have Russellian acquaintance with our own conscious states and that therefore “we know what consciousness is by means of knowledge by acquaintance.”4 Citing Russell, McGinn contends that “when I am directly aware of my own consciousness, I know it perfectly and completely.” Furthermore, he affirms that some of our concepts are acquaintance-based, and by this he means that our grasp of such concepts is dependent on acquaintance with what they stand for. He then suggests: The general concept of consciousness is acquaintance-based: we are acquainted with consciousness and our concept of consciousness depends upon this acquaintance. To have the concept of consciousness is to know what consciousness is (in one sense), and this knowledge is produced by our acquaintance with consciousness.5 In McGinn’s view, the epistemology of our concept of consciousness is enough to generate the mind-body problem, “for if we know the essence of consciousness by means of acquaintance, then we can just see that consciousness is not reducible to neural or functional processes (say)—just as acquaintance with the color red could

3 Brian Loar, “David Chalmers’s The Conscious Mind,” Philosophy and Phenomenological Research 59 (1999), p. 471; cf. Brian Loar, “Phenomenal States,” in The Nature of Consciousness: Philosophical Debates, ed. Ned Block, Owen Flanagan, and Guven Güzeldere (Cambridge, MA: MIT Press, 1997); for a similar view, see David Papineau, Thinking about Consciousness, pp. 96–140. 4 Bertrand Russell, “Knowledge by Acquaintance and Knowledge by Description,” Proceedings of the Aristotelian Society 11 (1910), pp. 108–28. 5 Colin McGinn, “What Constitutes the Mind-Body Problem,” p. 11.

31 Phenomenal Concepts and the Explanatory Gap

ground our knowledge that redness not the same as greenness.”6 The essence or nature of consciousness is transparent to anyone who has a concept generated by acquaintance with it, and thus this concept yields knowledge of what consciousness is in itself. Moreover, concepts such as ‘C-fiber firing’ and ‘pain’ are so different that there can be no a priori entailments between them, and, in addition, no concepts that our cognitive faculties might provide can bridge this gap. For this reason, “we aren’t going to solve the mind-body problem, as we are currently cognitively constituted.”7 In response, a plausible account of phenomenal concepts shows this sort of objection can be contested. Although it is attractive to suppose with McGinn that an acquaintance-based general concept of consciousness gives us knowledge of the essence of consciousness and to think that a similar claim is true for acquaintance-based concepts of specific phenomenal properties, we should not be confident that this is true. It is open that introspection does not accurately represent the qualitative nature of a phenomenal property and that therefore an acquaintance-based concept of a phenomenal property will not provide us with knowledge of this qualitative nature as it is in itself. However, we should not assume that on this possibility phenomenal concepts would simply misrepresent phenomenal properties. By analogy, on Locke’s proposal, secondary quality concepts do not misrepresent secondary qualities at all. What is it, exactly, that conceptual analysis reveals about phenomenal concepts? An attractive template, deriving from Putnam, is that the structure of certain concepts is a conjunction of conditionals.8 On a model inspired by Jackson and Chalmers, the antecedents of the conditionals specify possible worlds considered as actual—that is, possible worlds considered as the way things actually turn out—and the consequents indicate what the concept in question then correctly applies to (the next chapter considers these and related notions in more detail). Which conditional actually applies depends on the way the actual world is; it is the one whose antecedent is actually true. This structure is discerned by reflection on possible scenarios—Jackson and Chalmers make an impressive case that the sort of reflection on possible scenarios that we see in Putnam’s work might be thought of as conceptual analysis.9 6 Colin McGinn, “What Constitutes the Mind-Body Problem,” p. 11. 7 Colin McGinn, “What Constitutes the Mind-Body Problem,” pp. 20–21. 8 Hilary Putnam, “The Meaning of ‘Meaning,’” in his Philosophical Papers, vol. 2 (Cambridge: Cambridge University Press, 1975), pp. 240–41. Some of the essential elements of this view are developed by Rudolf Carnap in “Meaning and Synonymy in Natural Languages,” Philosophical Studies 6 (1955), pp. 33–47. A position of this general type is endorsed by Ned Block and Robert Stalnaker, “Conceptual Analysis, Dualism, and the Explanatory Gap,” Philosophical Review 108 (1999), pp 1–46, at p. 36; by David Chalmers and Frank Jackson, “Conceptual Analysis and Reductive Explanation,” esp. pp. 322, 340–41; and by George Bealer, “Modal Epistemology and the Rationalist Renaissance,” in Conceivability and Possibility, ed. Tamar Szabó Gendler and John Hawthorne (Oxford: Oxford University Press, 2002), pp. 77–125, at p. 109. 9 Chalmers and Jackson (“Conceptual Analysis and Reductive Explanation,” p. 322) write: “When given sufficient information about a hypothetical scenario, subjects are frequently in a position to identify the extension of a given concept, on reflection, under the hypothesis that the

32 Consciousness and the Prospects of Physicalism

For example, given that all of our samples of the watery stuff in our environment are composed of H2O, and this chemical structure explains the properties we associate with water, our concept ‘water’ correctly applies (just) to H2O, and water = H2O. But suppose that it turned out instead that the watery stuff, like our samples of jade, had two distinct kinds of composition, each at least fairly common. Then claiming that ‘water’ correctly applies only to H2O would be implausible, and, like jade, it would have turned out that water was a disjunctive kind.10 Or imagine that instead it turned out that the watery stuff had many distinct compositions with no salient similarities among their intrinsic features, while each sample nevertheless exemplified a wellbehaved functional characterization. Then, like ‘catalyst’ and ‘enzyme,’ we might have rightly counted water as a functional kind. Or suppose it turned out that Berkeley’s view of the universe was correct and that water was composed just of sensations directly produced in our minds by God. Then we might have classified water as an appearance kind, so that ‘water’ applied correctly to anything that appeared in a particular way under certain conditions and in different particular ways under other conditions.11 On this proposal, analysis reveals that our concept ‘water’ has a structure of the following sort: If a world is actual in which the watery stuff in the environment has a unique sort of composition, then the concept ‘water’ correctly applies to a unique compositional stuff; and if a world is actual in which the watery stuff has a small number of sorts of composition, then the concept ‘water’ correctly applies to a disjunctive compositional stuff; and if a world is actual in which the watery stuff has many sorts of composition, and in which there are no salient similarities among the intrinsic properties of these compositions, while each sample of the watery stuff exemplifies a well-behaved functional characterization, then the concept ‘water’ correctly applies to a functional kind,

scenario in question obtains. Analysis of a concept proceeds at least in part through consideration of a concept’s extension within hypothetical scenarios, and noting the regularities that emerge.” Cf. Frank Jackson, From Metaphysics to Ethics (Oxford: Oxford University Press, 1998), pp. 28–86. 10 The ‘would have turned out that S, had it turned out that W’ locution derives from Stephen Yablo, “Shoulda, Woulda, Coulda,” in Conceivability and Possibility, pp. 441–92, at p. 454. 11 David Braddon-Mitchell makes a related point in “Qualia and Analytical Conditionals,” Journal of Philosophy 100 (2003), pp. 111–35, at p. 115. Braddon-Mitchell also suggests that the analysis of some concepts might be a conjunction of conditionals. Berkeley’s view is developed in A Treatise Concerning the Principles of Human Knowledge, ed. Jonathan Dancy (Oxford: Oxford University Press, 1998); and in Three Dialogues between Hylas and Philonous, ed. Jonathan Dancy (Oxford: Oxford University Press, 1998).

33 Phenomenal Concepts and the Explanatory Gap

and if a world is actual in which each instance of the watery stuff is a collection of sensations produced directly in minds by God, then the concept ‘water’ correctly applies to an appearance kind. . . .12 For certain concepts, the plausibility of this picture serves as a corrective to the idea that conceptual analysis alone can determine, in effect, that a specific conditional actually applies. It is sometimes assumed, for example, that conceptual analysis alone shows that ‘water’ refers to a unique compositional stuff. But this would then not be so, since analysis would reveal a conjunction of conditionals. Which of the conjuncts actually applies would be settled by the actual world, and we would know which conjunct actually applies only by our investigation of the actual world. This model permits a concept to remain the same through changes in our scientific theories about what it correctly applies to, or (more salient for present purposes) through a more rudimentary change from a situation in which we rely only on the manifest image for its conditions of correct application to one in which we are informed by an empirically confirmed scientific theory. The model allows that the concept persists through such changes, while the conditional held actually to apply varies.13 Returning to our color analogy, one might at first assume: C1. The concept ‘red’ correctly applies to the property of objects that resembles sensations of red. Scientific investigation might then lead one to see that our concept also permits: C2. The concept ‘red’ correctly applies to the property of objects that is the normal cause of their looking red (where ‘the normal cause of their looking red’ functions merely as a reference-fixer). Further, one might imagine it turning out that there are many different sorts of causes of looking red, and there are no salient similarities among the intrinsic properties of these causes. Reliance on C2 might then predict that there is no such property as 12 Alternatively, such conditionals might be formulated nonmetacognitively; for example: if a world is actual in which the watery stuff in the environment has a unique sort of composition, then water is a unique compositional stuff. 13 On a reading inspired by Stalnaker, such a conjunction of conditionals would not amount to semantic analysis, or an analysis of the meaning of a term or concept, but rather a metasemantics: “an account of what the facts are in virtue of which expressions have the semantic values they have.” See Robert Stalnaker, “On Considering a Possible World as Actual,” Proceedings of the Aristotelian Society 75, no. 1 (2001), pp. 141–56; reprinted in his Ways a World Might Be: Metaphysical and AntiMetaphysical Essays (New York: Oxford University Press, 2004), pp. 188–200. But the upshot relevant to the present discussion is the same. On either account, what a term like ‘water’ refers to or what a concept such as ‘water’ stands for depends on which of these various conditionals actually applies, and this depends on how the world turns out to be.

34 Consciousness and the Prospects of Physicalism

redness, or at least that redness is not instantiated—if wildly disjunctive candidates are ruled out, for example. Perhaps the same would then need to be said about any proposed response-dependent property whose categorical basis was wildly disjunctive, and this would not be credible. We would plausibly conclude that our concept also allows: C3. The concept ‘red’ correctly applies to whatever properties cause (or could cause) instances of looking red. Or suppose that because Berkeley’s theory turned out to be true, God is the normal cause of objects’ looking red, and we knew it. Then we would not say that the concept ‘red’ correctly applies to God, but more likely that it correctly applies to an appearance property. Thus while conceptual analysis of color concepts might initially seem to reveal something like C1, a more thorough analysis would yield a complex conjunction of conditionals. There is a moral here for the analysis of phenomenal concepts. One might at first be convinced that conceptual analysis reveals that phenomenal concepts correctly apply to properties that resemble our introspective representations of them, so that P1. The concept ‘phenomenal red’ correctly applies to the property that resembles the introspective representation of phenomenal redness. But by analogy, consider an Aristotelian who holds that conceptual analysis reveals that C1. The concept ‘red’ correctly applies to the property of objects that resembles sensations of red. Imagine that she is confronted with a scientific demonstration that physical objects have no such properties. She might conclude that redness is not instantiated in the physical world and that the concept ‘red’ does not correctly apply to anything in the physical world, as Galileo did.14 However, many of us hold that a response of this sort is mistaken, and that a distinct conjunct of our concept ‘red,’ such as the conditional that features C2 as its consequent, would then actually apply. Or even if we are initially strongly disposed to the reaction Galileo had, after overcoming the initial shock resulting from the displacement of our instinctive belief, we might become habituated to a different conception of color. Similarly, investigation and reflection might indicate that the conditional of a phenomenal concept that actually applies is not one that has

14 Galileo Galilei, The Assayer, in Discoveries and Opinions of Galileo, tr. Stillman Drake (New York: Doubleday Anchor, 1957), pp. 217–80, at pp. 274–77: “Hence I think that tastes, odors, colors, and so on are nothing but mere names so far as the object in which we place them is concerned, and that they reside only in the consciousness” (p. 274).

35 Phenomenal Concepts and the Explanatory Gap

it standing for a property that is accurately represented introspectively. Suppose it turns out that no instantiated properties are accurately represented in this way. One might then conclude that phenomenal concepts fail to apply to any instantiated properties. But as in the case of color concepts, a radical conclusion of this sort is not clearly forced. It could well be that there are alternative options reflected in other conditionals in the analysis of phenomenal concepts. One might object that while it is possible to devise phenomenal concepts whose analysis is complex in this way, still our ordinary phenomenal concepts are simple in the sense of being nonconjunctive, and their analyses yield only specifications like P1. Thus whether there are phenomenal properties on the ordinary understanding depends on whether there are properties that fit P1. In response, we might envision Aristotelians about color having made an analogous claim: “One might devise color concepts whose analysis is a complex conjunction of conditionals, but ordinary color concepts are simple, and their analyses yield only specifications like C1.” However, as history has shown, the initial attractiveness of C1 as an exhaustive characterization of the concept ‘red’ is defeasible. The case of phenomenal redness, I suggest, is parallel; the initial attractiveness of P1 as an exhaustive characterization of the concept ‘phenomenal red’ is also defeasible. Instead, it is an open possibility that the analysis of the concept of phenomenal redness reveals a complex conjunction of conditionals, and that the conditional that actually applies renders true a different sort of characterization, such as: P2. The concept ‘phenomenal red’ correctly applies to the property that is the normal cause of introspective representations of phenomenal redness (where ‘the normal cause of introspective representations of phenomenal redness’ functions merely as a reference-fixer), or else P3. The concept ‘phenomenal red’ correctly applies to whatever properties cause (or could cause) instances of the introspective representation of phenomenal redness. Thus even if on an ordinary understanding phenomenal concepts are exhaustively characterized by specifications like P1, it might be that the dispensability of this understanding is certified by the correct analysis of these concepts. Notice that even if introspective phenomenal modes of presentation represent phenomenal properties as having qualitative natures that they actually lack, given P2 the concept of phenomenal redness need not misrepresent phenomenal redness. By analogy, even if red sensations represent red as having a qualitative nature that it actually lacks, given C2 our concept of red need not misrepresent redness. In each case, the concept can represent accurately by picking out the straightforwardly physical normal cause of the relevant mental state. In the view I will now set out, however, phenomenal property representation has a dual content, so that if the qualitative

36 Consciousness and the Prospects of Physicalism

inaccuracy hypothesis is true, phenomenal content and concepts misrepresent in one respect and correctly represent in another.

PERFECT AND ORDINARY PHENOMENAL CONTENT The analogy with secondary quality representation can be developed further to strengthen the challenge from the qualitative inaccuracy hypothesis to the intuition that Mary learns something new when she sees the tomato. Consider Chalmers’s view of the content of visual color representation. He first argues that the account of such content that is most adequate to the phenomenology of color perception is a kind of primitivism: The view of content that most directly mirrors the phenomenology of color experience is primitivism. Phenomenologically, it seems to us as if visual experience presents simple intrinsic qualities of objects in the world, spread out over the surface of the object. When I have a phenomenally red experience of an object, the object seems to be simply, primitively, red. The apparent redness does not seem to be a microphysical property, or a mental property, or a disposition, or an unspecified property that plays an appropriate causal role. Rather it seems to be a simple qualitative property, with a distinctive sensuous nature. We might call this property perfect redness: the sort of property that may have been instantiated in Eden.15 Our experience of red is not as an unspecified property. Instead, its qualitative nature is wholly revealed in our sensory experience of it. In addition, to sensory experience red appears simple in the sense of not seeming to be constituted by a number of more fundamental properties, for example, as not having a complex internal causal or dispositional structure.16 The content of a phenomenal color representation associated with primitiveness Chalmers calls its Edenic content.17 However, science and philosophical reflection provide us with good reason to believe that there is no instantiated property to which this Edenic content correctly applies; there are no instantiated primitive color properties. But Chalmers agrees that we should not conclude that there are no colors. There is a veridical content of phenomenal color representation that well enough matches its perfect content, which he calls its ordinary content. Edenic content functions as a kind of regulative ideal in

15 David Chalmers, “Perception and the Fall from Eden,” p. 66. 16 In the Garden of Eden, Chalmers specifies, “We had unmediated contact with the world. We were directly acquainted with objects in the world and with their properties. Objects were simply presented to us without causal mediation, and properties were revealed to us in their true intrinsic glory” (“Perception and the Fall from Eden,” p. 48). Here Chalmers is specifying an ideal; he does not deny that primitivism about visual color representation can accommodate a causal theory of such representation, as in the Aristotelian view. 17 David Chalmers, “Perception and the Fall from Eden,” pp. 69–71.

37 Phenomenal Concepts and the Explanatory Gap

determining the ordinary content of our color experiences—it is the standard that matching ordinary content must most closely approximate—but its being merely a regulative ideal allows for matching ordinary content that is veridical.18 Note that this account seems to commit Chalmers to qualitative inaccuracy in visual color perception. Such perception represents physical objects as primitively colored, but they are not primitively colored, while at the same time they are colored. Visual color perception sorts colors quite correctly but represents something else about them inaccurately. The best candidate for what is inaccurately represented is the color’s qualitative nature. A story parallel to Chalmers’s account of the content of color representation can be provided for our introspective representations of phenomenal properties. When I have an introspective representation of phenomenal redness, what I apprehend seems to be a simple qualitative property, with a distinctive sensuous nature that is wholly revealed in an introspective experience of it. This property might be primitive phenomenal redness, or perhaps just a property that has a qualitative nature accurately represented by introspection. We might call the content of introspective phenomenal redness associated with such a property its Edenic content. However, it may be that these representations are also qualitatively inaccurate in the sense that their Edenic content correctly applies to no instantiated properties. Still, there might be an ordinary content of these representations that matches their Edenic content closely enough, with the consequence that there are instantiated phenomenal properties to which this matching content correctly applies. These properties might be physical properties, such as the physical property that is the normal cause of introspective representations of phenomenal redness. Consider two possible proposals for the ordinary content of representations of phenomenal redness (derived from P2 and P3, defined previously): OC-P2: an ordinary content that correctly applies to the property that is the normal cause of introspective representations of phenomenal redness (where ‘the normal cause of introspective representations of phenomenal redness’ functions merely as a reference-fixer). OC-P3: an ordinary content that correctly applies to whatever properties cause (or could cause) instances of the introspective representation of phenomenal redness. On OC-P2, it is nomologically possible for a state to be introspected as phenomenal redness while phenomenal redness is not then instantiated, since a property that on some occasion causes the introspective representation of phenomenal redness might not be the property that is its normal cause. This is perhaps an unintuitive result. A contrasting characteristic of OC-P3 that accordingly counts in favor of its being the closest match to the regulative ideal is that it renders it impossible for a state to be 18 David Chalmers, “Perception and the Fall from Eden,” pp. 69–84.

38 Consciousness and the Prospects of Physicalism

introspected as phenomenal redness while phenomenal redness is not instantiated. Then if a state seems conscious in the what-it-is-like-to-see-red way, it will be conscious in this way. As applied to pain, for example, this might recommend OC-P3, since it is at least initially strongly unintuitive that a state be introspectively represented as pain and not be pain. On the other hand, it may count in favor of OC-P2 that it would allow our classification of phenomenal properties to cut nature at its causal joints after the manner of Kripkean natural kind terms or concepts, while OC-P3 is not designed to do so. I favor OC-P2. Among other virtues, this allows for pain, and for phenomenal properties generally, not to be functional in the sense of properties whose essences are exclusively causal relations, and not to be essentially response-dependent properties. In the view that I will develop in chapters 7 and 8, phenomenal properties, and mental properties generally, are not functional in this sense but instead compositional, that is, properties things have solely by virtue of intrinsic features of their parts, either proper or improper, and relations these parts have to one another. As for the unintuitive consequence of OC-P2, recall the example of the fraternity pledge discussed in the previous chapter. Someone says he is going to cut his throat with a razor but instead administers the icicle. Here it might well be that the sensation of cold is mistaken for pain. A state would then be introspectively represented as pain but not be pain.19 When Mary leaves her room and sees the red tomato, we imagine her now having a belief of the form: (A) Seeing red has R. On an Edenic content interpretation for the concept ‘R,’ and supposing (A) is true, Mary would learn something new when sees the red tomato. But on the open possibility we are considering, given an Edenic content interpretation this belief is in fact false. So then Mary would not learn something new. On several ordinary content interpretations (for example OC-P2 and OC-P3), there would be no less reason to believe that the belief of form (A) is derivable from what pre-emergence Mary knows than there is to think that ‘rivers and lakes contain water’ or ‘some physical objects are red’ is so derivable. Then again, Mary would not learn anything new when she sees the tomato, but for a different reason: she already had the belief when she was in the room. Thus, on our open possibility, for both Edenic and ordinary content interpretations of (A), Mary does not learn anything new when she leaves the room, and the knowledge argument faces a challenge. To this overall picture, one might press the objection that the strength of the intuition that Mary learns something about phenomenal redness upon having an experience of the tomato indicates that it’s a conceptual truth after all that phenomenal redness is a property that resembles our introspective representation of it, whereupon (something like)

19 Thanks to Kati Balog for raising the objection that occasioned this account.

39 Phenomenal Concepts and the Explanatory Gap

P1. The concept ‘phenomenal red’ correctly applies to the property that resembles the introspective representation of phenomenal redness. constitutes an exhaustive conceptual analysis. One might add that this claim is also supported by the strength of intuition that zombies are conceivable, to be examined in the next chapter. On the view that Jackson and Chalmers develop, the objection predicts that if we considered as actual a scenario in which we introspect phenomenal redness as we in fact do but there is no instantiated property that resembles this introspective representation, we would be strongly disposed to judge that phenomenal redness is not instantiated. Let me respond with an additional consideration, developed by Janice Dowell, which to my mind has especially significant force.20 On the Jackson-Chalmers view, someone who possesses the concept ‘red’ has an implicit knowledge of the various conditionals that compose it. This implicit knowledge can become explicit when we consider appropriate scenarios as actual and then ask what the concept applies to in those scenarios, if anything. However, as Dowell contends, resistance to discerning the correct application of a concept in such a situation can be provided by a strongly held belief about how things actually are. In one of her examples, at a certain point in the history of chemistry experts believed that, as a matter of conceptual fact, an acid must contain oxygen. It was actual empirical and scientific examination of the nature of HCl that occasioned the widespread rejection of this view and allowed the conceptual possibility that an acid not contain oxygen to become explicit.21 For another possible instance, prior to the twentieth century we might have thought that as a matter of conceptual fact our space is Euclidean. It was scientific reflection on the empirical evidence for a Riemannian account that allowed us to become explicitly aware of the conceptual possibility that our space be non-Euclidean. Returning to the secondary quality analogy, one might speculate that a process of this sort occurred for Descartes, who in the Meditations of 1641 seriously considered eliminativism about such qualities, claiming not to know whether our ideas of them are “of things or of nonthings,” while in the Principles of Philosophy of 1644 he confidently affirms that secondary qualities are “simply dispositions in those objects which make them able to set up various kinds of motions in our nerves which are required to produce all the various sensations in our soul.”22 Not implausibly, scientifically informed reflection

20 Janice Dowell, “Empirical Metaphysics: The Role of Intuitions about Possible Cases in Philosophy,” Philosophical Studies 140 (2008), pp. 19–46. 21 Janice Dowell, “Empirical Metaphysics: The Role of Intuitions about Possible Cases in Philosophy.” Dowell notes that the example comes from Jessica Wilson. Dowell develops an interesting Kripkean line on the conceptual analysis relevant to the knowledge and conceivability arguments in “A Priori Entailment and Conceptual Analysis: Making Room Type-C Physicalism,” Australasian Journal of Philosophy 86 (2008), pp. 93–111. 22 René Descartes, Meditations on First Philosophy, in The Philosophical Writings of Descartes, vol. 2, p. 30 (AT VII 44); Principles of Philosophy, Part IV, 198, in The Philosophical Writings of Descartes, vol. 1, p. 285 (AT VIIIA 323).

40 Consciousness and the Prospects of Physicalism

occasioned his realization that primitivism about secondary qualities is not demanded by our secondary quality concepts. Even if one is now resistant to its being open that our concept of phenomenal redness has an ordinary content that correctly applies to the property that is the normal cause of introspective representations of phenomenal redness, Dowell points out that it is an open possibility that our actual future reactions to future empirical discoveries will show us that the analysis that best fits the pattern in our application of our concept of phenomenal redness is such a content. We would then come to realize that our phenomenal concepts permit contents like OC-P2.23 To my mind, this reflection strongly indicates that it is a serious open possibility that our phenomenal concepts allow for such contents and also that the qualitative inaccuracy hypothesis is true, given that it would be if content like OC-P2 actually applies.

A TYPE OF PROJECTIVISM? According to projectivist views, our minds project onto things of certain kinds properties that they do not in fact have.24 On one variety of projectivism about color, we perceptually represent external objects as colored only because our minds project the primitive color (or primitive phenomenal color) of our experiences onto those objects. Such views about color have their adherents, but they are controversial. Sydney Shoemaker wonders whether the qualitative inaccuracy hypothesis is committed to an implausible type of projectivism.25 Here is his definition of the general idea: Projectivism: Our experience represents external things as having properties that they do not in fact have. In his analysis, two varieties are salient:

23 Janice Dowell, in correspondence, and “Empirical Metaphysics: The Role of Intuitions about Possible Cases in Philosophy.” 24 The most famous characterization of projectivism is due to Hume: Thus the distinct boundaries and offices of reason and of taste are easily ascertained. The former conveys the knowledge of truth and falsehood: the latter gives the sentiment of beauty and deformity, vice and virtue. The one discovers objects as they really stand in nature, without addition or diminution: the other has a productive faculty, and gilding or staining all natural objects with the colours, borrowed from internal sentiment, raises in a manner a new creation. (David Hume, Enquiry Concerning the Principles of Morals [Oxford: Oxford University Press, 1975], Appendix 1, p. 294) 25 Sydney Shoemaker, in conversation and in “Qualities and Qualia: What’s in the Mind?” reprinted in his The First-Person Perspective and Other Essays (Cambridge: Cambridge University Press, 1996), pp. 97–120.

41 Phenomenal Concepts and the Explanatory Gap

Literal projectivism: Our experience represents external things as having properties that they do not have but are had only by our experiences. Figurative projectivism: Our experience represents external things as having properties that they do not have and in fact do not belong to anything. On the open possibility I’ve set out, just as our visual experience represents the colors of external things as having qualitative natures that they do not possess and, in fact, do not belong to anything, so similarly, introspection represents phenomenal properties as having qualitative natures that they do not have and are not instantiated. This might be interpreted as a variety of figurative projectivism that applies to both colors of external objects and properties of phenomenal states. A first concern of Shoemaker’s is this: “As for literal projectivism, I cannot myself make any sense of the idea that any property I perceive as belonging to the surface of the tomato, when I perceive its color, is in fact a property of the experience itself.”26 This seems right, but a toned-down version of literal projectivism can survive this criticism. When I think of the raspberry as having a certain taste, I might be thinking that the raspberry itself has a property whose qualitative nature resembles the qualitative nature of the taste property of my experience. True, when I think of the raspberry in this way, I am conceiving the property of the raspberry as an external physical property, while I am thinking of the property of the experience as a mental property. So in this respect, Shoemaker’s claim would be correct. But this is compatible with our representing these distinct sorts of properties as having a similar qualitative nature; such properties would in this way be analogues of one another. Thus one might imagine a kind of literal projectivism in which analogues of properties that introspection represents phenomenal states as having are perceptually represented as belonging to external objects. In accord with this suggestion, one possibility is that we perform mental acts that project onto external things analogues of properties with qualitative natures that we represent phenomenal properties as having, although this qualitative nature is nowhere instantiated. I tentatively favor this figurative projectivist version of the open possibility. But another account is motivated by a response of Shoemaker’s to a projectivist threat to his own position: I mentioned two features of my view that might seem to commit me to projectivism. One was that my view says that in some sense we project similarities and differences between experiences onto things in the world. This might seem to imply the literal projectivist view that our experiences project onto objects features of them, qualia, in virtue of which these phenomenal similarities and differences hold. But all that it need be taken to imply is that what similarity and difference relations we perceive in the world is a function of what relations of phenomenal similarity and difference relations hold among our experiences, and that does not imply that we project the properties of the experiences.27 26 Sydney Shoemaker, “Qualities and Qualia: What’s in the Mind?” p. 117. 27 Sydney Shoemaker, “Qualities and Qualia: What’s in the Mind?” p. 118.

42 Consciousness and the Prospects of Physicalism

On the alternative view inspired by this response, our experience does not actually project properties from one level to another. Rather, there are similarities between the visual color representation system and the introspective phenomenal color representation system that explain why it is that these two systems (mis)represent entities as having properties with similar qualitative natures. Then we would not, for example, perform an act of projection whereby a phenomenal property is transformed and projected onto the tomato as a physical property. I mildly prefer the first view, since my sense is that it provides a better explanation of the fact that the phenomenal property and the color are represented as having a common qualitative nature, but I think this second view is a contender as well. Shoemaker raises a further worry: As for figurative projectivism, it is a mystery, to say the least, how the content of our experience can include reference to properties whose actual instantiation we have never experienced or had any other epistemic access to—properties we know neither “by acquaintance” nor “by description,” unless we have some sort of nonsensory acquaintance with a Platonic realm of uninstantiated properties.28 I prefer to resist the epistemological picture that Shoemaker assumes. We are familiar with the qualitative nature that we misrepresent phenomenal redness as having just because of the way our introspective system represents phenomenally red states, and thus there is no need for this qualitative nature to be instantiated for us to become familiar with it. Then, perhaps we project this qualitative nature as we represent it onto external objects, whereupon we are familiar with the qualitative nature we misrepresent redness as having just because of the way our introspective system misrepresents phenomenally red states. Or else we become familiar with the qualitative nature we misrepresent redness as having just because of the way our visual system represents red things. It is ways things are represented that would, fundamentally, provide us with access to this qualitative nature, even if it turns out that they misrepresent things, with the consequence that there isn’t anything that has this nature. A final issue raised by Shoemaker’s challenge is if the qualitative nature we represent phenomenal redness as having is nowhere instantiated, we will have no account of how our representations of phenomenal redness acquire content at all.29 However, as we saw in the previous section, just as our representations of redness can have an ordinary content, even assuming that our visual color sensations misrepresent its qualitative nature, so our representations of phenomenal redness can have an ordinary content, even if we introspectively misrepresent its qualitative nature. The content of our introspective representation of phenomenal redness might be one that correctly applies to the property that is the normal cause of introspective representations of

28 Sydney Shoemaker, “Qualities and Qualia: What’s in the Mind?” p. 117. 29 Thanks to Colin McLear for discussion of this point.

43 Phenomenal Concepts and the Explanatory Gap

phenomenal redness, and for this to be so, by exact analogy with color representation, it is not required that introspection accurately represent the qualitative nature of phenomenal redness.

THE EXPLANATORY GAP AND ELIMINATIVISM Chalmers contends that several commentators who have attempted to undermine the knowledge argument (and the zombie argument) by the old fact–new guise response have failed to show how it might be that the distinct modes of presentation, physical and phenomenal, might represent the same thing. This issue is especially pressing for Loar, who claims that physicalism is true, while phenomenal concepts express the properties they refer to, which commits him to the claim that introspective phenomenal modes of presentation accurately represent the qualitative natures of the properties they pick out, since these modes of presentation just are the properties they pick out. On Chalmers’s reading, Loar in fact maintains “that phenomenal and physical concepts (i) are cognitively distinct, and (ii) both express the property they refer to.”30 Chalmers argues that if both (i) and (ii) are accepted, nothing can justify the claim that the phenomenal and physical concepts corefer. Perhaps it is actually impossible that both (i) and (ii) are each satisfied on the supposition that the modes of presentation corefer, for the reason that the phenomenal concept correctly applies to a primitive phenomenal property or a property that has a qualitative nature accurately represented by introspection, while this is not the property the physical concept expresses. However, if introspective representations of phenomenal properties are qualitatively inaccurate, Chalmers’s explanatory burden—which is part of the burden of the explanatory gap—can be discharged. Then it need no longer be explained how a qualitative nature that resembles the what-it-is-like-to-sense-red mode of presentation can correctly be attributed to a phenomenal property, while that property is at the same time physical, and a description of its real qualitative nature is derivable from ‘P,’ the complete physical truth. With regard to this explanatory gap, Robert Adams argues that materialism has no adequate response to the demand to explain why particular kinds of phenomenal properties are correlated with particular kinds of physical properties: For suppose a materialist claims that [physical property] R and the phenomenal appearance of red are one and the same property of brains, identified as R on the basis of its place in the physical system, and as the appearance of red on the basis of the way it seems to us when our brains have it. We can still ask why R seems to us the way it does, rather than the way Y (the physical brain state which “is” the appearance of yellow) does. This is quite recognizably our original question, and it remains unanswered.31 30 David Chalmers, “Materialism and the Metaphysics of Modality,” Philosophy and Phenomenological Research 59 (1999), pp. 473–96, at pp. 487–88. 31 Robert Adams, “Flavors, Colors, and God,” p. 259.

44 Consciousness and the Prospects of Physicalism

Adams’s demand is for a contrastive explanation: why does physical property R seem the way it does, and not the way physical property Y seems? Supposing that different brain states appear in different ways to introspection, the physicalist needs to explain why any one brain state appears to introspection in one way, and not in some other way, for example, in the way some other brain state appears to introspection. Adams believes that the physicalist has no adequate response to this demand. But we can reply: it is an open possibility that there is a discrepancy between the real nature of the property how R seems and the qualitative nature we introspectively represent this property as having. It is then an open possibility that how R seems is a straightforwardly physical property, call it RS, despite how we introspectively represent it. The same can be said of how Y seems—it might be a straightforwardly physical property— call it YS, despite the qualitative nature we introspect it as having. If this open possibility is actual, then the physicalist can meet the demand for contrastive explanation, which might then be formulated as: why does physical property R cause physical property RS and not physical property YS? We’re assuming that Mary, while she is in the room, has mastered purely physical explanations of this sort. So on the open possibility under consideration, the physicalist can meet Adams’s demand for an explanation.32 Adams further contends that a materialistic explanation of correlations between physical and phenomenal properties would require a materialism of a radical sort: One would have to adopt a very radical materialism indeed, rejecting not only the dualism of substances, but also the dualism of properties, and even the distinction of first- and third-person aspects or ways of identifying the sensible qualities, as well as the notion of a way in which conscious states seem to us when we are in them, as opposed to their place in the physical scheme of things. Thus one would have to eliminate phenomenal qualia, or reduce them in a most extreme way to physical qualities.33 However, the materialism suggested by our open possibility can retain the distinction between first- and third-person ways of identifying the sensible qualities, and also the notion of a way in which conscious states seem to us when we are in them. Despite the

32 As mentioned in note 5 of chapter 1, Stoljar’s version of the knowledge argument specifies that pre-emergence Mary possesses all the phenomenal concepts, while she nevertheless lacks knowledge of how correct applications of phenomenal concepts are correlated with physical states (“Physicalism and Phenomenal Concepts”). The antiphysicalist might then contend that while she is in the room, Mary would not be able to produce contrastive explanations of the sort Adams demands and that for this reason physicalism is false. But now we can see that on the open possibility under consideration, Mary would be able to produce these explanations. One might press on here by asking: why does RS appear to us as it does and not otherwise? Here again we can ascend a level and suggest the open possibility that our introspective representation of the appearance of RS—call it ARS—is inaccurate and that ARS is a physical property. Then, by virtue of having mastered all the physical explanations, Mary understands why RS causes ARS and not some other relevant alternative physical property. 33 Robert Adams, “Flavors, Colors, and God,” p. 259.

45 Phenomenal Concepts and the Explanatory Gap

discrepancy between the real qualitative nature of phenomenal properties and how they are introspectively represented, there is a first-person, introspective point of view on phenomenal properties, and a way they appear to us when we are in the states that have them. True, the real qualitative nature of those properties would be accessible from the third-person point of view, so the first-person perspective does not provide genuine information about the qualitative nature of those properties that is not accessible from the third-person perspective. But this is not enough to make the materialism in question a radical one, since a claim of this sort would be required for any materialism. At the same time, denying that a qualitative nature that resembles introspective phenomenal modes of presentation is correctly attributed to phenomenal properties might well not amount to eliminativism for phenomenal properties. One could, in agreement with Galileo, argue for eliminativism about temperature as a property of physical objects on the grounds that the temperature of physical objects does not resemble our sensory representation of it.34 An Edenic content of our temperature concept could even be defined that applies only to a property that resembles temperature sensations—and it might then be pointed out that there is no actually instantiated property to which this content correctly applies. But as history has shown, highly plausible noneliminativist options for temperature itself remain. On the open possibility that I have been discussing, noneliminativist options also remain for phenomenal properties. Something that many believe to exist would be eliminated: certain features that are accurately represented introspectively. Indeed, one might define a notion of the Edenic content of phenomenal concepts that would correctly apply only to such features, which would then correctly apply to no properties that are actually instantiated. But this is not to say that phenomenal properties would thereby be eliminated or that our phenomenal concepts fail to apply to anything real, for they might have an ordinary content that does.35 Even then, the Edenic content might still function as a regulative ideal, as on the model for color concepts Chalmers develops. Thus the advocate of the knowledge argument might claim against the qualitative inaccuracy hypothesis that it proposes eliminativism about phenomenal properties. However, it is no more eliminativist about phenomenal properties than most contemporary theories of secondary qualities are about color. In each case, primitivism about the properties at issue, for example, is denied, but denying primitivism does not amount to eliminativism about phenomenal properties, only about primitive versions of these properties. On the other hand, because they invoke primitivism about phenomenal properties, or else the weaker claim that they have a qualitative nature that we accurately represent introspectively, each of which could well be false, these antiphysicalist arguments are less forceful than one might initially believe. This is not to 34 Galileo Galilei, The Assayer, in Discoveries and Opinions of Galileo, pp. 274–77. 35 This position would thus count as a kind of real materialism on Galen Strawson’s characterization, that is, the sort that does not deny the existence of the experiential; see Galen Strawson, “Real Materialism,” in his Real Materialism and Other Essays (Oxford: Oxford University Press, 2008), pp. 19–51.

46 Consciousness and the Prospects of Physicalism

say that they lack force altogether, since these claims about phenomenal properties are attractive. But it is significant that their antiphysicalist force is limited by the degree to which they are plausible.

SUMMARY In these first two chapters, against the knowledge argument I have argued that it is an open possibility that introspective representations of phenomenal properties are in a sense qualitatively inaccurate. Specifically, we introspectively represent phenomenal properties as having a certain qualitative nature, but the attribution of this qualitative nature to these properties might be incorrect. As a result, it may be that the real nature of phenomenal properties is straightforwardly physical, and complete information about it is derivable from what Mary knows before emerging from the room, despite the appearance that she acquires information about the qualitative nature of phenomenal-redness when she leaves the room that was not available to her earlier. Moreover, there is a plausible view of the nature of phenomenal concepts that fits this conception. Kant might well have endorsed the open possibility of this kind of qualitative inaccuracy.36 By contrast, Descartes arguably maintained that qualitatively accurate and complete introspective representations of our mental states generally are available to us.37 Perhaps we should say that both the Cartesian and Kantian options, applied to phenomenal states, are live open possibilities; neither has been ruled out. But as long as the seriousness of the Kantian open possibility is not undermined, the knowledge argument against physicalism faces a significant challenge.

36 Immanuel Kant, Critique of Pure Reason, e.g., Bxxiv–xxvii. 37 René Descartes, Meditations on First Philosophy, in The Philosophical Writings of Descartes, vol. 2, pp. 16–23 (AT VII 23–34).

3 CONCEIVABILITY ARGUMENTS AND QUALITATIVE INACCURACY

Conceivability arguments against physicalism might also be challenged by the proposal that while phenomenal properties are introspectively represented as having natures of a particular qualitative sort, it is an open possibility that these qualitative natures are incorrectly attributed to them. Conceivability arguments, advanced by René Descartes and more recently by Saul Kripke, David Chalmers, and George Bealer, typically claim first that certain mental phenomena can be conceived without the physical entities at issue or that the relevant physical entities can be conceived without certain mental phenomena, derive from this that such scenarios are metaphysically possible, and conclude that physicalism is false.1 Sometimes the conceivability of such scenarios is replaced by their seeming possibility or by an intuition of their possibility. These arguments have had the lead role in the confrontation with physicalism since the early modern period. They have in addition raised important and interesting issues about the relation of conceivability to possibility and about modal epistemology more generally. I will focus on the zombie argument, the version of this type of challenge developed by Chalmers.2 In short, this argument hinges on the claim that it is conceivable, in an

1 René Descartes, Meditations on First Philosophy, in The Philosophical Writings of Descartes, vol. 2, p. 54 (AT VII 78); “Fourth Replies,” in The Philosophical Writings of Descartes, vol. 2, pp. 154–62 (AT VII 219–31); Saul Kripke, Naming and Necessity, pp. 144–53; George Bealer, “Modal Epistemology and the Rationalist Renaissance.” For expositions of Descartes’s argument, see Margaret Wilson, Descartes (London: Routledge, 1978); Stephen Yablo, “The Real Distinction between Mind and Body,” Canadian Journal of Philosophy 16 (1991), pp. 149–201; Marleen Rozemond, Descartes’s Dualism (Cambridge, MA: Harvard University Press, 1998); and Joseph Almog, What Am I? Descartes and the Mind-Body Problem (New York: Oxford University Press, 2002). 2 David Chalmers, The Conscious Mind, “Consciousness and Its Place in Nature,” and “The TwoDimensional Argument against Materialism.” 47

48 Consciousness and the Prospects of Physicalism

appropriately sophisticated way, that a world that is (nothing but) an exact physical duplicate of the actual world features no phenomenal consciousness —in other words, that a zombie world is conceivable. From this premise, the argument reasons to the conclusion that the complete physical truth does not necessitate the complete phenomenal truth, or even any phenomenal truth, and that therefore physicalism is false. The factor that gives rise to complex mechanics in the argument is that not all conceivable scenarios are metaphysically possible. Sometimes one can conceive a scenario only because one is deficient in reasoning, as when one conceives of a right triangle the square of whose hypotenuse is not equal to the sum of the squares of each of the two sides.3 Conceiving is then less than ideal. Or else, as Saul Kripke argued, sometimes what is really being conceived is misreported or mischaracterized, as when someone reports conceiving of water that is not H2O but is really conceiving of something that merely appears to be water or only has the evident causal role water has in our world. Chalmers’s strategy is to show that none of the available ways of explaining how deficiency in conceivability fails to establish metaphysical possibility is applicable to the conceivability of a physical duplicate of the actual world absent phenomenal consciousness, and that it therefore sustains valid reasoning to the conclusion that a zombie world is metaphysically possible. One response to this argument appeals to the same central claim as the foregoing reply to the knowledge argument. Physicalism requires that phenomenal truths be derivable from the complete physical truth about the actual world. Chalmers and Jackson argue that the relevant sort of derivability is a priori, and to my mind, the case they make is impressive and has not been successfully challenged.4 However, if

3 This is the example Antoine Arnauld directs at Descartes’s conceivability argument for dualism; “Fourth Objections,” in The Philosophical Writings of Descartes, vol. 2, p. 142 (AT VII 202). 4 David Chalmers and Frank Jackson, “Conceptual Analysis and Reductive Explanation.” Chalmers and Jackson argue that the nonfundamental truths are a derivable a priori, and not merely a posteriori, from a base consisting of the fundamental truths, which in their exposition are specified as the fundamental physical truth (P), the fundamental phenomenal truth (Q), the ‘that’s all’ provision (T), and the indexical truths (I). On the physicalist view, all the nonfundamental truths would be derivable from a base of PTI instead. As they explain, this claim for a priori derivability can be supported by arguing that all of the fundamental information relevant to deriving the nonfundamental truths can be driven into the base, whereupon no other information, let alone empirical information, will be required to perform the derivation. As a result, the derivation itself will be a priori. Their contentions are directed against the contrary view as advocated by Ned Block and Robert Stalnaker in “Conceptual Analysis, Dualism, and the Explanatory Gap.” My sense is that the most compelling objection that Block and Stalnaker provide for the a priori derivability thesis is expressed in this passage: This seems to be armchair reasoning, reflection that does not include any obvious reference to real experiments, so it is tempting to conclude that this reflection just unfolds our concepts in a totally a priori way. But what this conclusion misses is that our reasoning about the proper epistemic response in various counterfactual situations is informed not only by our concepts, but by implicit and explicit theories and general methodological principles that

49 Conceivability Arguments and Qualitative Inaccuracy

introspection did represent phenomenal states inaccurately, on analogy with our ordinary visual color representations on Locke’s conception, then the truth of physicalism would not require propositions about phenomenal properties as they are represented introspectively to be derivable a priori from the complete physical truth about the actual world. If the qualitative natures these phenomenal properties are introspectively represented as having are incorrectly attributed to them and are features these properties actually lack, then it might well be that all facts about the real natures of phenomenal states are derivable a priori from the complete physical truth after all. It may then be that a zombie scenario is not conceivable in the ideal case, since conceiving such a scenario involves an error that would be eliminated in ideal reflection. On this open possibility, when I attempt to conceive a zombie scenario, it turns out that what I end up conceiving is something identical to the actual world physically yet bereft of phenomenal properties as they would be represented introspectively. But when I come to understand that the qualitative natures that introspection represents phenomenal properties as having reflect merely how these properties appear and not how they really are, then I realize that my concepts of phenomenal properties do not pick out such properties as they appear introspectively, but rather properties all truths about which are derivable a priori from the complete physical truth. As a result, once I form a conception of a scenario that reflects the complete physical truth, it will be evident that the negation of any phenomenal truth is ruled out a priori, and the ideal rational conceivability of the zombie scenario is undercut. These thoughts yield a strategy for explaining why, in certain cases, the sort of conceivability on offer is not sufficient to secure metaphysical possibility. The possibility of qualitative inaccuracy of introspective representation, together with ignorance that such misrepresentation is occurring, is compatible with some sort of zombie-conceivability, but the fact that misrepresentation might be occurring precludes its being the ideal kind that issues in metaphysical possibility. Let me now develop this sketch in detail.

we have absorbed through our scientific culture—by everything that the “we” who are performing these thought experiments believe. What people should rationally say in response to various hypothesized discoveries will vary depending on their experience, commitments and epistemic priorities. (Ned Block and Robert Stalnaker, “Conceptual Analysis, Dualism, and the Explanatory Gap,” p. 43) Chalmers and Jackson’s response (“Conceptual Analysis and Reductive Explanation,” pp. 345– 50) is in essence that the relevant theories and methodological principles that we have absorbed through our scientific culture can themselves be derived from the fundamental truths alone. The general principle that the world is simple, for instance, is plausibly derivable from the fundamental truths alone, and thus is a priori derivable from these truths (p. 347). For another sustained critical treatment of the view Block and Stalnaker propose, see Brie Gertler, “Explanatory Reduction, Conceptual Analysis, and Conceivability Arguments about the Mind,” Noûs 36 (2002), pp. 22–49.

50 Consciousness and the Prospects of Physicalism

CHALMERS’S CONCEIVABILITY ARGUMENT: PRELIMINARIES Chalmers’s account formalizes several of the key notions at play in the zombie argument.5 To begin, he distinguishes several dimensions of conceivability: prima facie versus ideal, positive versus negative, and primary versus secondary. First, S is prima facie conceivable when S is conceivable on first appearances, and ideally conceivable when it is conceivable on ideal rational reflection. He proposes a way of spelling out ‘conceivable on ideal rational reflection’: “S is ideally conceivable when there is a possible subject for whom S is prima facie conceivable, with justification that is undefeatable by better reasoning.”6 Second, S is negatively conceivable when one cannot rule S out. Thus S is ideally negatively conceivable when one cannot prima facie rule S out with justification that is undefeatable by any possible better reasoning.7 Positive conceivability is difficult to characterize, but one paradigmatic variety involves the ability to form, by imagination, a mental picture of a scenario in which S is true.8 Third, we can understand primary and secondary conceivability as correlates of two distinct notions of possibility.9 In Chalmers’s view, one way to think about a possible world is as a kind of epistemic possibility, that is, as the way the world might actually turn out to be, given everything we can know a priori. When we do this, we consider that possible world as actual. So then, S is possible in this sense just in case S is true in some world considered as actual.10 For example, when one considers as actual the world in which all of the ‘water’ samples are not H2O but XYZ instead, then ‘water = XYZ’ is true in that world, and for this reason ‘water ≠ H2O’ is primarily possible. More formally, the term ‘considering as actual’ is linked to Chalmers’s proposal that to determine whether a statement S is primarily possible, one needs to evaluate indicative conditionals of the form ‘if possible world W is actual, then S.’ In this case, the indicative conditional ‘if the XYZ world is actual, then water ≠ H2O’ comes out true, and thus ‘water ≠ H2O’ is primarily possible.11

5 David Chalmers, “Consciousness and Its Place in Nature,” pp. 255–56. Chalmers’s formalization is controversial and has given rise to criticism; see, for example, George Bealer, “Modal Epistemology and the Rationalist Renaissance,” pp. 87–99; Stephen Yablo, “Shoulda, Woulda, Coulda.” 6 David Chalmers, “Does Conceivability Entail Possibility?” in Conceivability and Possibility, pp. 145–200, at p. 148. 7 David Chalmers, “Does Conceivability Entail Possibility?” p. 147. 8 David Chalmers, “Does Conceivability Entail Possibility?” pp. 148–49. 9 David Chalmers, “Does Conceivability Entail Possibility?” p. 157. 10 Chalmers’s scheme is similar to Jackson’s with a difference in terms: Jackson uses, for example, ‘A-intension’ for Chalmers’s ‘primary intension,’ and ‘C-intension’ for Chalmers’s ‘secondary intension’; Frank Jackson, From Metaphysics to Ethics, pp. 46–52. See also David Chalmers, “Foundations of Two-Dimensional Semantics,” in Two-Dimensional Semantics, ed. M. Garcia-Carpintero and J. Macia (Oxford: Oxford University Press, 2006), pp. 55–140. 11 Whether this last method for determining the sort of possibility at issue succeeds has been challenged by Yablo, but he advances an alternative conditional test in the spirit of Chalmers’s suggestion; see Stephen Yablo, “Shoulda, Woulda, Coulda,” pp. 452–54.

51 Conceivability Arguments and Qualitative Inaccuracy

An alternative criterion Stephen Yablo discusses is that S is possible in the sense at issue—which he calls conceptual possibility—just in case it could have turned out that S.12 Chalmers sometimes uses this formulation as well. A further test, proposed by Kripke, is to consider S in situations that are in some specified sense epistemically identical to ours and then to evaluate whether S is true in at least one such situation.13 Bealer formalizes this idea, and he suggests that the relevant sort of epistemic identity is evidential qualitative identity.14 Some find this kind of test more intuitively appealing than the indicative conditional proposal, but both Chalmers and Yablo reject it.15 The indicative conditional proposal, the ‘it could have turned out that . . .’ criterion, and the evidential qualitative identity test are all in a sense indirect, since none specifies explicitly and exactly which aspects of the concepts in S or the meanings of the terms in S are to be held fixed when evaluating S in alternative scenarios.16 On the other hand, one might instead consider world W as counterfactual. One then holds the nature of the actual world fixed and thinks of W as a way things might have been. If one thinks of the XYZ world in this way, then at the XYZ world ‘water = XYZ’ and ‘water ≠ H2O’ turn out false. In Chalmers’s framework, S is secondarily possible just in case S is true in some world considered as counterfactual. Hence, ‘water ≠ H2O’ is primarily possible, but not secondarily possible. Secondary possibility is what is more commonly known as metaphysical possibility. More formally, the term ‘considering as counterfactual’ derives from Chalmers’s proposal that to determine whether a statement S is possible in this sense, one needs to evaluate subjunctive conditionals of the

12 Stephen Yablo, “Shoulda, Woulda, Coulda,” pp. 452–54. 13 About the epistemic possibility that Hesperus is not Phosphorus, Kripke says: “And so it’s true that given the evidence that someone has antecedent to his empirical investigation, he can be placed in a sense in exactly the same situation, that is a qualitatively identical epistemic situation, and call two heavenly bodies ‘Hesperus’ and ‘Phosphorus,’ without their being identical. So in that sense it might have turned out either way” (Saul Kripke, Naming and Necessity, pp. 103–4); discussed by George Bealer, “Modal Epistemology and the Rationalist Renaissance,” p. 82. 14 Bealer designates the epistemic possibility at issue the ‘could’-of-qualitative-evidential-neutrality, which he defines as follows: “it is possible that p in the sense of qualitative-evidential-neutrality if and only if it is possible for there to be a population c with attitudes towards p and it is possible for there to be a population c’ with attitudes toward p whose epistemic situation is qualitatively identical to that of c such that the proposition which in c’ is the epistemic counterpart of p in c is true” (George Bealer, “Modal Epistemology and the Rationalist Renaissance,” p. 80). 15 Yablo argues that on the test suggested by Kripke, appearances will be necessary in the sense at issue, which is an unwelcome consequence: Mixed in with the semantical material we want to hold fixed will be nonsemantic circumstances that should be allowed to vary. One doesn’t want to hold fixed that there seems to be a lectern present, or there seeming to be a lectern present will be classified as conceptually necessary. That is clearly the wrong result. Appearances are conceptually contingent if anything is. (Stephen Yablo, “Shoulda, Woulda Coulda,” p. 446) Chalmers advances a related argument, “Does Conceivability Entail Possibility?” p. 169. 16 Stephen Yablo, “Shoulda, Woulda, Coulda,” pp. 441–54.

52 Consciousness and the Prospects of Physicalism

form ‘if possible world W were actual, then S.’ One might resist the idea that conditionals of this sort can have a key role in determining this kind of possibility, and claim instead that the right test involves holding an appropriate aspect of S fixed—perhaps what is said by S in the actual world, or the proposition expressed by S in the actual world—and then to evaluate whether there is some W in which this is true.17 These two notions of possibility can now be viewed as yielding characterizations of primary and secondary conceivability. Retaining Chalmers’s preferred ways of construing these notions, S is primarily conceivable just in case S can be conceived as true in some world considered as actual, and S is secondarily conceivable just in case S can be conceived as true in some world considered as counterfactual. An alternative but arguably equivalent approach crucially involves the epistemic notions of the a priori and the a posteriori. Chalmers specifies that whether a statement is primarily conceivable is solely a matter of a priori reflection: “primary conceivability is always an a priori matter. We consider specific ways the world might be, in such a way that the true character of the actual world is irrelevant.”18 How would ‘water ≠ H2O’ be conceived by way of a priori reflection? Inequality is plausibly fully transparent or understandable a priori. But not so for water; a priori reflection has no access to the ample empirical information that extends beyond what is involved in grasping the concept ‘water.’ Jackson and Chalmers suggest that in this case the information available a priori is restricted to certain facts about the causal role of water; for Jackson it features platitudes such as: water fills rivers and lakes, we drink water when we’re thirsty, and water puts out fires. Unavailable a priori is, for example, whether our water samples are all of a single chemical compound or whether they are rather all samples of H2O. As a consequence, a priori reflection does not rule out the truth of ‘water ≠ H2O,’ and for this reason this proposition is (negatively) primarily conceivable. More generally, a proposition or sentence S is (negatively) primarily conceivable when one cannot rule S out a priori. By contrast, secondary conceivability incorporates a posteriori investigation when appropriate. For ‘7 + 5 = 12’ there is no difference between primary and secondary conceiving, since (by hypothesis) only a priori and no a posteriori investigation is relevant to understanding the proposition. But matters are otherwise for ‘water ≠ H2O,’ since in this case a posteriori investigation reveals that water is identical to a specific chemical compound, H2O. The question now is whether holding these a posteriori results fixed, water can be (ideally) conceived as not being H2O. Here the answer is negative. Consequently, ‘water ≠ H2O’ is not (ideally) secondarily conceivable.

CHALMERS’S CONCEIVABILITY ARGUMENT FORMULATED AND EXPLAINED With these preliminaries in place, here is the zombie argument. Let ‘P’ be a statement that details the complete physical truth about the actual world, and ‘T’ a “that’s all” statement, so that ‘PT’ enumerates all the physical truths about the actual world with 17 Yablo discusses such a view in “Shoulda, Woulda, Coulda,” p. 446. 18 David Chalmers, “Does Conceivability Entail Possibility?” p. 158.

53 Conceivability Arguments and Qualitative Inaccuracy

the specification that there are no further truths (that is, other than those entailed by those physical truths).19 Further, let ‘Q’ be an arbitrary phenomenal truth. Russellian monism is the view that underlying the physical properties, which Chalmers suggests are all relational, are categorical and intrinsic phenomenal properties, or else categorical and intrinsic protophenomenal properties, so-called because while not phenomenal properties themselves, they nonetheless account for them. (This issue is discussed in chapters 5 and 6.) Then (1) ‘PT and ~ Q’ is ideally, positively, primarily conceivable. (2) If ‘PT and ~ Q’ is ideally, positively, primarily conceivable, then ‘PT and ~ Q’ is primarily possible. (3) If ‘PT and ~ Q’ is primarily possible, then ‘PT and ~ Q’ is secondarily possible or Russellian monism is true. (4) If ‘PT and ~ Q’ is secondarily possible, materialism is false. (5) Materialism is false or Russellian monism is true.20 (In the exposition that follows, I’ll often, like Chalmers, assume the ‘that’s all there is’ condition ‘T’ while not explicitly indicating it.) In premise (1) Chalmers specifies that ‘P and ~ Q’ is ideally, primarily, and not only negatively but also positively conceivable. If one is under the impression that ‘P and ~ Q’ can be ruled out a priori, one’s justification for believing this can be defeated by better reasoning, reasoning that in this context must be a priori, since it is primary conceivability that is at issue; moreover, one can form a positive conception of a scenario in which ‘P and ~ Q’ is true. Exactly what information is included in ‘P’? On a microphysical/macrophysical option, ‘P’ specifies complete information about every actual-world entity that is either microphysical or else wholly microphysically constituted, including every law that relates these entities.21 On a microphysical alternative, ‘P’ features only microphysical information, including information about microphysical laws. Since Chalmers employs this microphysical option in his discussions of the zombie argument, and I prefer it myself, I’ll assume it here as well. Rejecting premise (1) involves claiming that an ideal reasoner could derive a priori the arbitrarily selected actual phenomenal truth ‘Q’ from ‘P,’ supposing she has the

19 About ‘PT’ Chalmers and Jackson say: Intuitively, this statement says that our world contains what is implied by P, and only what is implied by P. More formally, we can say that world W1 outstrips world W2 if W1 contains a qualitative duplicate of W2 as a proper part and the reverse is not the case. Then a minimal P-world is a P-world that outstrips no other P-world. (David Chalmers and Frank Jackson, “Conceptual Analysis and Reductive Explanation,” p. 317) 20 David Chalmers, “Does Conceivability Entail Possibility?” p. 198, with equivalent terminology sometimes substituted; cf. “Consciousness and Its Place in Nature,” pp. 256–57. 21 See chapter 7 for my account of constitution.

54 Consciousness and the Prospects of Physicalism

minimal information required to ensure adequate possession of the phenomenal concepts involved in representing ‘Q.’ Chalmers is right to contend that there is a strong intuition that this claim is false. It is quite intuitive that given the complete microphysical conception of the actual world, together with the minimum phenomenal conceptual information, the falsity of ‘Q’ would not be ruled out, no matter how much better one’s reasoning about that conception became. However, this requires that one can rule out by a priori reflection with the (right sort of) concept of some phenomenal property that the qualitative nature this property is introspectively represented as having is not as it appears. It also requires that one can rule out by a priori reflection that this qualitative nature is completely neural. Here is where I think the opposition has an opening. Chalmers then defends Premise (2): if ‘P and ~ Q’ is ideally, positively, and primarily conceivable, then ‘P and ~ Q’ is primarily possible; that is, it is true in a world considered as actual—there is a metaphysically possible world W, such that if W is actual, then ‘P and ~ Q’ is true. Denying it amounts to endorsing what he calls a strong necessity: “a statement that is falsified by some positively conceivable situation (considered as actual), but which is nevertheless true in all possible worlds (considered as actual).”22 Chalmers contends that it is advantageous to preserve the entailment from ideal, positive, primary conceivability to primary possibility, for on the hypothesis that there are instances where this entailment fails, there would in such cases be no explanation as to why primary conceivability does not entail primary possibility.23 If a statement were ideally, positively, primarily conceivably false and yet true in all worlds considered as actual, then there would be no resources for explaining the resulting mismatch between conceivability and possibility. Chalmers argues in addition that there are no convincing counterexamples to the thesis that ideal, positive, primary conceivability entails primary possibility, that is, that there are no credible examples of strong necessities.24 An equally important step in this argument is: Premise (3): If ‘PT and ~ Q’ is primarily possible, then ‘PT and ~ Q’ is secondarily possible or Russellian monism is true.

22 David Chalmers, “Does Conceivability Entail Possibility?” p. 189; cf. “Materialism and the Metaphysics of Modality,” p. 480. 23 Bealer argues that what is represented as a claim that p is conceivable is better thought of as a rational intuition that p is possible. In his view, such intuitions are often fallible, and when they are, what is intuited to be possible is not really possible. On this analysis, the connection to possibility is more evident than it is in Chalmers’s formulations; see “Modal Epistemology and the Rationalist Renaissance,” pp. 73–77. Chalmers prefers to avoid this sort of move, since he thinks it would risk trivializing the connection to possibility in the ideal case; see “Does Conceivability Entail Possibility?” p. 156. 24 David Chalmers, “Materialism and the Metaphysics of Modality,” pp. 480–91; “Does Conceivability Entail Possibility?” pp. 189–94.

55 Conceivability Arguments and Qualitative Inaccuracy

For Chalmers, this is equivalent to: If ‘P and ~ Q’ is true in a world considered as actual, then either ‘P and ~ Q’ is true in a world considered as counterfactual, or Russellian monism is true. To understand this premise, it helps to note, as Chalmers indicates, that his zombie argument formalizes certain key aspects of Kripke’s antimaterialist argument in Naming and Necessity.25 Kripke argues that identity claims involving natural kind terms such as ‘water = H2O’ are necessarily true if true, with the consequence that showing that such a proposition is contingent is sufficient to show that it is false. But it appears conceivable, or seems possible, that water not be H2O, which suggests that ‘water = H2O’ is indeed contingent and therefore false. Yet on Kripke’s account, this reflection does not present a successful challenge to the claim that water = H2O, for the reason that in this case the proposition’s apparent contingency can be explained away. Here what one is really conceiving is a liquid that is not H2O, whose appearance is like that of water, or the everyday qualitative evidence for whose nature is the same as our everyday qualitative evidence for the nature of water, and this is consistent with water nevertheless being identical to H2O. Within Chalmers’s framework, the failure of this sort of challenge to the claim that water = H2O is accounted for by the fact that the statement ‘water ≠ H2O’ is ideally primarily conceivable and primarily possible but not ideally secondarily conceivable or secondarily possible. ‘Water ≠ H2O’ is conceivable as true, and is in fact true, in a possible world considered as actual—in, for example, the XYZ world. Yet at no possible world considered as counterfactual is this statement true; there is no possible world in which it is compatible with both the a priori truths and the relevant a posteriori facts. Although there is a world considered as actual in which ‘water = H2O’ is false—and this fact ultimately explains our sense that this statement is contingent—there is no world considered as counterfactual in which it is false, and this in the last analysis explains (put in Kripke’s terms) why the contingency is merely apparent.26 Kripke also argued that, by contrast, the apparent contingency of ‘pain = C-fiber firing’ cannot similarly be explained away, for the reason that any state that seems to be pain is in fact pain. Making an analogous point about consciousness in general, Chalmers argues that if ‘there is consciousness’ is true in W considered as actual, then in W considered as counterfactual ‘there is consciousness’ is true, and vice versa; if ‘there is consciousness’ is true in W considered as actual, then “it contains something that at

25 Saul Kripke, Naming and Necessity, pp. 144–53. 26 Here Chalmers’s notions of primary and secondary intension come into play: the primary intension of S assigns truth values to S in possible worlds considered as actual, while the secondary intension assigns a truth value to S in possible worlds considered as counterfactual. It is the fact that the primary and secondary intensions of ‘water = H2O’ come apart that explains its apparent contingency; see David Chalmers, “Does Conceivability Entail Possibility?” p. 163.

56 Consciousness and the Prospects of Physicalism

least feels conscious, and if something feels conscious, then it is conscious.”27 However, in the last analysis, Chalmers does not want to rest his case solely on the relationship between how things seem or feel and how things are. In his discussion of what he calls pure phenomenal concepts, Chalmers makes a claim crucial to the way he prefers to develop the zombie argument. A pure phenomenal concept “characterizes the phenomenal quality as the phenomenal quality it is.”28 This sort of concept is epistemically rigid: It picks out the same referent in every epistemically possible scenario (considered as actual). By contrast, ordinary rigid concepts are merely subjunctively rigid, picking out the same referent in every possible scenario (considered as counterfactual).29 Here one might define a similar notion more closely allied with those we’ve examined so far: Concept C is primarily rigid just in case C has the same referent in every possible world considered as actual.30 I will frame the discussion in terms of primary rather than epistemic rigidity. Chalmers maintains that unlike the concept ‘water,’ a pure phenomenal concept refers to the same thing, in this case to the same phenomenal property, in every scenario that is not ruled out a priori. By contrast, the concept ‘water’ is not primarily rigid, since there are applications of the term ‘water’ in scenarios not ruled out a priori in which it does not refer to H2O—for instance, in the XYZ world.31 Hence, there will be worlds considered as actual in which the word ‘water’ fails to refer to H2O. By extension, statements and propositions can be primarily rigid by having the same truth value in every world considered as actual. It would follow from phenomenal concepts’ being primarily rigid that they are also secondarily rigid; that is, they pick out the same property in every world considered as counterfactual.

27 David Chalmers, “The Content and Epistemology of Phenomenal Belief,” in Consciousness: New Philosophical Perspective, ed. Q. Smith and A. Jokic (Oxford: Oxford University Press, 2003). 28 By Chalmers’s characterization, a pure phenomenal concept is difficult to express in language. The term ‘phenomenal red’ might seem to express the relational concept ‘the phenomenal quality typically caused in normal subjects within my community by paradigmatic red things,’ but a pure phenomenal concept is not relational. The term ‘phenomenal red’ might appear to express the demonstrative concept ‘this property of my experience’ in appropriate circumstances, but a pure phenomenal concept is not demonstrative; see David Chalmers, “The Content and Epistemology of Phenomenal Belief.” 29 “The Content and Epistemology of Phenomenal Belief,” p. 18. 30 For a statement to be primarily necessary is for it to be a priori true; for a concept to be primarily rigid is for its reference to be a priori fixed. 31 In correspondence, Chalmers notes the wrinkle that there might be some truths about what a primarily rigid concept refers to that can’t be known nonempirically; for example, it might be empirical that such a concept refers to Y, for although it’s nonempirical that it refers to X, it’s empirical that X is Y. Conversely, although it’s empirical that the concept refers to Y, it is still primarily rigid.

57 Conceivability Arguments and Qualitative Inaccuracy

Setting aside Russellian monism for now, Premise (3) states that if ‘P and ~ Q’ is primarily possible, it is also secondarily possible. Crucial to the argument for this premise is the claim that ‘Q’—and for now we are assuming ‘P’—are primarily rigid. Then, for any world considered as actual in which ‘P and ~ Q’ is true, it will also be true in that world considered as counterfactual. So if ‘P and ~ Q’ is in fact true in a world considered as actual, it will also be true in that world considered as counterfactual, which is to say that it would then be secondarily possible.32 Premise (3) would thus be true as a matter of the logic of primary rigidity. If (3) is in fact false, it would have to be because primary rigidity is incorrectly attributed to some relevant concept or statement.

A CHALLENGE TO THE CONCEIVABILITY OF ZOMBIES One can now see that the force of Chalmers’s conceivability argument is considerable. To reject Premise (1) is to claim that by ideal a priori reasoning one could derive the selected phenomenal truth ‘Q’ from ‘P’ (and ‘T’), and against this, there is a strong intuition. Denying Premise (2) requires claiming that even if the positive conception of ‘P and ~ Q’ is not ruled out a priori, so that the justification for this cannot be defeated by better a priori reasoning, still there is no possible world considered as actual in which ‘P and ~ Q’ is true. Given the persuasive force of Chalmers’s extensive discussion of this issue, the burden of proof is on the side of those who want to deny this connection, and it is not easy to see how this might be done.33 Setting aside Russellian monism, rejecting Premise (3) requires denying that pure phenomenal concepts are primarily rigid. However, the qualitative inaccuracy hypothesis indicates that Premise (1) (really ‘PT and ~ Q’) is vulnerable, and this is the idea I will now develop. In addition, I should note that in my view the reasons for rejecting Premise (1) also count against Premise (3). But as Chalmers correctly points out, there is a resilient alternative premise that can substitute for Premise (3), which sidesteps the threat to Premise (1). I discuss this issue in the appendix to this chapter.34 To begin, it is important to distinguish the thesis that phenomenal concepts are primarily rigid from the claim that these concepts accurately represent the qualitative nature of phenomenal properties. Chalmers may be suggesting this second idea when he contends that “one might say very loosely that [for pure phenomenal concepts] the referent of the concept is somehow present inside the concept’s sense, in a way much stronger than in the usual cases of ‘direct reference,’” and “in the phenomenal case, the epistemic content itself seems to be constituted by the referent.”35 As I understand it, Chalmers’s position is that the sense of a pure phenomenal concept contains the qualitative nature of the phenomenal property as it is introspectively represented. For 32 David Chalmers, “Does Conceivability Entail Possibility?” p. 197. 33 David Chalmers, “Materialism and the Metaphysics of Modality,” pp. 480–91; “Does Conceivability Entail Possibility?” pp. 189–94. 34 David Chalmers, “Does Conceivability Entail Possibility?” p. 197. 35 David Chalmers, “The Content and Epistemology of Phenomenal Belief,” pp. 13–14.

58 Consciousness and the Prospects of Physicalism

example, the sense of pure phenomenal concept ‘R,’ a concept of R, that is, phenomenal redness, contains the what-it-is-like-to-sense-red qualitative nature as it is introspectively represented. Thus it will be a priori that this concept refers to a property whose qualitative nature is accurately represented by introspection, if it refers at all. Chalmers maintains that if a property lacks the qualitative nature of R as we introspectively represent it—that is, as pure phenomenal concept ‘R’ represents it—then it does not qualify as phenomenal property R. However, for this last claim to be established, it would have to be shown that it is not an open possibility for our introspective phenomenal representation of the qualitative nature of phenomenal redness to be inaccurate. The more specific proposal I have in mind is this: while phenomenal concept ‘R’ correctly picks out phenomenal property R, introspectively we represent R as having a certain qualitative nature, and it is an open possibility that R actually lacks it. This proposal gives rise to a challenge to the zombie argument’s first premise: Premise (1): ‘PT and ~ Q’ is ideally, positively, primarily conceivable. Assume that ‘Q’ is a truth about phenomenal property R. Securing Premise (1) does not require that if something feels conscious, it is conscious, exactly, but rather something like if something does not feel conscious, then it is not conscious. More precisely, it requires that if a property does not have a qualitative nature that is accurately represented by pure phenomenal concept ‘R,’ then the property represented is not R. If this phenomenal concept’s representation of R could indeed be inaccurate in this way, then the fact that R as represented introspectively is not a priori derivable from ‘PT’ fails to show that a true proposition about the real qualitative nature of R, or about R as it really is, is not a priori derivable from ‘PT.’ Thus it also fails to show that ‘Q,’ our selected truth about R, is not a priori derivable from ‘PT.’ Furthermore, if ‘Q’ is a priori derivable from ‘PT,’ then the ideal, positive, primarily conceivability of ‘PT and ~ Q’ is in serious jeopardy. What we were conceiving as the zombie-world would not be one in which ‘Q’ is false after all. By analogy, imagine Galileo assessing the claim that it is ideally, positively, primarily conceivable that a scenario be exactly as ours is microphysically, but without physical objects being colored. He would have agreed that this claim is in fact true, for he after all maintained that the actual world is one in which the physical objects are not colored. But while he would be right to hold that it is ideally, positively, primarily conceivable that there is a scenario exactly as ours is microphysically but without physical objects having colors whose qualitative natures are accurately represented by our visual color sensations, or without having primitive colors, he was, by Chalmers’s own plausible analysis (see chapter 2), mistaken to hold that there is a scenario exactly like ours microphysically but without physical objects being colored.36 An analysis of our color concepts will specify that in this 36 A qualitative accuracy hypothesis claims that when (in the normal way) we sense or introspect a secondary quality or a phenomenal property, we thereby represent it as having a specific qualitative nature that it actually has; on the stronger primitivist accuracy hypothesis, when we sense or introspectively represent a secondary quality or a phenomenal property, we thereby accurately represent its complete qualitative essence; cf. chapter 1.

59 Conceivability Arguments and Qualitative Inaccuracy

world physical objects are in fact colored, despite the qualitative inaccuracy of visual representation, and their not being primitively colored. I propose that while a scenario exactly like ours microphysically but without phenomenal properties whose qualitative natures are accurately represented introspectively, or without primitive phenomenal properties, is ideally, positively, primarily conceivable, at the same time a scenario exactly like ours microphysically, but without phenomenal properties themselves, is not ideally, positively, primarily conceivable. On analogy with the color case, the correct analysis of our phenomenal property concepts might well specify that in such a scenario there are phenomenal properties, even though there are no phenomenal properties whose qualitative nature is accurately represented introspectively, and no primitive phenomenal properties. Let’s say that a property is quasi-primitive just in case its qualitative nature is accurately represented by sensation or introspection (when a property is primitive, it is also quasi-primitive). In accord with the suggestions of the last chapter, analysis of the concept ‘red’ would reveal something like the following conjunction of conditionals: (C1+) If a world is actual in which physical objects are color-wise exactly as humans normally visually represent them to be, then: the concept ‘red’ correctly applies to quasi-primitive redness, and (C2+) If a world is actual in which no physical objects instantiate quasiprimitive redness, but there is a unitary property that is the normal cause of objects’ looking red, then: the concept ‘red’ correctly applies to the property of physical objects that is the normal cause of their looking red (where ‘the normal cause of their looking red’ functions merely as a reference-fixer); and (C3+) If a world is actual in which no physical objects instantiate quasiprimitive redness, but there are many different sorts of causes of looking red, and there are no salient similarities among the intrinsic properties of these causes, then: the concept ‘red’ correctly applies to whatever properties of physical objects cause (or could cause) instances of looking red . . . When Galileo claims that a scenario is ideally, positively, primarily conceivable that is just like ours microphysically but without physical objects being colored, he is at least implicitly assuming that conceptual analysis of ‘red’ discloses only the consequent of C1+. But conceptual analysis reveals conditionals that feature C2+ and C3+ as well, and in a scenario exactly similar to ours microphysically but without any quasi-primitively red physical objects, one of these conditionals (or a similar alternative) plausibly applies. If C2+ actually applies then given the a priori derivability of instantiations of straightforwardly physical nonfundamental properties, like being water, generally from ‘P,’ ‘P and no physical objects are red’ will not be ideally, positively, primarily conceivable. In this scenario, ‘PT and no physical objects are red’ will have the same

60 Consciousness and the Prospects of Physicalism

status as ‘PT and there is no water.’ Chalmers and Jackson argue that ‘PT and there is no water’ is ruled out a priori because all of the water truths are a priori derivable from ‘PT.’ Similarly, on their view, when one surveys a scenario correctly described by ‘PT,’ one can infer, without further empirical information, that physical objects in that world are colored. But now, since it’s a serious open possibility that introspective phenomenal representations are qualitatively inaccurate, it’s plausible that conceptual analysis would reveal the following sort of conjunction of conditionals for the concept ‘phenomenal red’: (P1+) If a world is actual in which experiences are qualitatively exactly as we introspectively represent them to be, then: the concept ‘phenomenal red’ correctly applies to quasi-primitive phenomenal redness, and (P2+) If a world is actual in which no experiences instantiate quasi-primitive phenomenal redness, but there is a unitary property that is the normal cause of their introspectively appearing phenomenally red, then: the concept ‘phenomenal red’ correctly applies to the property that is the normal cause of the introspective appearance of phenomenal redness (where ‘the normal cause of introspective representations of phenomenal redness’ functions merely as a reference-fixer), and (P3+) If a world is actual in which no experiences instantiate quasi-primitive phenomenal redness, but there are many different sorts of causes of their introspectively appearing phenomenally red, and there are no salient similarities among the intrinsic properties of these causes, then: the concept ‘phenomenal red’ correctly applies to whatever properties cause (or could cause) instances of the introspective appearance of phenomenal redness . . . When it seems intuitive that a scenario is ideally, positively, primarily conceivable that is exactly similar to ours microphysically but without instances of phenomenal redness, one may be implicitly supposing that conceptual analysis of ‘phenomenal red’ would reveal only something like the consequent of P1+. But because qualitatively inaccurate introspective representations of phenomenal properties cannot be ruled out a priori, conceptual analysis discloses conditionals like P2+ and P3+ as well. In a world just like ours microphysically but without any instances of quasi-primitive phenomenal redness, a conditional of this sort plausibly applies. Phenomenal color properties would then be instantiated by experiences in that scenario, and they would be straightforwardly physical. Given the a priori derivability of instantiations of straightforwardly physical properties such as chemical and neural properties from ‘PT,’ an ideal reasoner could then derive a priori from ‘PT’ that phenomenal properties are

61 Conceivability Arguments and Qualitative Inaccuracy

instantiated. Consequently, the ideal, positive, primary conceivability of ‘PT and ~ Q’ is jeopardized.37 Thus the zombie argument faces the following objection. Since the qualitative inaccuracy hypothesis is an open possibility, the analysis of phenomenal concepts discloses a conjunction of conditionals such as the conjunction that includes the conditionals P1+, P2+, and P3+. In a scenario microphysically just like ours (and ‘T’ holds) but without instantiated quasi-primitive phenomenal properties, phenomenal properties would nevertheless be instantiated. If such a scenario were actually realized, there would be no less reason to believe that ‘PT and ~ Q’ would be ruled out by ideal a priori reasoning than to believe that ‘PT and no physical objects are red’ or that ‘PT and there is no water’ would be so ruled out. As a result, the status of the ideal, positive, primary conceivability of ‘PT and ~ Q’ will not differ from that of ‘PT and there is no water’ and ‘PT and no physical objects are red.’ Thus, since the qualitative inaccuracy hypothesis is an open possibility, Premise (1) is insecure.38 Resistance to this objection might be explained away as we did analogous opposition to the related objection to knowledge argument. It is natural to assume at the outset that our phenomenal concepts are primitive or quasi-primitive (i.e., that their content is exhaustively Edenic), whereupon ‘PT and ~ Q’ would be conceivable. But

37 In “The Rationalist Foundations of Chalmers’s 2-D Semantics” (Philosophical Studies 118 [2004], pp. 227–55), Laura Schroeter remarks: There are two possible explanations for the fact that our commonsense understanding of physical and phenomenal predicates doesn’t allow us to derive a contradiction from the relevant sentence [P and ~Q]. One explanation is that according to our commonsense understanding, the properties picked out by the 1-intensions of phenomenal terms like ‘pain’ and microphysical terms like ‘quark’ are logically independent. A different explanation is that our ordinary understanding of physical and phenomenal predicates does not commit us one way or another: a priori considerations leave us undecided about whether these properties are logically independent. If this second explanation of our failure to derive a contradiction is right, then the zombie intuitions by themselves would show nothing about our commonsense commitments as to what’s possible and what’s not. On the proposal we’re now considering, it is left open by a priori considerations whether ‘Q’ is a priori derivable from ‘P,’ and thus a priori considerations alone leave us undecided about whether these truths are logically independent. 38 In Robert Stalnaker’s “What Is It Like to Be a Zombie?” in Conceivability and Possibility, pp. 385–400, his character Anne holds (in effect) that conceptual or metasemantic analysis reveals a conjunction of more than one conditional for ‘phenomenal red,’ and which conditional actually applies can only be established (if it can be) a posteriori. I agree with Anne and thus (presumably) with Stalnaker. But Stalnaker says that Anne will be a Type-B materialist, that is, someone who concurs with Chalmers that ‘P and ~ Q’ is ideally, positively, primarily conceivable, but denies that it is metaphysically possible. I don’t think that she needs to be, at least not on account of the fact that analysis of phenomenal concepts reveals a conjunction of more than one conditional, while which one actually applies is determinable only a posteriori. Rather, as I have just argued, she might reject Premise (1).

62 Consciousness and the Prospects of Physicalism

on the basis of the analogy with color concepts, it’s arguably an open possibility that this assumption about our phenomenal concepts is false. Moreover, as Dowell contends, the openness of this possibility can be grounded in our envisioning future empirical discoveries that will indicate to us that our phenomenal concepts feature nonprimitive ordinary content.39 Since ideal conceivability of a proposition reckons with the entire conceptual structure of its constituent concepts, and on this open possibility nonprimitivist conditionals such as P2+ and P3+ will be part of the structure of phenomenal concepts, ‘PT and ~ Q’ would consequently not be ideally conceivable. Erroneously taking ‘PT and ~ Q’ to be ideally, primarily, positively conceivable is then to be explained by the assumption of a natural but mistaken conception of the structure of phenomenal concepts. Note that even given our open possibility we can agree that ‘PT and ~ Q’ is ideally, primarily, positively conceivable on the assumption that our phenomenal concepts are primitive. But from this we should not conclude that ‘PT and ~ Q’ is ideally, primarily, positively conceivable tout court; our phenomenal concepts’ having imperfect content blocks this inference. By analogy, we would not allow that ‘P and no physical objects are colored’ is ideally, primarily, positively conceivable on the ground that this claim is ideally, primarily, positively conceivable on the assumption that our color concepts are primitive. On the dual-content view for color perception that Chalmers advocates, it is the imperfect content of our color concepts that blocks this inference.

HIGHER LEVELS Still, the crucial point, one might object, has been overlooked. Phenomenal concepts differ in an important respect from color concepts. When one claims that our visual color concepts might not provide us with an accurate representation of the qualitative nature of colors, one can retain the view that a visual mode of presentation of a color is a merely mental phenomenon—not one that accurately attributes a property to something in the external world. However, when one contends that the what-it-is-liketo-sense-red introspective mode of presentation inaccurately represents the property of phenomenal redness, it seems that the challenge to the physicalist arises again, for one must still affirm this mode of presentation as an instantiated mental feature. As was the case for the knowledge argument, one might think that this point about the conceivability argument merely displaces the problem for consciousness. Let R* be the higher-level property of being the introspective mode of presentation of phenomenal-redness, and ‘Q*’ a truth about R*. The suggestion is that a successful conceivability argument can now be constructed with (1*) ‘PT and ~ Q*’ is ideally, positively, primarily conceivable. as its first premise. 39 Janice Dowell, “Empirical Metaphysics: The Role of Intuitions about Possible Cases in Philosophy.”

63 Conceivability Arguments and Qualitative Inaccuracy

The response is the same: it is also an open possibility that our introspective representation of R* is qualitatively inaccurate, whereupon the analysis of our concept ‘R*’ would contain conditionals similar in form to P2 and P3. Hence, in a scenario microphysically identical to ours but without instantiated quasi-primitive introspective modes of presentation of phenomenal properties, R* would nevertheless be instantiated. Thus there is no less reason to believe that ‘PT and ~ Q*’ is ruled out by ideal a priori reasoning than to believe that ‘PT and no physical objects are red’ or that ‘PT and there is no water’ is so ruled out. The ideal, positive, primary conceivability of ‘PT and ~ Q*’ is therefore vulnerable. Because it is open that there is no relevant disanalogy between ‘red’ or ‘water’ and the higher-level phenomenal concept, Premise (1*) is also insecure.

APPENDIX: A CHALLENGE TO THE TRANSITION FROM PRIMARY TO SECONDARY POSSIBILITY ? It is natural to suspect that these same considerations would also yield an objection to Premise (3): (3) If ‘PT and ~ Q’ is primarily possible, then ‘PT and ~ Q’ is secondarily possible, or Russellian monism is true. First, if it is an open possibility that introspective representations of phenomenal properties are qualitatively inaccurate, then phenomenal concepts will have the sort of complex profile ‘red’ or ‘water’ does. A priori conceptual reflection will then not determine the reference of ‘R,’ whether it is, for example, a nonphysical property or any of a variety of physical properties, for a priori conceptual reflection lacks the resources to determine which of these various possibilities is actual. And if the reference of a concept would have to be determined by empirical investigation, it will not be primarily rigid. Here, then, is the challenge to Premise (3) (I’ll leave out the ‘T’ for simplicity). Assume again that ‘Q’ is an actual truth about phenomenal property R and that phenomenal concept ‘R,’ by means of which R is represented in statement ‘Q,’ is not primarily (or epistemically) rigid. Then, supposing that ‘P and ~ Q’ is not ruled out by a priori conceptual reflection, it might nevertheless be ruled out by what we discover about the reference of ‘R’ by a posteriori investigation. For instance, we might make the empirical discovery that ‘R’ refers to a property whose entire nature is straightforwardly physical and is a priori derivable from ‘P.’ In any event, as in the ‘water’ case, the secondary possibility of ‘P and ~ Q’ would no longer be derivable from its primary possibility. However, at this point Chalmers advances an alternative to Premise (3) that appeals to the notions of primary and secondary intension. The primary intension of S is that set of worlds in which, when considered as actual, S is true; the secondary intension of S is that set of worlds in which, when considered as counterfactual, S is true. The primary intension of ‘there is water,’ for example, includes H2O worlds and certain XYZ

64 Consciousness and the Prospects of Physicalism

worlds, whereas its secondary intension includes only the H2O worlds. In addition, Chalmers specifies that “the primary intension of S is true in W if the material conditional ‘if W is actual, then S’ is a priori: that is, if the hypothesis that W is actual and S is not the case can be ruled out a priori.”40 So the primary intension of ‘there is water’ is true in the XYZ world, since the material conditional ‘if the XYZ world is actual, then there is water’ is a priori; the hypothesis that the XYZ world is actual and there is no water can be ruled out a priori. Chalmers then argues as follows: One can observe that if ‘P’ and ‘Q’ both had identical primary and secondary intensions (up to centering), then Premise (3) would be straightforwardly true. Further, it is very plausible that the most important phenomenal concepts do indeed have the same primary and secondary intensions . . . so that Q at least can be accommodated here. And even if this is false, Q’s primary intension can be seen as the secondary intension of some other truth Q*, which stands to Q roughly as ‘watery stuff ’ stands to ‘water.’ As long as P has the same primary and secondary intension, then the primary possibility of P and ~ Q* will entail the secondary possibility of P and ~ Q*, which will itself entail the falsity of materialism.41 In correspondence, Chalmers embellishes this argument. We need first to canvass some further terminology. If S is true in W considered as actual, then W verifies S; if S is true in W considered as counterfactual, then W satisfies S. Centered worlds are possible worlds with a point of reference marked, proposed by David Lewis to provide a semantics for indexical and demonstrative statements. For example, a world marked with me at the center can be used to provide semantics for my statements using the term ‘I.’42 Here is Chalmers: Let’s allow that P and ~ Q is primarily conceivable and that primary conceivability entails primary possibility. Then there’s a centered world that verifies P and ~ Q. Or equivalently, there’s a centered world that satisfies P* and ~ Q*, where these are the primary intensions of P and Q respectively. Let’s allow with you that Q isn’t epistemically rigid and so may have quite different primary and secondary intensions. So Q* may be quite different from Q in modal profile. Nevertheless we have a world that (i) satisfies P and (ii) in which some truth about our world fails to hold. That’s enough for physicalism to be false in our world. The key claim—call it Premise (3*)—is that

40 David Chalmers, “Does Conceivability Entail Possibility?” p. 163. 41 David Chalmers, “Does Conceivability Entail Possibility?” p. 197 (with Q* substituted for Q′). 42 David Lewis, “Attitudes De Dicto and De Se,” Philosophical Review 88 (1979), pp. 513–43.

65 Conceivability Arguments and Qualitative Inaccuracy

(A) there’s a centered world considered as actual in which ‘P and ~ Q’ is true, is equivalent to (B) there’s a centered world considered as counterfactual in which ‘P* and ~ Q*’ is true where the secondary intensions of ‘P*’ and ‘Q*’ are the primary intensions of ‘P’ and ‘Q.’ The advantage of this claim, given the objection to Premise (3) I’ve suggested, is that it does not require a phenomenal concept or a phenomenal truth whose primary and secondary intensions are the same. And Premise (3*) seems clearly true. For ‘S’ to be primarily possible is just for there to be a metaphysically possible world in which the primary intension of ‘S’ is true. Given my concern about the primary rigidity of phenomenal concepts, this is a more advantageous way of developing the argument. In consequence, my criticism of the zombie argument focuses just on Premise (1), and not also on the third premise.

4 QUALITATIVE INACCURACY AND RECENT OBJECTIONS TO CONCEIVABILITY ARGUMENTS

Evaluation of several recent criticisms of conceivability arguments against physicalism highlights the advantages of the challenge to arguments of this kind based on the qualitative inaccuracy hypothesis. This challenge crucially involves the claim that we introspectively represent phenomenal properties as having certain qualitative features, while it is an open possibility that they actually lack them. Robert Stalnaker, John Hawthorne, and David Braddon-Mitchell each contend that the apparent conceivabilities at issue in these arguments presuppose the falsity of physicalism, and as a result, they fail to pose a strong objection to the physicalist. Daniel Stoljar argues that due to our ignorance, the conception of the physical operative in these arguments is in effect not ideal and that this undercuts their force. Stephen Yablo calls into question the Kripke-inspired thought that once it is established that conceivability arguments are not subject to confusion about which propositions are being conceived, then they are in the clear. In his analysis, a defect that may still persist is a failure of the ideality condition on conceivability. In this chapter, I argue that in each case the qualitative inaccuracy hypothesis can correct or embellish these contentions profitably. In the discussion of Yablo’s contribution, I develop a more comprehensive diagnosis of why conceivability arguments might seem compelling while at the same time be unsound.

STALNAKER , HAW THORNE, AND BRADDON-MITCHELL ON ZOMBIE CONCEIVABILITY First, Stalnaker, Hawthorne, and Braddon-Mitchell have argued that Chalmers’s zombie argument can be undermined by the following sort of consideration.1 We can imagine finding out that the actual world is merely physical—for example, as 1 Robert Stalnaker, “What Is It Like to Be a Zombie?”; John Hawthorne, “Advice for Physicalists”; David Braddon-Mitchell, “Qualia and Analytical Conditionals.” 66

67 Qualitative Inaccuracy and Recent Objections to Conceivability Arguments

Hawthorne specifies, God or an oracle may tell us that it is. Under these circumstances, our phenomenal terms refer to physical properties, and from this it would follow, for Kripkean reasons, that zombies are impossible. But we can also imagine finding out that the actual world is not merely physical and that all phenomenal properties are nonphysical properties, whereupon phenomenal terms refer to nonphysical properties, and zombies would be metaphysically possible. As a result, on Hawthorne’s suggestion, although we can now conceive of zombies, we should not be confident that zombies are metaphysically possible, nor should we be confident that zombies would remain conceivable (in the right sense), were we to be fully informed about the actual world.2 In Stalnaker’s view, if in this world phenomenal properties coincided with nonphysical properties (“a-properties”), then the following counterfactual would be true: “if we didn’t have any of the a-properties, but the world were physical just as it is, then we wouldn’t be conscious—we would be in a zombie world.” But “if the materialists are right, and we live in a z-world (a merely physical world), then there are no possible worlds correctly describable as zombie worlds.”3 He then concludes: Whether or not Dave’s [the Chalmers character in Stalnaker’s paper] dualism is true, if we can coherently suppose that it is true, then we can coherently suppose that zombies are possible, and so can form a coherent conception of zombies. But if this is the only sense in which zombies are conceivable, their conceivability will provide no argument against materialism, since we must assume that materialism is false to be justified in inferring that zombies are possible from the fact that they are conceivable.4 Chalmers’s reply to this objection is that the sense in which zombies are conceivable it invokes is (merely) 1-2 conceivability, where S is 1-2 conceivable “if it is primarily conceivable that S is secondarily possible.”5 He concurs that the conceivability of zombies in this sense “does not directly entail the falsity of materialism.” Chalmers contends, however, that this sort of conceivability has no role in his own arguments: “What is relevant is simply the 1-conceivability of P and not-Q . . . that is, the claim that P and not-Q is primarily positively conceivable.” But “Stalnaker says nothing to cast doubt on this claim . . . and he says nothing to cast doubt on the inference from primary conceivability to primary possibility. So his discussion leaves this argument untouched.”6 Do Stalnaker’s and Hawthorne’s reflections serve to explain away the zombie intuition, the sense that (a) ‘P and ~ Q’ is ideally primarily, positively conceivable?

2 3 4 5 6

John Hawthorne, “Advice for Physicalists,” p. 25. Robert Stalnaker, “What Is It Like to Be a Zombie?” p. 399. Robert Stalnaker, “What Is It Like to Be a Zombie?” p. 399. David Chalmers, “Does Conceivability Entail Possibility?” p. 198–99. David Chalmers, “Does Conceivability Entail Possibility?” p. 199.

68 Consciousness and the Prospects of Physicalism

The objection has it that the relevant conceivability is: (i) Given the background belief that phenomenal properties are nonphysical, it is conceivable that ‘P and ~ Q’ is metaphysically possible,7 and that, by contrast with the ideal, positive, primary conceivability of ‘P and ~ Q,’ this does not yield a plausible argument for the conclusion that ‘P and ~ Q’ is metaphysically possible (or even that it is primarily possible). Rather, whether ‘P and ~ Q’ is metaphysically possible depends on whether physicalism or dualism is true about phenomenal properties, and hence nothing in the neighborhood of a zombie intuition provides an independent route to the falsity of physicalism. But what is it about this account that would explain away the zombie intuition? Braddon-Mitchell proposes that a conditional like (i) is the “shadow” of (a). The thought is that (i) is true, (a) is false, and the conflation of (a) with (i) explains why people think that (a) is true. However, as Torin Alter argues, it’s implausible that this confusion is being made by proponents of the zombie argument, and by Chalmers in particular.8 In addition, it isn’t dialectically effective to claim that the zombie conception surreptitiously involves a background supposition of a dualist world considered as actual. On a charitable assessment, the zombie conception does not involve any supposition as to whether the actual world is physicalist or not, and assuming this charitable reading is required for an effective response to the argument. As Alter sets up the objection, Hawthorne and Braddon-Mitchell maintain that the following conditional claims are true as a matter of a priori conceptual analysis: If the world contains appropriate nonphysical states, then phenomenal concepts refer to them. If the world is merely physical, then phenomenal concepts refer to physical states. One of the most telling of Alter’s criticisms of this version of the objection is that it would rule out the ideal primary conceivability (and primary possibility) of a zombie world on the basis of a priori conceptual analysis alone. Alter contends, rightly, I think, that this is highly implausible. Some further argument would be required to overcome this implausibility. In addition, Hawthorne’s version of the proposal uses the vehicle of imagined oracular revelation to lend support to the primary conceivability (and possibility) of certain of his claims; in his scenario, God or an oracle tells us that the actual world is merely physical.9 A problem for this strategy is that when a claim is a priori false, but 7 Nico Silins (in conversation) suggests this reading, by contrast with the slightly stronger version I initially proposed: Given the supposition that phenomenal properties are nonphysical, it is conceivable that ‘P and ~ Q’ is metaphysically possible. 8 Torin Alter, “On the Conditional Analysis of Phenomenal Concepts,” Philosophical Studies 134 (2007), pp. 235–53. 9 Alter effectively argues that this method cannot be sustained; see Torin Alter, “On the Conditional Analysis of Phenomenal Concepts.”

69 Qualitative Inaccuracy and Recent Objections to Conceivability Arguments

not evidently so, we can still imagine an oracle telling us that it is true. For instance, we might imagine an oracle telling us that Goldbach’s conjecture is true, even though it is a priori false, or that it is false, even though it is a priori true. So then perhaps we could imagine an oracle telling us that this world, in which there are pains, is wholly physical, even if it is not primarily possible that the world be physical and feature pains. For these reasons, my sense is that the kind of objection Stalnaker, Hawthorne, and Braddon-Mitchell have raised can’t be the right one. I propose instead that zombie conceivability does not involve a background supposition of nonphysicalism, but rather that introspective representations of phenomenal properties are qualitatively accurate, or perhaps primitive. This one might well naturally suppose without at the same time assuming nonphysicalism. And indeed, if conceptual analysis were to reveal that phenomenal concepts refer to primitive phenomenal properties or, more cautiously, those with qualitative features that are accurately represented introspectively, then ‘P and ~ Q’ and zombies would be ideally, positively, primarily conceivable. On the Stalnaker, Hawthorne, and Braddon-Mitchell type of objection, when confusions are eliminated, there is no conceivability phenomenon that rationally counts in favor of a nonphysicalist view. But given that zombie conceivability assumes only qualitative accuracy, which is at least initially very compelling, this is not so. I suggest that the supposition of qualitative accuracy or of primitivism is the linchpin of the argument and that it, rather than a background nonphysicalist supposition, should be the focus of an objection to it. If the accuracy supposition can be doubted, then the ideal, positive, primary conceivability of ‘P and ~ Q’ can be questioned as well.

STOLJAR’S EPISTEMIC STRATEGY Second, Stoljar’s response to the knowledge and conceivability arguments features most prominently the claim that it is due to our ignorance that we tend to find these arguments persuasive.10 This type of challenge to arguments of this sort has a long history. A version was provided by Antoine Arnauld in response to Descartes’s conceivability argument,11 and another was advanced more recently by Paul Churchland in opposition to the knowledge argument.12 But Stoljar provides its most thorough development. In the case of Chalmers’s conceivability argument, his proposal is to explain the conceivability or the seeming possibility of a zombie world as resulting from ignorance of relevant facts. I agree with Stoljar insofar as I think that relevant ignorance is part of the explanation of the conceivability and seeming possibility at issue, but his epistemic view would be strengthened by a more specific hypothesis about how we might be ignorant in a respect that would facilitate this sort of explanation, and this is a role the qualitative inaccuracy hypothesis can play. 10 Daniel Stoljar, Ignorance and Imagination (New York: Oxford University Press, 2006). 11 Antoine Arnauld, Fourth Set of Objections, in The Philosophical Writings of Descartes, vol. 2, p. 142 (AT VII 202). 12 Paul M. Churchland, “Reduction, Qualia and the Direct Introspection of Brain States,” Journal of Philosophy 82 (1985), pp. 8–28.

70 Consciousness and the Prospects of Physicalism

Here is Stoljar’s particular take on the problem of consciousness. Many philosophers maintain that the experiential supervenes on the nonexperiential, holding that in no sense is the experiential fundamental, and that a complete account of reality can in principle be provided in nonexperiential terms. But the conceivability and knowledge arguments contest these claims. It’s conceivable that there be a world identical to this one in all nonexperiential respects but distinct in experiential respects, and from this, with some additional reasoning, it can be inferred that this world is genuinely possible. In addition, we can imagine a scientist who knows all that there is to know about the nonexperiential but who lacks certain kinds of experiences herself and lacks some knowledge of the experiential. The scientist lacks the whole truth about the experiential. When she comes to have these experiences, she intuitively learns something. From these arguments, it is concluded that the experiential is something in some sense over and above the nonexperiential. Stoljar contends that the most reasonable physicalist response to the problem of consciousness involves supposing that we are ignorant of a certain type of physical or nonexperiential truth. This supposition, he thinks, explains why the arguments that drive the problem are as compelling as they are and why they are not genuinely persuasive. To advance this ignorance response, Stoljar profitably employs analogies of beings with inferior sensory and cognitive capacities as analogies to its central point.13 One such analogy illustrates the claim that the sort of ignorance that would be relevant to the antiphysicalist arguments is of fundamental features of nonexperiential reality, for example, ignorance of fundamental forces and particles or of the categorical properties that underlie fundamental dispositional properties; Stoljar call this fundamental-level ignorance. Another illustrates the possibility that the relevant sort of ignorance is of how features of the fundamental level combine to result in features at higher levels; he calls this intermediate-level ignorance. The following example illustrates fundamental-level ignorance: The Slugs and the Tiles. Imagine a mosaic completely constructed from two sorts of tiles, triangles and pieces of pie. The mosaic may have many different shapes in it, so long as those shapes are constructed by transparent principles from the basic ones: circles, figure-eights, half-moons, rectangles, rhombuses, and so on. Now imagine a population of intelligent slugs who live on the mosaic and are cognitively sophisticated but whose perceptual access to it is limited to two shape-detecting systems: the first scans the mosaic and detects triangles, the second scans it and detects circles. Given their epistemological access to the mosaic, it would be natural for these slugs to think that, at least so far as the tiles of the mosaic are concerned, it was constituted only by triangles and circles. This is a mistake, but it would be a natural one in the situation.

13 Daniel Stoljar, Ignorance and Imagination; “Précis of Ignorance and Imagination,” Philosophy and Phenomenological Research 79 (2009), pp. 748–55.

71 Qualitative Inaccuracy and Recent Objections to Conceivability Arguments

In this example, the slugs might advance the following argument: It is conceivable that there is a zombie mosaic, that is, a mosaic exactly like the actual mosaic in noncircular respects but that lacks circles entirely. If this is conceivable, it is possible. This argument is not genuinely persuasive, and the slugs’ ignorance explains why it seems persuasive to them. The second example illustrates intermediate-level ignorance: The Moths and the Beams. Imagine an array of light beams, completely constructed by red and green lights. The beams may be any color, so long as those colors are derived from an admixture of red and green. Yellow beams can be derived from some admixture of red and green beams. Suppose for simplicity that there are only red, green, and yellow lights. Imagine a population of intelligent moths who live among the beams. The moths possess visual systems capable of detecting beams of red, green, and any admixture of these. However, while they are cognitively sophisticated, the moths are ignorant of the principles whereby various lights combine together to create different colored beams. Given their epistemological access to the array, it would be natural for these moths to think that its basic elements are three sorts of beams: red, green, and yellow. This is a mistake, but it would be a natural one for the moths to make in the situation. In this example, the moths advance the following argument. It is conceivable that there is a zombie array, that is, an array exactly like the actual array in terms of beams that aren’t yellow but different from it in terms of yellow beams. If this is conceivable, it is possible. This argument is again not genuinely persuasive, but the moths’ ignorance explains why it seems persuasive to them. Thus the difference between moth example and the slug example is that for the moths ignorance is of an intermediate-level truth about how light beams of certain colors combine to yield differently colored light beams, while the slugs are ignorant of a fundamental-level truth relevant to the truth about circles. According to Stoljar’s epistemic view, we are ignorant of a type of nonexperiential (or physical) truth relevant to the truth about experience, and that truth might be either at the fundamental level or at an intermediate level. Schematically: The ignorance hypothesis: we are ignorant of a type of nonexperiential experience-relevant truth.

72 Consciousness and the Prospects of Physicalism

gives rise to: The epistemic view, which consists of two claims: E1. If the ignorance hypothesis is true, the conceivability argument is unpersuasive. E2. The ignorance hypothesis is true. Which specific mistakes are the moths and the slugs making? Stoljar identifies three possibilities: Proposition confusion: p may have been confused with some other proposition q. Mode confusion: p may be conceivable but in the wrong way or mode. Defeater neglect: The inference from conceivability to possibility may be subject to defeaters that have not been taken into account. Given the ignorance hypothesis, it is plausible that we are making one of (or a combination of) the standard mistakes. In the first analogy, the slugs are guilty (at least) of proposition confusion: they confuse the target proposition (1) with a contrasting proposition (2): (1) There is a mosaic exactly like the actual mosaic in noncircular respects, but it lacks circles entirely. (2) There is a mosaic exactly like the actual mosaic in triangular respects, but it lacks circles entirely. It is plausible in the story that the slugs have conceived, in a requisitely strong sense, that (2) is true. But they need the conceivability of (1) to justify a conclusion that threatens the supervenience of the circular on the noncircular. To bring this illustration to bear on the conceivability argument, consider: (3) There is someone identical to me in respect of all nonexperiential truths but who differs from me in respect of some experiential truth. (4) There is someone identical to me in respect of all known nonexperiential truths but who differs from me in respect of some experiential truth. Perhaps we have conceived that (4) is true in the requisitely strong sense. But we need the similarly strong conceivability of (3) to derive a conclusion that threatens the supervenience of the experiential on the nonexperiential. (E2), the claim that the ignorance hypothesis is true, is credible on general grounds. We are in fact empirically ignorant about the nature of conscious experience, and our current epistemic situation has historical precedents. Stoljar develops several interesting analogies and partial hypotheses designed to support the ignorance hypothesis. The plausibility of fundamental-level ignorance is illustrated by the example of a claim C. D. Broad made during the 1920s, that there is no possible prediction or account of

73 Qualitative Inaccuracy and Recent Objections to Conceivability Arguments

the chemical properties of water solely on the basis of a complete specification of the properties of its constituents, hydrogen and oxygen, and how they are put together, and that hence these chemical properties are emergent.14 Physical theory developed soon thereafter showed that the persuasiveness of this argument was due to ignorance. Stoljar illustrates intermediate-level ignorance by Descartes’s contention that material things could never have the human capacities for reason and language.15 It is credible that this argument was compelling for Descartes because he was ignorant of certain intermediate-level physical truths, in particular, truths about physical computational possibilities. As Karen Bennett points out, one way to test the plausibility of such an ignorance hypothesis is by discerning how it might differ from unpersuasive strategies that employ ignorance in a similar way.16 Stoljar’s position can be represented as a kind of skepticism, justified on the basis of our ignorance, about the ideal, positive, primary conceivability of zombies. We might compare it to a type of skepticism, similarly justified on the basis of our ignorance, about a well-confirmed scientific theory. So consider first: We can reasonably claim that the physical evidence that there is to be had strongly supports the truth of quantum mechanics. It is generally agreed that quantum mechanics (QM) is well supported by the relevant evidence (EV); if you disagree, substitute another appropriate theory, perhaps the DNA theory of transmission of genetic information. However, the skeptic points out that because our cognitive capacities for understanding physics are limited, we are in no position rationally to dismiss the claim that there is a currently unspecified theory distinct from QM that is metaphysically more plausible and that explains EV as well as QM does. From this, he contends, we can draw the conclusion that we have no good reason to believe that QM is more likely on EV than not.17 In support of his position, our QM-skeptic cites the general fact of human ignorance, the obvious fact that we are at least to some significant degree ignorant of the physical evidence that there is to be had, and parallel examples, such as we used to think that our evidence strongly supports the truth of the ether hypothesis about the propagation of light, but we now think that our evidence does not support this theory.

14 Daniel Stoljar, Ignorance and Imagination, pp. 135–40; C. D. Broad, Mind and Its Place in Nature (London: Kegan Paul, 1925), pp. 62–63 (see chapter 7 for a discussion of emergence). 15 Daniel Stoljar, Ignorance and Imagination, pp. 124–34; René Descartes, Discourse on Method V, in The Philosophical Writings of Descartes, vol. 1, p. 140 (AT VI 56). 16 Karen Bennett, “What You Don’t Know Can Hurt You,” Philosophy and Phenomenological Research 79 (2009), pp. 766–74, at pp. 770–74; Derk Pereboom, “The Problem of Evil,” in The Blackwell Guide to Philosophy of Religion, ed. William E. Mann (Oxford: Blackwell, 2004), pp. 148–70. 17 I discuss this type of issue in the context of a critical evaluation of skeptical theism in Derk Pereboom, “The Problem of Evil.”

74 Consciousness and the Prospects of Physicalism

Now consider: We can reasonably claim that the first premise of the zombie argument, that ‘P and ~ Q’ is ideally, positively, primarily conceivable, is true. To reasonably claim that the primary conceivability of ‘P and ~ Q’ is ideal, one must have good reason to believe that the most thorough understanding of P would not undermine ‘P and ~ Q’s primary conceivability. That is, one must have good reason to believe that the most thorough understanding of P would not allow one to see that ‘P and ~ Q’ is ruled out a priori. Stoljar argues that against this ideality claim we can cite the general fact of human ignorance, the obvious fact that we are at least to some significant degree ignorant of the physical, and parallel examples such as Descartes’s language case and Broad’s chemistry example, where, due to our increased physical knowledge, we can now conceive what Descartes and Broad claimed to be inconceivable. On the assumption that the QM-skeptic’s case is not especially strong, one might want to show how Stoljar’s ignorance response is relevantly different from this skeptic’s strategy. It is not enough merely to point out that we are in no position rationally to dismiss the claim that there are facts about the nonexperiential that would render inconceivable beings that are identical to us physically but lack consciousness. After all, we are also in no position rationally to dismiss the claim that there is physical evidence to be had that would support a currently unspecified theory distinct from QM that is metaphysically more plausible and that explains EV as well as QM does. To enhance his case, the QM-skeptic would need to adduce a reason to increase our confidence that there really is a theory that rivals QM. Articulating a partially filled-out hypothesis, by contrast with an unspecified one, would help serve that purpose. In addition, it seems intuitive that the lower the probability of a QM-skeptical hypothesis, the smaller its lowering effect on the claim that QM is true. So to make the strategy work, the QM-skeptic would require a partially filled-out hypothesis with significant probability—one, let’s say, that could really be true. By analogy, Stoljar’s case would benefit from a partially specified hypothesis about the nonexperiential that that could really be true. About such a hypothesis we would need some reason to think that if it were true, and we understood and believed it, zombies would be inconceivable. Stoljar’s examples from Broad and Descartes are relevant analogies. But one worry is that they do not involve the experiential or the phenomenal in particular, which, according to the proponents of the conceivability argument, are sui generis. Brie Gertler voices this concern: Support for the manifest supervenience thesis stems from reflection on particular scientific advances: some manifest features of the world (such as macroscopic phenomena) have been explained by the non-manifest (e.g., the behavior of microscopic particles, which is in turn explained by the presence and nature of quarks, leptons, etc.). These cases help to establish the manifest supervenience argument only if we may legitimately generalize from such explanations to conclude that all manifest phenomena can, in principle, be

75 Qualitative Inaccuracy and Recent Objections to Conceivability Arguments

explained by the non-manifest. Is that generalization legitimate? Note that the cases of successful scientific explanation concern phenomena that are nonexperiential and are manifest to perception. One need not be a primitivist to recognize a significant difference between these phenomena and phenomena that are experiential and are manifest to introspection. As Stoljar himself notes, the science of consciousness is in its infancy, and has not yielded any clear explanations of the experiential in non-manifest terms. The disparity between the manifest nonexperiential and the (manifest) experiential casts doubt on whether our success in explaining the former provides reason to believe that the latter could yield to a similar explanation.18 Stoljar’s nonintrospective analogies do not generate a partially specified hypothesis specifically about the experiential or the phenomenal that would lend significant support to the claim that zombies are not conceivable in the requisitely strong sense. My suggestion for answering this concern is to cite the hypothesis—which is at least partially specified, and which I’ve argued could really be true—that introspection represents phenomenal properties as having certain qualitative features and that these properties actually lack them, or else a similar hypothesis about primitive phenomenal properties. Under the supposition that this open possibility of qualitative inaccuracy is actually realized and that we grasped this fact, we should be convinced that the ideal, positive, primary conceivability of ‘P and ~ Q’ is in jeopardy. Indeed, if we knew that the open possibility was in fact realized, the key reason for believing that ‘Q’ is not a priori derivable from ‘P’ (and ‘T’) would be removed. Returning to our actual epistemic situation, and considering the open possibility just as a hypothesis about how things might turn out to be, not assuming that it is actually realized, the analogy to better established characteristics of secondary-quality representation provides significant (although not decisive) reason to believe that this hypothesis is in fact true. Ordinary secondary-quality sensation plausibly represents secondary qualities as having certain qualitative features that they actually lack. The fact that our various secondary-quality sensory systems might well produce representations that are qualitatively inaccurate in this respect affords credibility to the claim that our representations of introspective phenomenal properties are inaccurate in a similar way. This open possibility is serious enough to provide us with an appreciable reason to doubt the first premise of Chalmers’s conceivability argument, that ‘P and ~ Q’ is ideally, positively, primarily conceivable. My sense is that this open possibility is sufficiently credible to preclude rational conviction that this premise is true.

YABLO AND THE PRIMITIVIST HYPOTHESIS Third, the primitivist enhancement of the qualitative inaccuracy hypothesis yields a response to Kripke’s argument for dualism that meets a standard for which Stephen Yablo has recently argued.19 Kripke contends that by contrast with the seeming possibility of 18 Brie Gertler, “Daniel Stoljar’s Ignorance and Imagination,” Noûs 43 (2009), pp. 378–93, at p. 387.

76 Consciousness and the Prospects of Physicalism

(1) heat ≠ high molecular kinetic energy (HME), the seeming possibility of (2) pain ≠ C-fiber firing cannot be explained away by the genuine possibility of something that merely appears like the referent of the term on the left-hand side of the nonidentity statement not being identical to the referent of the term on its right-hand side. It is genuinely possible that (1*) something that feels like heat but is not in fact heat, for example, low molecular kinetic energy, ≠ HME, and because one might not properly distinguish (1) and (1*), the genuine possibility of (1*) (or the recognition of it) can explain away the seeming possibility of (1). But no similar explaining-away is in the offing for (2), since whatever feels like pain is in fact pain, and therefore the seeming possibility of (2) cannot be explained away by the proposal that the possibility really being accessed is not (2) but instead (2*) something that feels like pain but is not in fact pain ≠ C-fiber firing.20 Yablo argues, however, that the seeming possibility of (1) cannot plausibly be explained away by Kripke’s strategy. With one amendment, I agree. Yablo then suggests an alternative method for explaining away the seeming possibility of (1) and that the seeming possibility of (2) might be explained away in the same way as the seeming possibility of (1) after all. On the primitivist open possibility, this is in fact so. Yablo develops two compelling concerns for Kripke’s claim that the seeming possibility of (1) can be explained by way of the possibility of (1*). The first is that to explain away the seeming possibility to me or to us that something other than HME could be heat, it had better be that there is something I have in mind other than HME that would feel like heat to me or to us. It couldn’t be, for example, that what I’m imagining is an otherworldly being, whose sensory system is wired differently, to whom low molecular kinetic energy (LME) feels like heat. Here Yablo introduces otherworldly counter-Steve, who has heatish phenomenology when sensing LME and coldish phenomenology when sensing HME: “I am liable to confuse A with B because they look the same to me” sounds quite plausible. If things look the same, then one is quite liable to confuse them. “I am liable to confuse A with B because the same looks result if it is me looking at A or counter-Steve looking at B.” There is no chance at all that I am confusing 19 Stephen Yablo, “No Fool’s Cold,” in Two-Dimensional Semantics, pp. 327–45. 20 Saul Kripke, Naming and Necessity, pp. 144–53.

77 Qualitative Inaccuracy and Recent Objections to Conceivability Arguments

myself with counter-Steve, even if his phenomenology is just the same. Counter-Steve is by definition a person who sees things differently than I do.21 At this point, Yablo invokes what he calls the psychoanalytic standard: “Assuming the conceiver is not too self-deceived or resistant, ◊F explains E’s seeming possibility only if he/she does or would accept it as an explanation, and accept that his/her intuition testifies at best to F’s possibility, not E’s.” So when it seems possible to Yablo that heat ≠ HME, he would not accept as an explanation of this seeming possibility that there is an otherworldly counterpart for whom LME feels like heat. The upshot is this generalization: Facsimile principle: to explain why this, understood to present like so, seems like it could turn out to be Q, one needs a possible scenario in which something superficially indistinguishable from it does turn out to be Q. The counterfactual thing has to look the same, not to the counterfactual folks, but to us.22 A facsimile of an actual thing, by Yablo’s lights, is something that appears like the actual thing to me, constituted as I am, or to us, constituted as we are. Constituted as I am, and as we are, LME can’t feel like heat to us. To my mind, Yablo’s diagnosis is close to correct, but perhaps there is a little more leeway for what constitutes an adequate explanation of a seeming possibility than his facsimile principle allows. Still, I don’t think that there will be enough leeway to save the explaining away of (1) by (1*). Supposing that p1 and p2 are physical properties—say, spectral reflectance profiles— imagine the color scientists tell us that: brown = p1 green = p2 Sue then thinks, “It seems that p2 could have been brown.” The background story is that there’s an article in the World Book Encyclopedia with two pictures: one of a scene in natural color, and the other of what that scene looks like to someone who is red-green color-blind. Sue muses, “The things that look green to me look brown to someone who is red-green color-blind.” There are people in Sue’s community who are red-green color-blind, and Joe, Sue’s brother, is one of them. Sue thinks, “I could have been color-blind just like Joe is, and then things that actually look green to me would have looked brown to me. What’s more, every human being, for all of history, might have had color vision just like Joe’s, and then things that actually look green to me would have looked brown to everyone. But then, p2 would have been brown.” The explaining away for this seeming possibility—of p2’s being brown—will be something like: what Sue is thinking is that there is a not-very-far-off scenario in which p2 21 Stephen Yablo, “No Fool’s Cold,” p. 337. 22 Stephen Yablo, “No Fool’s Cold,” p. 337.

78 Consciousness and the Prospects of Physicalism

looks brown to everyone. But this is not really a scenario in which p2 is brown or, more cautiously, not a scenario in which p2 is a different color from the one it actually is. It’s my sense that this counts as an explaining away of the seeming possibility of p2’s being brown that meets a plausible psychoanalytic standard. But p2 cannot look brown to Sue as she is, or to us as we are, with our existing sensory endowment. So Yablo’s facsimile principle would appear to be too strong to be required for an adequate explanation of a seeming possibility. Part of what’s going on in this example is that the counterparts aren’t very seriously otherworldly but instead not very far off. Perhaps if the counterparts were seriously otherworldly, the explanation of the seeming possibility wouldn’t meet a plausible psychoanalytic standard. But if counterparts are not very far off, then the explanation might meet such a standard after all.23 Note, however, that this modification does not challenge Yablo’s claim that the seeming possibility of LME’s being heat cannot be explained away by a Kripkean facsimile account. As far as I know, there is no nearby heat-cold sensitivity abnormality, parallel

23 Yablo distinguishes Type 1 intuitions that pivot on a perceptually available x, to the effect that its hidden nature could have been y, from Type 2 intuitions that pivot on a hidden nature y, to the effect that it could have supported or underwritten perceptually available x. Versions of the Sue-Joe case can generate Kripkean explanations without facsimiles for both Type 1 and Type 2 seemingpossibility intuitions: Type 1: We can set up the case so that Sue’s intuitions pivot on a perceptually available x—brown— to the effect that its hidden nature could have been y, p2. She was thinking: “That very color, brown, which is perceptually available to me, might have had a different hidden nature, namely, p2. This is because in the scenario in which human beings all had the colorblind visual system, the hidden nature of brown would have been p2.” On this setup, the case seems to show that we don’t need a facsimile to explain away the seeming possibility, on the supposition that here a facsimile requires that, holding fixed Sue’s actual perceptual and cognitive endowment, that which looks brown to her can be p2. The explanation is: what Sue is thinking is that there is a not-very-far-off scenario in which that which looks brown to everyone is p2. But this is not really a scenario in which brown is p2 or, more cautiously, not a scenario in which a color identical to the one we call brown is p2. This explaining away seems to meet the psychoanalytic standard, despite: that which looks brown to Sue, holding fixed her actual perceptual and cognitive endowment, cannot be p2. Type 2: We could also set up the case so that Sue’s intuitions pivot on a hidden nature, p2—a spectral-reflectance profile—to the effect that it could have supported a perceptually available x: brown, instead of green. Sue was thinking: “p2, that hidden nature, could have been brown. This is because if we all had the colorblind visual system, then p2, that hidden nature, would be brown.” On this setup, the case also seems to show that we don’t need a facsimile to explain away the seeming possibility, on the supposition that here a facsimile requires that to Sue, given her perceptual and cognitive endowment, p2 could look brown. The explanation is: what Sue is thinking is that there is a not-very-far-off scenario in which p2 looks brown to everyone. But this is not really a scenario in which p2 is brown or, more cautiously, not a scenario in which p2 is a different color from the one it actually is. This explanation seems to meet the psychoanalytic standard, despite: spectral-reflectance profile p2 cannot look brown to Sue, holding fixed her perceptual and cognitive endowment.

79 Qualitative Inaccuracy and Recent Objections to Conceivability Arguments

to color blindness, that would allow this facsimile account to meet a plausible psychoanalytic standard. Yablo’s second criticism of Kripke’s strategy is directed against another “scientific” way of generating a seeming possibility, not by the intuition that LME could feel or could have felt like heat, but just by the brute intuition that it is possible, even in a world without perceivers, that heat be LME. Yablo notes that it also seems possible in this way that heat be something entirely alien; call it ABC (alien basis caliente). He then asks: although one might propose to explain the illusion in Kripke’s “facsimile” way, would the explanation be correct? I am not sure that it would, for the following reason. Our feeling that heat could have turned out to be something else is indifferent to whether the something else is alien ABC or actual LME. It would be very surprising if the feeling had two radically different explanations depending on the precise form of the something else. The LME form of the illusion cannot be explained by pointing to a possible facsimile of heat that is really LME. . . . Therefore the ABC form of the illusion ought not to be explained with a possible facsimile either.24 I agree with Yablo that, plausibly, the “scientific” intuition that heat’s being ABC is possible can no more be explained away by the possibility of its seeming to me, or someone sufficiently similar to me, that ABC feels like heat, than the “scientific” intuition that heat’s being LME is possible can be explained away by its seeming to me, or someone sufficiently similar to me, that LME feels like heat. So the task is to find an explanation of the seeming possibility of (1) heat ≠ HME that would suffice to explain away the “scientific” seeming possibilities of heat’s being LME and heat’s being ABC, and at the same time does not violate an appropriately liberalized version of the facsimile principle. I suggest that a salient feature of a phenomenal property such as phenomenal redness is that it appears to introspection as primitive, or else it appears in such a way as to occasion the belief that it is primitive. Part of what it is for a phenomenal property to be taken as primitive in either of these ways is for it to be taken as not identical to a more fundamental property whose qualitative nature or essence is not revealed in the ordinary introspective experience of phenomenal redness, and as not constituted by multiple more fundamental properties. Hence, because we take phenomenal redness to be primitive, it might well appear metaphysically possible that phenomenal redness exist without any such more fundamental properties, and that in this way it can stand on its own.

24 Stephen Yablo, “No Fool’s Cold,” p. 340.

80 Consciousness and the Prospects of Physicalism

For similar reasons, the apparent primitiveness of heat generates an explanation of why (1) heat ≠ HME seems possible. Our ordinary tactile sensations of heat represent it as a primitive secondary quality or are such as to occasion the belief that it is a primitive secondary quality. When we represent a property in this way, we represent it as if it is not identical to any physical property whose qualitative nature is not revealed in the sensory experience, or, alternatively, the way we represent it occasions a belief with this content. Accordingly, we believe heat not to be anything like HME. However, as a consequence, we take it that the existence of heat does not require anything like HME. So the seeming possibility of (1) is not to be explained away by the fact that I can represent LME (for instance) as feeling like heat to me or to someone sufficiently like me. Rather, it’s because we mistakenly take heat to be a property whose entire qualitative nature is revealed in the relevant sensory experiences, and thus as not being identical to anything like HME, that (1) seems possible. But why then does it seem possible that heat be LME or ABC instead, given the primitivist intuition about heat? One possibility (what I’d say is going on in my mind) is that we begin with the intuition that heat is a primitive property. It’s then difficult to see how heat might actually be HME, but once we’re told by the scientists that it is, or we become aware of the derivation of the heat observations from ‘heat = HME’ and Newtonian mechanics, we come to accept it. One available way to reconcile this new reductionist belief with the primitivist intuition—that is, without giving up the primitivist belief—is to think that heat is merely correlated with HME. We then imagine that given suitable alterations of the causal laws or the empirical facts, heat would be correlated with LME or ABC instead. So if heat “is” HME, then it’s possible that it “be” LME or ABC just as well. But when we understand what the truth of ‘heat = HME’ really comes to, and that identity is not just correlation, and we subsequently shed the intuition that heat is a primitive property, we come to believe that there may be a distinction between the real nature of heat and how it is represented by ordinary tactile sensation. Then while we perhaps formerly might have thought the concept of heat to be the concept of a property whose entire qualitative nature is revealed in our sensory experiences of heat, we can now see that the concept of heat allows for more possibilities, that it permits the reference of ‘heat’ to be fixed by the description “the typical cause of heat sensations,” where that cause is not primitive heat. With all of this firmly in mind, and understanding that HME is actually the typical cause of heat sensations, it ceases to seem metaphysically possible that heat be LME or ABC, although we may recognize that these are still conceptual or primary possibilities. So similarly, introspection of phenomenal properties has us taking them to be primitive phenomenal qualities and thus not constituted of more fundamental properties, and this explains why (2) pain ≠ C-fiber firing

81 Qualitative Inaccuracy and Recent Objections to Conceivability Arguments

seems possible to us. But now, by analogy with the case of heat, we might come to shed the intuition that phenomenal properties are primitive and to affirm instead that there may be a distinction between the real qualitative nature of pain and how it is represented introspectively. While we formerly thought that the concept of pain is the concept of a property whose entire qualitative nature is accurately represented by our introspective representation of it, we now come to believe that the concept of pain allows it to refer to different properties, in particular, that it allows the reference of ‘pain’ to be fixed by the description “the typical cause of our introspective representation of pain,” where that typical cause is not a primitive phenomenal property. Phenomenal concepts and secondary-quality concepts will then underwrite the same sort of explaining away of relevant seeming possibilities. Most important, just as the seeming possibility of (1) can be explained away by the apparent primitiveness of heat, so the seeming possibility of (2) can be explained away by the apparent primitiveness of pain. This is not the full story about how the seeming possibilities of (1) and (2) are to be explained away. With Stoljar, I endorsed the idea that the pertinent seeming possibilities’ not being genuine, or a failure of the conceivabilities under consideration to issue in possibilities, can be explained away partly by ignorance of relevant facts. But Yablo also notes that in general, ignorance produces epistemic possibilities, not seeming metaphysical possibilities.25 Yablo’s idea is compelling. So if ignorance in some particular case is to explain away seeming metaphysical possibility, then there must be something distinguishing about it. In response, about the seeming possibility of (1) heat ≠ HME and of (2) pain ≠ C-fiber firing we can now say this. Given the truth of the qualitative inaccuracy hypothesis, these seeming possibilities are undercut by the following: (i) due to the nature of our sensation and introspection, we take heat and pain to have qualitative natures that they in fact lack; (ii) more boldly, we take heat and pain to be primitive properties, a point which in turn has two components: that their entire qualitative natures are revealed in our sensory or introspective experiences of them, and that they are represented as not being constituted of more fundamental properties, while they are in fact not primitive properties; and (iii) we are ignorant in each case of such misrepresentation, and even if we were no longer ignorant of it, at least some of the force of the

25 In his presentations of “No Fool’s Cold” at the University of Vermont in the fall of 2006 and at the Pacific Division Meetings of the American Philosophical Association in the spring of 2007.

82 Consciousness and the Prospects of Physicalism

primitivist conception might well persist. Thus: when a property is represented as or believed to be primitive, it is taken not to be identical to a distinct, more fundamental property whose qualitative nature is not revealed in sensory or introspective experience of the property, and the state that has the property as its essence is represented as not requiring such more fundamental properties to exist. Supposing we are ignorant of the inaccuracy in our taking phenomenal properties to be primitive, it might well seem metaphysically possible to us that phenomenal states can exist without such more fundamental properties. Similarly, when Descartes represents his mind (i.e., himself) as not requiring extension to be a complete existing thing, it might well be that his taking mental properties as primitive is doing the work.26 If he is ignorant of the inaccuracy in his taking mental properties as primitive, this ignorance will explain why it appears possible to him that his mind, an entity characterized solely by such properties, exists without being spatially extended, while this is in fact impossible.27 Thus in response to Yablo, it is not simply ignorance that is to be cited to explain away the targeted seeming possibility, but rather ignorance of the fact that phenomenal properties are inaccurately taken to be primitive. When these properties are mistakenly taken to be primitive, the illusion is generated that it is metaphysically possible that states or substances characterized by them should exist without more fundamental properties, and ignorance of this misrepresentation allows the illusion to persist. Moreover, even if this ignorance were dispelled, the primitivist conception might well continue to exert its force, which would explain why even many of those persuaded of physicalism retain a dualist intuition.28

26 René Descartes, Meditations on First Philosophy, in The Philosophical Writings of Descartes, vol. 2, p. 54 (AT VII 78); “Fourth Replies,” in The Philosophical Writings of Descartes, vol. 2, pp. 154–62 (AT VII 219–31). 27 In “The Real Distinction between Mind and Body,” pp. 117–94, Yablo issues the following challenge to critics of Descartes’s claim that it is clearly and distinctly conceivable for me that I possess exactly my thought properties; that is, that I have a purely mental existence: come up with a proposition q such that q is true, if q then I am incapable of purely mental existence, and my ignorance of q’s truth explains my ability to conceive myself in a purely mental condition, on grounds independent of the supposition that I cannot exist in a purely mental condition. My suggestion for q is: I misrepresent my thought properties as primitive. My view does not quite fit Yablo’s schema, since I do not claim that q is true, only that it might well be. But if it is true, then it might well also be that I am incapable of purely mental existence, since mental properties might well then be physically composed instead; and the fact that I represent my thought properties as primitive and I am ignorant that when I do so I misrepresent them explains why I am able to conceive of myself in a purely mental condition, or why I think I am able to conceive this. This explanation is also independent of a supposition that I am incapable of purely mental existence. 28 Sydney Shoemaker, “On an Argument for Dualism,” in Identity, Cause, and Mind (Oxford: Oxford University Press, 2003), pp. 287–308.

83 Qualitative Inaccuracy and Recent Objections to Conceivability Arguments

FINAL WORDS The account that involves the open possibility that we introspectively represent phenomenal properties as having qualitative natures they actually lack, or else the stronger hypothesis that we misrepresent phenomenal properties as primitive, has several advantages over other ways of explaining away the seeming possibilities at issue. Over a Kripkean strategy, it has two advantages that Yablo cites: it does not attempt to explain away seeming possibility by way of a facsimile that fails to meet a credible psychoanalytical standard, and it provides a single explanation for the seeming possibility of (1) and (2). Over the Stalnaker, Hawthorne, and Braddon-Mitchell proposal, it does not attribute to the zombie-conceiver a presupposition that nonphysicalism, precisely the view in contention, is true. Their analysis seems mistaken because there is clearly something about our representation of the mental that motivates dualism. It’s implausible that people opt for dualism independently of considerations about how the mental is represented by us. On the account I propose, the seeming possibility at issue has its root precisely in how the phenomenal is represented: it is represented as having a qualitative nature that it might well lack, or more ambitiously, it is represented as primitive, or in such a way as to occasion the belief that it is primitive, while it might well not be. Finally, let me venture a brief remark about how the qualitative inaccuracy hypothesis might supplement the “phenomenal concepts” strategy. On Nagel’s version of the strategy, C-fiber firings are imagined perceptually: “we put ourselves in a conscious state resembling the state we would be in if we perceived it,” and pain is imagined sympathetically: “we put ourselves in a conscious state resembling the thing itself.” Because these two types of imagination are involved, ‘pain = C-fibre firing’ will seem contingent even if it is necessary. Hill’s variant has it that because our concept of C-fiber firing is theoretical and our concept of pain is phenomenological, and since between these two types of concepts there are no a priori ties, the identity claim will seem contingent.29 Yablo and Stoljar have argued that both Nagel’s and Hill’s strategies predict seeming possibilities where there are none. Yablo points out against Nagel that it does not seem possible that this rock is in pain, despite the rock’s being imagined perceptually and pain sympathetically; against Hill, Stoljar points out that it does not seem possible that if x is a number, then x is a red sensation, despite my concept ‘being a number’ being theoretical and my concept ‘being a sensation’ phenomenological.30 29 Thomas Nagel, “What Is It Like to Be a Bat?” Philosophical Review 83 (1974), pp. 435–50, note 11; Christopher Hill, “Imaginability, Conceivability, Possibility, and the Mind-Body Problem,” Philosophical Studies 87 (1997), pp. 61–85. 30 Stephen Yablo, “No Fool’s Cold,” pp. 328–29; Daniel Stoljar, “Physicalism and Phenomenal Concepts,” Mind and Language 20 (2005), pp. 469–94. Hill himself raises the objection that his strategy generates an implausibly general skepticism about conceivability evidence for modal claims; see “Imaginability, Conceivability, Possibility, and the Mind-Body Problem,” pp. 81–82. This is also a potential concern Yablo highlights for responses to conceivability arguments generally, for example, in “The Real Distinction between Mind and Body.” The diagnosis of the conceivability arguments I offer here would restrict such skepticism to conceivability evidence that involves a supposition of a primitive property, and so it would not result in modal skepticism that is significantly general.

84 Consciousness and the Prospects of Physicalism

If this strategy is to be reclaimed, there must be some other feature of phenomenal property representation that generates the seeming possibilities at issue. I propose that this additional feature is phenomenal properties’ being represented introspectively as having a qualitative nature that they lack or their being mistakenly represented as primitive properties, and that the plausibility of this suggestion rests on the case made in these first four chapters.

5 RUSSELLIAN MONISM I

While the qualitative inaccuracy hypothesis yields a coherent defense of physicalism against the knowledge and zombie arguments, and it is a serious open possibility, it is not clearly true. Prereflectively, most of us suppose that introspection does accurately represent the qualitative nature of phenomenal properties. Introspective modes of presentation of phenomenal properties represent them as having a specific qualitative nature, and we assume that the attribution of this nature to the phenomenal properties is correct. Correlatively, the contents of phenomenal concepts would be restricted to those that reflect such accurate introspective representation. Suppose one instead endorsed these assumptions. To emphasize the point, one might build them into the knowledge and conceivability arguments as premises. What prospects for physicalism would remain? Chalmers argues that an unconventional sort of physicalism might be developed that explains phenomenal properties, supposing this accuracy claim, by way of fundamentally intrinsic properties of the physical world. This idea has a complex history. In the early modern period, philosophers began to doubt whether we can make sense of the notion of mind-independent physical substance, and their misgivings gave rise to panpsychist and idealist views of reality. A distinctively Leibnizian version of this concern starts with the claim that physical properties are extrinsic, while none are intrinsic, at least in a fundamental sense of ‘intrinsic’. The core idea is that none of the physical properties to which our best physical theories refer is intrinsic in this fundamental sense, but at the same time these physical properties require grounding in fundamentally intrinsic properties. Leibniz concluded that physical properties have nonphysical intrinsic properties as a ground. In the context of his zombie argument, Chalmers develops this theme, but in an intriguing, less resolutely antiphysicalist way. Fundamentally intrinsic physical properties have indeed been proposed. The Aristotelians suggested prime materiality, and Locke and Newton advocated solidity. Chalmers envisions the possibility of physical (and also of nonphysical) properties of whose 85

86 Consciousness and the Prospects of Physicalism

nature we are currently ignorant in a significant respect and that have a dual foundational role: they not only ground the properties that current physical theory specifies but also are protophenomenal by virtue of explaining phenomenal properties. Because Bertrand Russell advocated a position of this general sort, Chalmers calls his view Russellian monism. (Restricting the fundamental properties to the protophenomenal ones would make the position strictly monist; Chalmers is clearly attracted to this view.) In this chapter and the next, we will examine the key ideas and arguments in this discussion, with an eye to formulating a more thorough characterization of Russellian monism and to testing the plausibility of this position. The plan is to begin by setting out a provisional account in terms of unknown categorical bases of dispositions and then to determine more precisely what it is that we might be ignorant about. We shall see that what is actually at issue is ignorance about properties that are intrinsic in a fundamental way. With the aid of Leibniz and Kant, I will characterize this claim in terms of the notion of an absolutely intrinsic property and propose a definition. In the next chapter, we will critically examine arguments by Kant and David Lewis for our lacking a significant sort of knowledge of such properties, whereupon I will suggest and endorse an alternative argument for ignorance of this kind, one from failure of abduction. The final task is to draw conclusions for an account of consciousness.

INTRODUCING RUSSELLIAN MONISM As noted in chapter 3, Chalmers contends that certain possibilities for categorical properties of physical states serve to mitigate the zombie argument’s antiphysicalist force. Here again is Chalmers’s formulation of the argument. ‘P’ is a statement that details the complete physical truth about the actual world; ‘T’ is a ‘that’s all there is’ statement, specifying that P describes a minimal P-world; and ‘Q’ is an arbitrarily selected actual phenomenal truth: (1) ‘PT and ~ Q’ is ideally, positively, primarily conceivable. (2) If ‘PT and ~ Q’ is ideally, positively, primarily conceivable, then ‘PT and ~ Q’ is primarily possible. (3) If ‘PT and ~ Q’ is primarily possible, then ‘PT and ~ Q’ is secondarily possible, or Russellian monism is true. (4) If ‘PT and ~ Q’ is secondarily possible, materialism is false. (5) Materialism is false or Russellian monism is true.1 Premise (3) and the conclusion (5) offer Russellian monism as a way of avoiding the antiphysicalist force of the argument. It turns out to be highly significant that there are more and less specific ways to characterize Russellian monism, but let us begin with a fairly general version that

1 David Chalmers, “Does Conceivability Entail Possibility?” pp. 145–200, at pp. 195–99; “Consciousness and Its Place in Nature.”

87 Russellian Monism I

foregrounds the distinction between dispositional and categorical properties.2 Fragility and flammability are paradigmatically dispositional, while shapes and sizes are often cited as nondispositional and categorical. The tenability of a sharp version of the distinction, on which no dispositional property is categorical and no categorical property is dispositional, is controversial. For example, C. B. Martin and John Heil contend that every property can be viewed as either categorical or dispositional, so in their conception the distinction is not sharp.3 But the sharp version of the distinction has many defenders, David Armstrong and Frank Jackson among them, and Troy Cross has of late provided a sophisticated defense of it against recent objections.4 Proponents of the sharp version of the distinction often also contend that all dispositional properties have categorical bases; that is, for any dispositional property of a thing, there are distinct properties that explain the thing’s having the dispositional property.5 As a simple model, take the tendency of a ball to roll when pushed to be a dispositional property, and its spherical shape a categorical property. The ball’s shape is a component of the explanation of why the ball has the tendency to roll and is thus part of the categorical basis of this dispositional property. But this type of claim about explanation does not depend on the viability of a sharp version of the dispositionalcategorical distinction. On Martin and Heil’s view in which all properties are at the same time dispositional and qualitative or categorical, the ball’s spherical shape counts as a dispositional property, or as a “vehicle of dispositionality,” while it also serves as an explanatory basis of the ball’s tendency to roll when pushed.6 So alternatively, this claim about explanation might instead be framed just in terms of tendencies and explanatory bases, where explanatory bases are themselves conceived as dispositional properties. We might well have two notions of a dispositional property: one on which the explanation of a thing’s having a dispositional property demands explanation by

2 Daniel Stoljar has also developed this position, in “Two Conceptions of the Physical,” Philosophy and Phenomenological Research 62 (2001), pp. 253–82, and in Ignorance and Imagination, pp. 106–22; see also Gregg Rosenberg, A Place for Consciousness (New York: Oxford University Press, 2004), pp. 13–30. 3 C. B. Martin, “On the Need for Properties: The Road to Pythagoreanism and Back,” Synthèse 112 (1997), pp. 193–221; C. B. Martin and John Heil, “The Ontological Turn,” Midwest Studies in Philosophy 23 (1999), pp. pp. 34–60; John Heil, “Dispositions,” Synthèse 144 (2005), pp. 343–56. 4 Frank Jackson, From Metaphysics to Ethics; David Armstrong, A Materialist Theory of Mind (London: Routledge, 1968); Troy Cross, “What Is a Disposition?” Synthèse 144 (2005), pp. 321–41. 5 Frank Jackson expresses this intuition when he argues against the view that the essences of properties, in general, are limited to their causal role: “This, to my way of thinking, is too close to holding that the nature of everything is relational cum causal, which makes a mystery of what it is that stands in the causal relations” (From Metaphysics to Ethics, p. 24); cf. David Armstrong, A Materialist Theory of Mind (London: Routledge, 1968); Elizabeth Prior, Robert Pargetter, and Frank Jackson, “Three Theses about Dispositions,” American Philosophical Quarterly 19 (1982), pp. 251–57; Elizabeth Prior, Dispositions (Aberdeen, Scotland: Aberdeen University Press, 1985); Michael Smith and Daniel Stoljar, “Global Response-Dependence and Noumenal Realism,” Monist 81 (1998), pp. 85–111; Simon Blackburn, “Filling in Space,” Analysis 50 (1990), pp. 62–65. 6 John Heil, “Dispositions,” p. 352.

88 Consciousness and the Prospects of Physicalism

virtue of a distinct property, and another on which a dispositional property includes the basis that explains the dispositional tendencies. Perhaps either way of speaking is fine, but what’s important to the discussion at hand is that it can legitimately be claimed that the ball has a property, its shape, which is an explanatory basis of its tendency to roll when pushed. Chalmers’s discussion, to which we now turn, retains the sharp categorical-dispositional distinction, but his contentions can be adapted to the alternative way of speaking. Chalmers argues that since primary conceivability is an a priori matter, features of physical properties available to a primary conception are restricted to features of their causal roles, which might be cast as dispositional properties. Significant information about the categorical bases of these causal role or dispositional properties is left out. This consideration gives rise to an alternative proposal for explaining away the apparent possibility of ‘P and ~ Q’ (I’ll omit the ‘T’ in this and subsequent formulations) and for explaining phenomenal consciousness: Here a loophole emerges: it is not clear that P has the same primary and secondary intension. It can reasonably be argued that physical concepts have their reference fixed by some dispositional role, but refer to an underlying categorical property. If so, their primary intensions pick out whatever plays a certain role in the world (irrespective of its categorical nature), while their secondary intensions pick out certain instances of a categorical property (irrespective of its role). If so, the purported ‘zombie world’ in which the primary intension of P and ~ Q holds may be a world in which the secondary intension of P is false, so we cannot infer the secondary possibility of P and ~ Q.7 Since ‘P’ details all of the physical information about the actual world, any scenario that is ideally primarily conceivable to which the primary intension of ‘P’ assigns “true” will preserve all of the actual-world physical causal-role or dispositional properties. Chalmers contends: “because the primary intension of P holds, this world must be structurally-dispositionally isomorphic to the actual world, with the same patterns of microphysical causal roles being played.”8 But this leaves out categorical properties. While a priori reflection on ‘P’ fixes all of the causal-role or dispositional properties designated by physical concepts, it does not determine categorical properties that underlie and explain them. So perhaps one can primarily conceive ‘P and ~ Q’ only because one is conceiving just causal-role or dispositional properties on the physical side, and it is an open possibility that if one were to replace ‘P’ with a more complete ‘P+’ that includes concepts that allow us to directly represent the currently unknown 7 David Chalmers, “Does Conceivability Entail Possibility?” p. 197. The primary intension of S assigns truth values to S in possible worlds considered as actual, while the secondary intension assigns truth values to S in possible worlds considered as counterfactual. For example, the primary intension of ‘water is XYZ’ assigns ‘false’ to this statement in the actual world and ‘true’ to it in a world in which the watery stuff is XYZ, while its secondary intension assigns ‘false’ in each world. 8 David Chalmers, “Does Conceivability Entail Possibility?” p. 197.

89 Russellian Monism I

or incompletely understood properties in their categorical basis, and the result, ‘P+ and ~ Q’ , would not be primarily conceivable. ‘Q’ would be a priori derivable from ‘P+,’ and P+ would explain phenomenal consciousness. In Chalmers’s proposal, these categorical properties are not directly specified by any concepts of current physics. In fact, we are at this point ignorant of the nature of these categorical properties. This ignorance creates epistemic space for speculative proposals about the nature of these categorical bases that might at the same time explain phenomenal consciousness.9 Thus, for physics to allow us to represent such categorical properties directly, new physical concepts would be required. The concepts added to ‘P’ to form ‘P+’ might be concepts of categorical phenomenal properties, perhaps giving rise to a panpsychist position, or else they might be concepts of categorical protophenomenal properties, not phenomenal but nonetheless explanatory of the phenomenal. In Chalmers’s conception, the protophenomenal option allows for physicalism, for even though no concepts in current physics specify protophenomenal properties directly, these properties might nonetheless be similar enough to those over which current physics quantifies to count as physical. Each of these positions is a version of Russellian monism. Provisionally, Russellian monism is any view that combines (1) categorical ignorance, the claim that physics, or at least current physics, leaves us ignorant of certain categorical bases of physical dispositional properties, with (2) consciousness- or experience-relevance, the proposal that these categorical properties have a significant role in explaining consciousness or experience.10

WHAT DOES PHYSICS LEAVE OUT? The idea that we lack knowledge of certain categorical bases of physical dispositional properties because physics does not specify these categorical bases directly has a long history, dating back to Kant, as we shall see. But first, Stoljar presents a version of a frequently-voiced contemporary argument for this view.11 Its structure is to catalogue the types of truths about physical objects to which we do have cognitive access and then to make plausible the conclusion that there must be a type of truth about categorical properties to which we would then lack cognitive access. Examining this argument will allow us to be more precise about what it might be that might be left out by physics.

9 Chalmers allows that our current physical concepts may refer to these categorical properties indirectly, but he thinks they do not specify or allow us to represent these properties directly: “there is a pretty strong discontinuity between our ordinary physical concepts (which pick out these [intrinsic] properties at best indirectly) and the sort of ‘physical’ concepts that would be required to represent these properties directly—so much so that these concepts seem different in kind from the concepts of physical theory, which are the concepts on which the argument turns” (from correspondence). 10 David Chalmers, “Does Conceivability Entail Possibility?” p. 198. 11 Daniel Stoljar, Ignorance and Imagination, pp. 106–22.

90 Consciousness and the Prospects of Physicalism

Stoljar notes that a claim typically first made in this context is one we’ve already encountered, that physics characterizes physical properties as dispositional, and leaves us ignorant of the categorical properties that would underlie these dispositions.12 This claim is sometimes grounded in a Humean metaphysics, according to which all the fundamental properties of the world are categorical, and physical dispositional properties are analyzed as regularities among instantiations of categorical properties (as is the case for modal properties in general, according to the Humean). But the explanatory demand at issue need not be motivated by a metaphysical view that reduces dispositional properties to categorical regularities of this sort. Even if dispositional properties are held not to be reducible in this way, for many there remains the intuition that they nevertheless must be explained by way of underlying nondispositional features. About which sort of categorical property might physics leave us significantly ignorant? Stoljar specifies three pertinent classes of truths to which we do not lack cognitive access: (i) spatiotemporal truths, that is, general truths about space and time, spatiotemporal position, and topology; (ii) truths about secondary qualities; and (iii) truths about primary qualities. Spatiotemporal truths are not the whole story, since they do not tell us anything about the objects that fill space. Secondary qualities are not a universal feature of physical objects: very small physical objects and very large objects like the universe don’t have them, so they are not the whole story either. Finally, there are the truths about primary qualities. In this context, a passage from Simon Blackburn is often cited: When we think of categorical grounds, we are apt to think of spatial configurations of things—hard, massy, shaped things, resisting penetration and displacement by others of their kind. But the categorical credentials of any item on this list are poor. Resistance is par excellence dispositional; extension is only of use, as Leibniz insisted, if there is some other property whose instancing defines the boundaries; hardness goes with resistance, and mass is knowable only by dynamical effects. Turn up the magnification and we find things like an electrical charge at a point, or rather varying over a region, but the magnitude of a field at a region is known only through its effects on other things in spatial relations to that region. A region with charge is very different from a region without. . . . It differs precisely in its dispositions or powers. But science finds only dispositions all the way down.13 12 John Hawthorne provides an instructive discussion of these issues in “Advice for Physicalists.” 13 Simon Blackburn, “Filling in Space,” pp. 62–63. For a similar view, see Thomas Holden, The Architecture of Matter, pp. 269–72: Scientific investigation into the nature of matter can only ever lead us to powers: to relational and dispositional properties. It cannot lead us to categorical or intrinsic properties, still less to their equally inscrutable ancestors, a quality-less substratum that stands behind all properties whatsoever. We never encounter a non-dispositional, non-relational, categorical property in the physicist’s material world. (p. 272)

91 Russellian Monism I

However, Blackburn’s reflections don’t show that all primary qualities are dispositional and that a primary quality conception leaves out categorical bases. Shapes and sizes are not dispositional properties, but rather (often) categorical, since they serve as explanatory bases for dispositions, and they are also properties that science finds. The sizes and shapes of molecules, for example, play a role in explaining the dispositional properties of molecules, and so these properties would count as categorical, and they are properties over which our physics and chemistry quantify. But note that even if shapes and sizes were properties of fundamental physical entities, they would seem to require a further property as a ground. Intuitively, a particle could not just have size and shape; rather, there would need to be some other feature that has or grounds its size and shape (but not everyone finds this convincing; Descartes doesn’t, for example). Another contrast is at work here—not just the distinction between dispositional and categorical properties, but perhaps a distinction between properties that are categorical and also intrinsic in some strong sense, and those that do not fall into both of these classes. This other contrast turns out to be very important for Chalmers’s view, since in the last analysis his positive proposals for Russellian monist theories of consciousness rest on the claim that physics specifies only structural and dynamical properties, and the missing properties, those that potentially explain consciousness, are categorical and in some sense intrinsic. In Chalmers’s conception, Russellian monism takes its inspiration from Bertrand Russell’s discussion of physics in The Analysis of Matter, according to which physics characterizes entities in its purview solely by their relations to one another and to us, while, crucially, it is silent on their intrinsic properties. Here is Russell: A piece of matter is a logical structure composed of events; the causal laws of the events concerned, and the abstract logical properties of their spatiotemporal relations, are more or less known, but their intrinsic character is not known. Percepts fit into the same causal scheme as physical events, and are not known to have any intrinsic character which physical events cannot have, since we do not know of any intrinsic character which could be incompatible with the logical properties that physics assigns to physical events. There is therefore no ground for supposing that percepts cannot be physical events, or for supposing that they are never compresent with other events.14 So the relevant unknown properties are not only categorical but also in some sense intrinsic. These are the ideas that have their roots in Leibnizian concerns about the very idea of matter, examination of which will yield a specific proposal for what it might be that physics leaves out. 14 Bertrand Russell, The Analysis of Matter (London: Kegan Paul, 1927), p. 384; see Galen Strawson’s exposition of Russell’s idea in Galen Strawson, “Real Materialism.”

92 Consciousness and the Prospects of Physicalism

LEIBNIZ AND THE DEMAND FOR ABSOLUTELY INTRINSIC PROPERTIES Intrinsic properties are nonrelational properties, and extrinsic properties are relational properties (although on some current classifications, these characterizations come apart).15 Leibniz contends that a conception of the physical world that does not include intrinsic properties of a certain fundamental sort is in an important sense incomplete.16 In his view, an examination of Descartes’ theory of matter leads us to see why this is so. Descartes maintains that the essence of matter—its single attribute—is extension in three spatial dimensions.17 Leibniz argues that this position is rationally unsatisfying for the reason that extension is in an important sense an extrinsic property and that any real thing cannot feature only properties that are extrinsic in this way, but must possess intrinsic properties, in a contrasting sense, as well: “there is no denomination so extrinsic that it does not have an intrinsic denomination at its basis. This is itself one of my important doctrines (kyriai doxai).”18 This claim suggests first of all that in Leibniz’s view properties can be more and less extrinsic. Plausibly, extrinsic properties can have intrinsic aspects. For example, being wise is an extrinsic property of Sophie since it involves a relation to a comparison class. But being wise also includes an intrinsic aspect—having a certain type and level of intelligence. Being wise is therefore a complex property that has at least one extrinsic and

15 There is a considerable literature on how to characterize intrinsic and extrinsic properties more exactly. For comprehensive discussions, see Lloyd Humberstone, “Intrinsic/Extrinsic,” Synthese 105 (1996), pp. 205–67; Brian Weatherson, “Intrinsic and Extrinsic Properties,” Stanford Encyclopedia of Philosophy. Jaegwon Kim proposed that an intrinsic property of X is one that X would have if it were the only concrete thing in the universe—if X would have it even if it were lonely; see Jaegwon Kim, “Psychophysical Supervenience,” Philosophical Studies 41 (1982), pp. 51–70. Intuitively, intrinsic properties satisfy this criterion, but as critics have pointed out, being the only concrete thing in the universe would not itself be an intrinsic property of whatever has it, but X’s having this property is compatible with X’s being lonely. David Lewis suggested that an intrinsic property of X is a property X has that is possessed by any duplicate of X; see David Lewis, “Extrinsic Properties,” Philosophical Studies 44 (1983), pp. 197–200; Rae Langton and David Lewis, “Defining ‘Intrinsic,’” Philosophy and Phenomenological Research 58 (1998), pp. 333–45. Again, intuitively, intrinsic properties pass this test, but being a duplicate of x is, or might seem to be, an extrinsic property, but any duplicate of X will possess it. Promising revisions to these proposals have more recently been advanced. But in a more skeptical vein, Michael Dunn and Lloyd Humberstone suggest that the best we can do is to specify that an intrinsic property of X is a property that X has in and of itself; see J. Michael Dunn, “Relevant Predication 2: Intrinsic Properties and Internal Relations,” Philosophical Studies 60 (1990), pp. 177–206; Lloyd Humberstone, “Intrinsic/Extrinsic.” A profitable discussion of the issues that lie ahead does not require these definitional issues to be resolved. 16 The material in this section is a revision of the account I set out in Derk Pereboom, “Kant’s Amphiboly,” Archiv für Geschichte der Philosophie, 73 (1991), pp. 50–70. 17 René Descartes, Principles of Philosophy, Part II, 1–22, in The Philosophical Writings of Descartes, pp. 223–32 (AT VIII, 40–52). 18 Leibniz to deVolder, April 1702, in G. W. Leibniz, Philosophical Papers and Letters, ed. L. E. Loemker (Dordrecht, The Netherlands: D. Reidel, 1969), (hereafter: Loemker), pp. 526–27; Gottfried Wilhelm Leibniz, G. W. Leibniz, Die philosophischen Schriften, ed. C. I. Gerhard, 7 vols. (Hildesheim, Germany: Olms, 1965), (hereafter: Gerhardt), p. 240.

93 Russellian Monism I

one intrinsic aspect; it is thus not thoroughly extrinsic. The limiting case is a property that is maximally or purely extrinsic: P is a purely extrinsic property of X just in case P is an extrinsic property of X and P has no intrinsic aspects. Being one among many is good candidate for a purely extrinsic property of, for instance, a point in space. To Leibniz’s charge against Descartes one might initially object that properties like having such and such an extension and being spherical are paradigmatically intrinsic properties of things. But Leibniz has in mind that a sphere’s extension is not intrinsic to it in a more fundamental sense.19 First, he maintains that there remains a respect in which the extension of a thing is extrinsic: Nor do I think that extension can be conceived in itself, but I consider it an analyzable and relative concept, for it can be resolved into plurality, continuity, and coexistence or the existence of parts at one and the same time.20 Leibniz proposes that the extension of the sphere can be analyzed as, or reduces to, the plurality, continuity, and coexistence of parts of the sphere. Properties of each of these three varieties are purely extrinsic properties of these parts. Being one of a collection of more than one thing, being continuous with other things, and coexisting with other things are all purely extrinsic properties of whatever has them. Thus it may be that P is an intrinsic property of X, while P is not in a sense fundamentally intrinsic to X, or, as James van Cleve points out, in Kant’s terminology, absolutely intrinsic to X.21 This occurs when X’s having P can be analyzed as, or reduces to, X’s parts having properties Q, R, S . . ., and these properties are purely extrinsic properties of these parts. We might say, then, that P is an absolutely intrinsic property of X just in case P is an intrinsic property of X, and X’s having P does not reduce to parts of X having purely extrinsic properties.22 19 Alyssa Ney makes this point in “Physicalism and Our Knowledge of Intrinsic Properties,” Australasian Journal of Philosophy 85 (2007), pp. 41–60, at p. 50. She also suggests that the next move to make is to define a more fundamental notion of intrinsic property. 20 Leibniz to De Volder, April 1699, Loemker, p. 516 = Gerhardt II, pp. 169–70. 21 Immanuel Kant, Critique of Pure Reason, A277/B333; the translations of passages from the Critique of Pure Reason in this and the next chapters are indebted to the translations by Norman Kemp Smith (London: Macmillan, 1929) and by Paul Guyer and Allen Wood (Cambridge: Cambridge University Press, 1987). James van Cleve, “Inner States and Outer Relations: Kant and the Case for Monadism,” in Doing Philosophy Historically, ed. Peter H. Hare (Buffalo, NY: Prometheus, 1988), pp. 231–47. 22 Slightly altered versions of these definitions accommodate a view like Peter van Inwagen’s, which for nonliving things countenances only the existence of simples. For instance, allowing X to be a plurality of existing things, and ‘entity’ to refer to such a plurality, P is an absolutely intrinsic property of X just in case P is an intrinsic property of X, and X’s having P does not reduce to entities that constitute X having purely extrinsic properties. Thanks to Andrew McGonigal for suggesting such more accommodating versions of these definitions.

94 Consciousness and the Prospects of Physicalism

By contrast, also in Kant’s terminology, P is a comparatively (or a relatively) intrinsic property of X just in case P is an intrinsic property of X, and X’s having P reduces to parts of X having purely extrinsic properties. Alternatively, the notions of absolutely and comparatively intrinsic properties might be expressed in terms of a priori derivability, which could be advantageous insofar as the notion of a priori derivability is clearer and less specifically ontologically committed than the notion of reduction or analysis in this context, and because these versions will be useful for negotiating the issues at hand with Chalmers: P is an absolutely intrinsic property of X just in case P is an intrinsic property of X, and the proposition that X has P is not a priori derivable from R, a proposition that details all and only the purely extrinsic properties of X’s parts. P is a comparatively intrinsic property of X just in case P is an intrinsic property of X, and the proposition that X has P is a priori derivable from R. Correlatively: P is an absolutely purely extrinsic property of X just in case P is an extrinsic property of X, and X’s having P reduces to proper parts of X having purely extrinsic properties, or if X has no proper parts, P is a purely extrinsic property of X, or else characterized in terms of a priori derivability: P is an absolutely purely extrinsic property of X just in case P is an extrinsic property of X, and the proposition that X has P is derivable a priori from S, a proposition that details all and only the purely extrinsic properties of X’s proper parts, or if X has no proper parts, P is a purely extrinsic property of X.23 23 James van Cleve, in his “Inner States and Outer Relations,” p. 235, proposes alternative definitions of the notions of comparatively and absolutely intrinsic properties, in accord with Kim’s notion of loneliness. First of all: P is a monadic property of X = df it is possible for something x to have P even if no individual distinct from x [i.e., not identical with x] exists; and, P is nonrelational = df it is possible for something x to have P even if no individual discrete from x [i.e., having no part in common with x] exists. He then characterizes absolutely intrinsic properties as nonrelational and monadic, and comparatively intrinsic properties as nonrelational but not monadic. Absolutely intrinisic properties of X are the intrinsic properties of X that X could have if it had no parts, or if the parts it does have did not

95 Russellian Monism I

So the extension of a sphere, if Leibniz is right about the reduction of the property of extension, turns out to be a comparatively intrinsic property of it. One might object that the Cartesian sphere’s extension does not reduce to parts of the sphere having purely extrinsic properties, for the reason that the parts have an intrinsic property that serves as the foundation for the extrinsic properties. But in the Cartesian theory of matter, the parts consist just in extension, and the extension of each of these parts is subject to the same reduction as the extension of the original body: the extension of each of these parts reduces to the plurality, continuity, and coexistence of its parts. The same reduction holds for the extension of the parts of these parts, on to infinity. In such a protracted analysis of the extension of the sphere, one never encounters anything other than purely extrinsic properties of parts. Next, Leibniz thinks that it is implausible that substantial things have only purely extrinsic properties: But it would appear from this that something must always be assumed which is continuous or diffused, such as the white in milk, the color, ductility, and weight in

exist, while the comparatively intrinsic properties of X are the other intrinsic properties of X. This is an interesting distinction, but I don’t think that it is the best one for illuminating Leibniz’s doctrine. The reason is that it has the consequence that every absolutely intrinsic property must be such that it can be possessed by a simple, partless entity. This is to add a requirement that should be controversial and not suggested by the intuition that underlies Leibniz’s doctrine. Imagine a multicolored sphere, a substance, in a world in which the colors of physical objects are primitive and intrinsic properties of things. One hemisphere of the sphere is all primitively green, the other hemisphere is all primitively blue. The sphere’s property of being primitively multicolored is analyzable into, or reduces to, intrinsic properties of two of its parts, namely, one hemisphere’s being primitively blue, and the other’s being primitively green. The parts of the primitively blue hemisphere are primitively blue, on to fundamental parts or infinity, and likewise, mutatis mutandis, for the primitively green hemisphere. Thus the sphere’s being multicolored does not reduce to purely extrinsic properties of its parts. On my proposed characterization, the sphere’s being primitively multicolored is an absolutely intrinsic property of it, since it does not reduce to purely extrinsic properties of parts of the sphere. By van Cleve’s definition, by contrast, the sphere’s being primitively multicolored is merely comparatively and not absolutely intrinsic because it is not a monadic property—the sphere could not be primitively multicolored if it had no parts distinct from the sphere itself. But would we want to say that the sphere’s property of being primitively multicolored is merely comparatively and not absolutely intrinsic, on the supposition that the notion of being comparatively intrinsic gets its cachet from the intuition that motivates Leibniz’s doctrine that extrinsic properties require absolutely intrinsic properties as a ground? The answer, I think, is no, for this intuition does not recommend that the sphere’s being primitively multicolored reducing to one part’s being primitively blue and the other’s being primitively green requires that the sphere have a further intrinsic property (or properties) as a ground for its being multicolored. More generally, there may be intrinsic properties of things that they could not have if they had no parts, but do not reduce to purely extrinsic properties of their parts, and are rather underlain by intrinsic properties of their parts, on to fundamental parts or to infinity. The content of Leibniz’s intuition does not counsel that for a thing that has such intrinsic properties to be substantial, further intrinsic properties are required. Rather, its point is that the substantiality of a thing demands that its thoroughly relational properties be accompanied by properties that are deeply intrinsic.

96 Consciousness and the Prospects of Physicalism

gold, and resistance in matter. For by itself, continuity (for extension is nothing but simultaneous continuity) no more constitutes substance than does multitude or number, where something is necessary to be numbered, repeated, and continued.24 There is a compelling thought here that is difficult to make precise. It might be put in this way: there must be some absolutely intrinsic property that confers, in some intuitive sense, substantive character on any substantial entity—one might call a property of this sort a substantival absolutely intrinsic property. In this passage, Leibniz represents the absolutely intrinsic property as that which has extension, in the sense that it is that which is continuous. Extension itself cannot have this substantival role, he thinks. Leibniz’s positive proposal is to ascribe force to matter as the missing property; the passage continues: “So I believe that our thinking is completed and ended in the concept of force rather than in that of extension. And we need seek no other concept of power or force than that it is the attribute from which change arises, and whose subject is substance itself.”25 Thus to avoid the result that matter has only purely extrinsic and comparatively intrinsic properties, Leibniz aims to find a feature that resists reduction to purely extrinsic properties of parts. But is force adequate to this role? Consider gravitational force, for instance. The gravitational force exerted by a sphere on another body is a function of the gravitational force exerted by its parts, but it is not obviously reducible to purely extrinsic properties of its parts. So one possibility is that there are properties of type T that are in some sense intrinsic to material thing X, and while X has P by virtue of its parts having certain properties, X has P by virtue of its parts having properties precisely of type T itself, and these properties are intrinsic to these parts. Furthermore, these parts have these properties by virtue of their parts having intrinsic properties of type T, ad infinitum. If force meets this condition, then material things’ having force will be an absolutely intrinsic property of them. It is important to note that, as the previous reasoning shows, force can be an absolutely intrinsic property even if there is no fundamental level, and thus no fundamental entity has force. This result is accommodated by this notion as I have characterized it. This is a welcome result, for the Leibnizian principle at issue, which I will provisionally formulate as follows: (Intrinsicness Principle, first pass) Any substantial entity must have at least one substantival absolutely intrinsic property, does not depend for its truth or plausibility on there being a fundamental level of reality—although Leibniz did believe for other reasons that there must be one.26 24 Leibniz to De Volder, April 1699, Loemker, p. 516 = Gerhardt II, p. 170; cf. G. W. Leibniz, Specimen Dynamicum, Loemker, pp. 435–52 = G. W. Leibniz, Mathemathische Schriften, ed. C. I. Gerhardt (Berlin and Halle, 1849–56), VI, pp. 234–54. 25 Cf. G. W. Leibniz, Specimen Dynamicum, Loemker, p. 445 = G. W. Leibniz, Mathematische Schriften, VI, p. 246. 26 G. W. Leibniz, “On Nature Itself,” Loemker, pp. 498–508 = Gerhardt IV, pp. 504–16; Jonathan Schaffer, “Is There a Fundamental Level?” Noûs 37 (2003), pp. 498–517.

97 Russellian Monism I

Significantly, Leibniz maintains that physical force is not an absolutely intrinsic property of a material substance. He calls physical force derivative, and he suggests that it is the phenomenal appearance of primitive force, which is an intrinsic mental property of a nonphysical monad. Primitive force is a law-governed tendency of a monad to pass from one perception to another.27 For Leibniz, the underlying ground of primitive force is found in the representational states of the monad, and it is these nonphysical representational states that provide the missing absolutely intrinsic properties. This account features no absolutely intrinsic physical properties. For Leibniz, this is part of the explanation for why physical things are not substantial or real in the fundamental sense and instead merely well-founded phenomena (phenomena bene fundata). The fact that derivative force has an appropriate foundation in absolutely intrinsic properties of a monad nevertheless allows physical things to be substantial in the lower-grade sense in which they are real, as well-founded phenomena. This story is of particular interest given our topic, for this is the first time we see an explicit formulation of the view that the fundamental intrinsic properties of the ultimately real world are mental. (Berkeley also held this view, but did not formulate it quite so explicitly.) Why did Leibniz advocate this position? Here is Kant’s diagnosis of Leibniz on this issue—read ‘absolutely intrinsic’ for ‘intrinsic’: As object of pure understanding, on the other hand, every substance must have intrinsic determinations and powers which pertain to its intrinsic reality. But what intrinsic accidents can I entertain in thought, save only those which my inner sense presents to me? They must be something which is either itself a thinking or analogous to thinking. For this reason Leibniz, regarding substances as noumena, took away from them, by the manner in which he conceived them, whatever might signify extrinsic relation, including also, therefore, composition, and so made them all, even the constituents of matter, simple subjects with powers of representation—in a word, MONADS.28 Kant’s thought is that the only absolutely intrinsic properties we can conceive are mental and that this is the source of Leibnizian idealism.

PERFECT SOLIDITY AS AN ABSOLUTELY INTRINSIC PROPERTY Locke and Newton disagree—they claim, in effect, to conceive of a substantival absolutely intrinsic physical property. Locke and Newton were corpuscularians, that is, atomists. The ancient atomistic hypothesis is that matter consists of physical entities without parts, that is, atoms.29 An atom’s having an intrinsic physical property would 27 G. W. Leibniz, Gerhardt II, p. 275. 28 Immanuel Kant, Critique of Pure Reason, A265–66/B321–22; cf. A274/B330, A283–84/B339–40. 29 See, for example, Lucretius, De Rerum Natura, translated as The Nature of Things, tr. A. E. Stallings (New York: Penguin, 2007).

98 Consciousness and the Prospects of Physicalism

not reduce to its parts’ having purely extrinsic properties, since atoms, by definition, have no parts. The shapes and sizes of atoms are thus absolutely intrinsic to them. But on Locke’s view, shape and size could not be substantival absolutely intrinsic properties, for we can form ideas of vacua with determinate shapes and sizes, and vacua are insubstantial. Rather, Locke and Newton in effect propose solidity as the absolutely intrinsic property of matter.30 By Locke’s characterization, solidity is “that which thus hinders the approach of two bodies when they are moving one towards another.” We acquire the idea of solidity from touch, from our tactile sense of resistance, but our idea of solidity outstrips any sensation we have of it: This resistance, whereby it keeps other bodies out of the space which it possesses, is so great, that no force, how great soever, can surmount it. All the bodies in the world, pressing a drop of water on all sides, will never be able to overcome the resistance which it will make, as soft as it is, to their approaching one another, till it be removed out of their way: whereby our idea of solidity is distinguished both from pure space, which is capable neither of resistance nor motion, and from the ordinary idea of hardness.31 One might object that Locke characterizes solidity dispositionally, as a tendency to keep other bodies out of a part of space, and this rules out its being absolutely intrinsic. Indeed, dispositions are often are cast as extrinsic properties, as mere relations to certain effects given certain circumstances, and the concern is that this is how Locke conceives of solidity.32 But while in Locke’s exposition solidity does seem at times to be described as an extrinsically conceived disposition to resist intrusion absolutely, at other times, as Thomas Holden argues, he intends solidity to be the categorical basis of such a disposition.33 It is then “that which hinders,” by contrast with the tendency to hinder—the vehicle of the dispositional tendency, rather than the tendency itself.34 This categorical solidity is the proposed substantival absolutely intrinsic property. However, all we have said by way of characterizing this property is that it is the categorical or underlying explanatory basis of absolute impenetrability. Do we have a conception of solidity that specifies its categorical nature directly and not merely as that which explains manifestations of absolute impenetrability? If we did, this would enhance the plausibility of the actuality of such a categorical feature. A perennial and relevant concern is that more generally we are acquainted only with dispositions in the extrinsic sense and not with underlying categorical bases, and this would also be the case for solidity. Consider Hume’s attack on the idea of causal power, the idea of an

30 Michael Ayers reports that Locke and Newton jointly proposed the solidity hypothesis, in his Locke (London: Routledge, 1991), vol. 2, Ontology, p. 59. 31 John Locke, Essay Concerning Human Understanding, II, iv, 2. 32 See, for example, Jennifer McKitrick, “The Bare Metaphysical Possibility of Bare Dispositions,” Philosophy and Phenomenological Research 66 (2003), pp. 349–69. 33 Thomas Holden, The Architecture of Matter, pp. 259–60. 34 John Heil, “Dispositions,” p. 352.

99 Russellian Monism I

intrinsic property that is at the same time dispositional in the sense that it is an aptness to produce effects, as Shoemaker specifies, where that aptness is not reducible to a feature of the world that is not itself an aptness of this sort, such as a regularity or counterfactual dependency among nondispositional property instances.35 Hume’s discussion suggests that such an idea might in some sense be difficult to grasp or express, whether the intrinsic dispositional property at issue is intuitively physical or mental. Still, this does not convince all of us that he is right to think that this idea of a causal power is not acceptable as it stands. Many of us think of our desires as tendencies to act, but also as involving an intrinsic component that is explanatory of these tendencies, and categorical in this sense. Perhaps the phenomenological feel of a desire directly acquaints us with such an explanatory component. Isolating it seems neither more nor less difficult than distinguishing the tendency-explaining or categorical component of solidity.36 Do we have any less reason to believe that we have a conception of categorical solidity than we have of, say, categorical shape? The shapes of objects have a dispositional component, but we also have a categorical conception of shape, which can be expressed mathematically, for example. Perhaps our conception of categorical solidity is no worse off. This property is not mathematically expressible in the sense that categorical shape is, but this is only because it is not a geometrical quality, and it is not scalar, since it does not admit of multiple distinct magnitudes, either extensive or intensive. And this categorical solidity, on the Locke-Newton proposal, is the substantival absolutely intrinsic property of the material world. We think of ordinary objects as solid, but physics tells us that the particles that constitute ordinary objects are noncontiguous. Perhaps if there are empty spaces among the particles in a cluster of a million, the cluster does not count as solid on the Lockean conception. Thus it may turn out that the ordinary objects that we think of as solid are

35 Sydney Shoemaker, “Causality and Properties,” in Time and Change, ed. P. van Inwagen (Dordrecht, The Netherlands: D. Reidel, 1980), pp. 109-35, and in conversation; a non-Humean theory of causality is developed by Rom Harré and E. H. Madden in Causal Powers (Oxford: Blackwell, 1975). In their view, to ascribe a causal power to do A to X is to claim that “X (will)/(can) do A, in the appropriate conditions, in virtue of its intrinsic character” (p. 86); see also Eric Watkins’s discussion in Kant and the Metaphysics of Causality, pp. 390–400. 36 Langton is arguably reflecting a Humean sensibility when she remarks about the suggestion that solidity is the intrinsic property that explains the dispositional property of impenetrability: “How does the positing of solidity help to explain anything? What more could there be to know?”; see Kantian Humility, p. 176. Holden expresses a similar view in The Architecture of Matter, p. 272, as does Ney in “Physicalism and Our Knowledge of Intrinsic Properties,” p. 55. By contrast, Kant holds that the absolutely intrinsic properties are the fundamental causal powers. The decisive passage is in the Critique of Pure Reason at A277–78/B333–34, which I quote and discuss in chapter 6, pp. 106-7 Schematically, for Humeans absolutely intrinsic properties are not causal powers, and to the extent that there are such powers, they are reducible to regularities or counterfactual dependencies among instances of such properties. On the contrasting conception of Kant, Locke, and Shoemaker, causal powers are fundamental and irreducible, and the absolutely intrinsic properties, at least the substantival ones, are such fundamental causal powers.

100 Consciousness and the Prospects of Physicalism

at best only imperfectly solid.37 But Locke’s idea is that of perfect solidity—a notion of solidity that (among other things) precludes intervening spaces. Our finding about ordinary objects would not rule out the possibility that matter is constituted of particles that are perfectly solid. The Locke-Newton proposal would then be that the absolutely intrinsic property of matter is categorical perfect solidity, and this is not ruled out by our finding about ordinary solid objects. Note also that a thing can be absolutely intrinsically perfectly solid even if it has parts. A perfectly solid composed entity will not be perfectly solid just by virtue of its parts standing in certain relations. The parts must also have perfect solidity, intuitively an intrinsic property of each of these parts. Thus one need not be an atomist to admit perfect solidity as an absolutely intrinsic physical property. One might consistently hold that certain physical things have perfect solidity as an absolutely intrinsic property and that all physical things are infinitely divisible.

KANT’S SKEPTICAL PROPOSAL Kant denies that we have knowledge or cognition (Erkenntnis) of any absolutely intrinsic properties of material things:38 All that we cognize in matter is nothing but relations (lauter Verhältnisse). What we call the intrinsic determinations of it are intrinsic only in a comparative sense (nur komparativ innerlich), but among these relations some are self-subsistent and permanent, and through these we are given a determinate object.39 In material things we find comparatively intrinsic properties but no absolutely intrinsic properties. This is not merely an epistemic claim. Kant contends that all properties of matter, substantia phaenomenon, even its apparently intrinsic properties, are absolutely purely extrinsic: “It is quite otherwise with a substantia phaenomenon in space; its intrinsic determinations are nothing but mere relations, and it itself is entirely made up of mere relations” (but this is consistent with some of these relations being “self-subsistent and permanent”).40 In the subsequent sentence, Kant mentions force 37 The perfect-imperfect terminology derives from David Chalmers, “Perception and the Fall from Eden”; see also chapter 2. 38 The material in this section is a revision of the account I develop in Derk Pereboom, “Kant’s Amphiboly.” 39 Immanuel Kant, Critique of Pure Reason, A285/B341. In a similar vein, David Armstrong writes: “If we look at the properties of physical objects that physicists are prepared to allow them such as mass, electric charge, or momentum, these show a distressing tendency to dissolve into relations that one object has to another;” see A Materialist Theory of Mind (London: Routledge, 1968), pp. 74–75. 40 Immanuel Kant, Critique of Pure Reason, A265/B321; cf. Metaphysical Foundations of Natural Science, tr. Michael Friedman (Cambridge: Cambridge University Press, 2004), AkB IV, p. 543. See Thomas Holden’s exposition of Kant’s position, and also of Roger Boscovich’s similar theory, in The Architecture of Matter, pp. 236–63.

101 Russellian Monism I

as a feature of matter: “We are acquainted with substance in space only through forces which are active in this and that space, either bringing objects to it (attraction), or preventing them penetrating into it (repulsion and impenetrability),” so for him force is in the last analysis an extrinsic property of material things.41 In Kant’s view, force is ultimately an extrinsic property because it is a relation among material items or, more abstractly, spatial points. The section on dynamics in Kant’s Metaphysical Foundations of Natural Science specifies two kinds of force: attractive and repulsive. Attractive force is by definition the cause by which two points approach one another, and repulsive force is by definition the cause by which two points recede from another.42 (Alternatively, Kant might be interpreted here as claiming that force is dispositional, and relational for that reason.) Kant admits that there is something unintuitive about his view that all of the properties of matter are relational: “It is certainly startling to hear that a thing is to be taken as consisting wholly of relations.”43 However, this apparent implausibility can be explained away: “Such a thing is, however, mere appearance, and cannot be thought through pure categories: what it itself consists in is the mere relation of something in general to the senses.”44 Because matter is mere appearance, it need not have any physical absolutely intrinsic properties. If matter were not merely appearance, but a thing in itself, then it would possess such absolutely intrinsic properties. In making these claims Kant indicates that he does not thoroughly reject the Leibnizian doctrine that intrinsic properties must ground extrinsic properties. If he rejected it, he would not feel the need to explain the plausibility of matter’s having only purely extrinsic and comparatively intrinsic properties by declaring that it is only appearance. What Kant accepts is that the extrinsic properties of mind-independently real substantial entities—things in themselves—must be grounded in absolutely intrinsic properties. This suggests a revised statement of the Intrinsicness Principle: (Intrinsicness Principle, final version) Any mind-independently real substantial entity must have at least one substantival absolutely intrinsic property, which I think best captures the metaphysical intuition that drives the positions we are discussing.

41 Immanuel Kant, Critique of Pure Reason, A265/B321. 42 Immanuel Kant, Metaphysical Foundations of Natural Science, Ak IV, pp. 498–91. 43 Immanuel Kant, Critique of Pure Reason, A285/B341; this passage conflicts with Thomas Holden’s claim (The Architecture of Matter, p. 261) that Kant was unmoved by the idea that matter must fill space by virtue of an intrinsic property. 44 Immanuel Kant, Critique of Pure Reason, A285/B341.

6 RUSSELLIAN MONISM II

In the previous chapter, I provisionally characterized Russellian monism as any view that combines (1) categorical ignorance, the claim that physics, or at least current physics, leaves us ignorant of certain categorical bases of physical dispositional properties, with (2) consciousness- or experience-relevance, the proposal that these categorical properties have a significant role in explaining consciousness or experience. Then with the aid of Leibniz and Kant, I proposed that what physics or current physics would leave us ignorant about are absolutely intrinsic properties: P is an absolutely intrinsic property of X just in case P is an intrinsic property of X, and X’s having P does not reduce to parts of X having purely extrinsic properties. Alternatively, formulated in terms of a priori derivability: P is an absolutely intrinsic property of X just in case P is an intrinsic property of X, and the proposition that X has P is not a priori derivable from R, a proposition that details all and only the purely extrinsic properties of X’s parts. What I called Leibniz’s Intrinsicness Principle is that every substantial entity has at least one absolutely intrinsic property and, more exactly, at least one substantival absolutely intrinsic property. What physics or current physics would leave us ignorant about is specifically which absolutely intrinsic properties, in particular which substantival ones, are actually instantiated. We will now critically examine the arguments for this ignorance claim.

102

103 Russellian Monism II

ARGUING FOR IGNORANCE ABOUT ABSOLUTELY INTRINSIC PROPERTIES

Ignorance from Failure of Deducibility Kant argues for Humility, the doctrine that we are significantly ignorant about which absolutely intrinsic properties our world features. The contention is not that we grasp all of the candidates, but just don’t know which of these understood properties are instantiated. Rather, there might well be candidates we do not comprehend and are even incapable of understanding. Rae Langton has developed an influential interpretation of Kant’s argument for Humility,1 and here is a version adapted from van Cleve’s revision of her interpretation. This adaptation substitutes ‘absolutely intrinsic’ for ‘intrinsic.’ The version Langton and van Cleve consider concludes ignorance about which intrinsic property simpliciter an arbitrary object instantiates. On my reading, this is not the conclusion Kant had in mind, nor is it one he would endorse, given that we have knowledge of shapes and sizes of bodies, which in his view are intrinsic—albeit merely comparatively intrinsic—properties of them. Here is the amended argument: 1. Receptivity: Human knowledge depends on sensibility, and sensibility is receptive: we can have knowledge of an object only insofar as it causes us to be in some state. 2. We know that A is F only if (i) A causes us to be in some state and (ii) from the fact that A’s causing us to be in that state, we can deduce that A is F. 3. If being F is an absolutely intrinsic property of A, we cannot make the required deduction: we cannot deduce what absolutely intrinsic properties things have from the fact that they cause certain effects in us. 4. Therefore, [Humility], if being F is an absolutely intrinsic property of A, we can never know that A is F.2 Van Cleve challenges this argument (actually, the one that concludes ignorance of all intrinsic properties) on the ground that the deducibility condition on knowledge it assumes (premise 2) is highly controversial and, in effect, too strict.3 It should not be that, in order to know that A is F, the possibility of all other options must be ruled out,

1 Rae Langton, Kantian Humility (Oxford: Oxford University Press, 1998); for interpretations of this material in Kant, see James van Cleve, “Inner States and Outer Relations: Kant and the Case for Monadism,” and Derk Pereboom, “Is Kant’s Transcendental Philosophy Inconsistent?” History of Philosophy Quarterly 8 (1991), pp. 357–71, and “Kant’s Amphiboly.” 2 James van Cleve, “Receptivity and Our Knowledge of Intrinsic Properties,” Philosophy and Phenomenological Research 65 (2002), pp. 218–37, at pp. 220 and 226. 3 One might doubt that Kant affirms a general deducibility condition on knowledge. Perhaps the discussion of empirical idealism in the Fourth Paralogism in A indicates that he does: Critique of Pure Reason, A366–A380. Still, he seems to allow for empirical knowledge justified by abductive argument, e.g., B274. One hypothesis is that for Kant a deducibility requirement holds for a priori knowledge but not for empirical knowledge.

104 Consciousness and the Prospects of Physicalism

a requirement presupposed by the claim that A is F be deducible from what we know. Van Cleve suggests that the stricture in question be counterfactual rather than deductive; for example, it might be sensitivity: if A weren’t F, we wouldn’t believe that it was F, or safety: we would believe that A is F only if it were F. He goes on to argue that supposing the deductive requirement, we would have knowledge of very little, including the relations that physical objects have to one another, since we cannot deduce the real natures of these relations from their effects on us. A plausible argument for ignorance about the properties in question cannot assume such a deducibility requirement on knowledge. David Lewis advances a related argument for this sort of ignorance.4 He has us imagine a final theory of the world that incorporates not only all of the actually instantiated causal-role properties but also all the categorical properties that underlie them, which would include the absolutely intrinsic properties of the fundamental entities. The theory leaves out the idlers, that is, the instantiated properties that have no causes or effects, and also the aliens, fundamental but uninstantiated properties. The theory is expressed in language T, which contains theoretical terms implicitly defined in T. The rest of our language is our old language, O, which is rich enough to express all of our observations (and much richer than that). Consider now the Ramsey sentence of T, which replaces all of the referring terms in T by existentially quantified variables.5 T implies all of the true O-language sentences. In addition, the Ramsey sentence says

4 David Lewis, “Ramseyan Humility”; see also the discussions of this argument by Rae Langton in “Elusive Knowledge of Things in Themselves,” Australasian Journal of Philosophy 82 (2004), pp. 129–36; by Jonathan Schaffer, “Quiddistic Knowledge,” in Lewisian Themes, ed. Frank Jackson and Graham Priest (Oxford: Oxford University Press, 2004), pp. 210–30; and by Dustin Locke, “A Partial Defense of Ramseyan Humility,” in Conceptual Analysis and Philosophical Naturalism, pp. 223–41. 5 James Ladyman provides this characterization of the Ramsey sentence: Ramsey’s method allows the elimination of theoretical terms from a theory by replacing them with existentially quantified predicate variables. . . . If one replaces the conjunction of assertions of a first-order theory with its Ramsey sentence, the observational consequences of the theory are carried over, but direct reference to unobservables is eliminated. If we formalize a theory in a first-order language: ∏(O1, . . . ,On; T1,  . . . ,Tm), where the Os are the observational terms and the Ts are the theoretical terms, then the corresponding Ramsey sentence is ∃t1,  . . .  , tm∏(O1,  . . . ,On; t1,  . . . , tm). Thus the Ramsey sentence only asserts that there are some objects, properties and relations that have certain logical features, satisfying certain implicit definitions. ( James Ladyman, “Structural Realism,” The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, at http://plato.stanford.edu/entries/structural-realism) Ramsey’s exposition of 1929 can be found in his “Theories,” in Frank P. Ramsey, The Foundations of Mathematics and Other Logical Essays, ed. R. B. Braithwaite (Paterson, NJ: Littlefield and Adams, 2001), pp. 212–36. Grover Maxwell, in the early 1960s, advocated the application of structuralism in its Ramsified form to scientific theories generally; see “The Ontological Status of Theoretical Entities,” Minnesota Studies in the Philosophy of Science 3 (1962), pp. 3–14; cf. “Structural Realism and the Meaning of Theoretical Terms,” Minnesota Studies in the Philosophy of Science 4 (1970), pp. 181–92.

105 Russellian Monism II

that T has at least one realization. It is thus open that it has more than one. Lewis thinks that it is possible for any underlying categorical property P1 to be switched with a distinct categorical property P2, in the sense that P2 has the exact causal role of P1. The result would be a theory that also implies all of the true O-language sentences. Sentences expressing our observations won’t determine whether it is P1 or P2 that actually has the causal role at issue. So for all we know, the Ramsey sentence has more than one realization, and if it does, our observations won’t tell us which is actual, and we will lack knowledge of which is actual. We therefore do not know which underlying categorical properties are actual, and this would include which absolutely intrinsic properties of fundamental entities are actual. But as Hawthorne, Langton, and Jonathan Schaffer point out, the requirement for knowledge assumed by this argument also seems too strict.6 Knowledge plausibly shouldn’t require the ruling out of all possibilities, only the salient ones—as Lewis himself argues.7 So Lewis’s ignorance claim also requires further substantiation. In addition, Lewis’s controversial metaphysics of properties makes switching especially easy. For him, different fundamental properties could have had the same causal role; he calls properties of that satisfy this description ‘quiddities.’8 But suppose, following Shoemaker, that we denied quiddities and instead endorsed a causal structuralist view of properties, according to which the causal role of a property constitutes its individual essence, so that if P1 and P2 have the same complete causal role, they are ipso facto the same property.9 Switching fundamental properties while not altering the evidence would then not be as clear a possibility. Still, as Hawthorne and Cross suggest, even given causal structuralism, ignorance is not foreclosed. The causal roles of two fundamental properties might differ only in virtue of possible but uninstantiated properties (aliens). We could then switch the two, and as long as the aliens weren’t also instantiated, the evidence would not be affected.10 Or perhaps more relevant to the issue at hand, the causal roles of two fundamental properties might differ only in virtue of properties that are instantiated but any effects of which are inaccessible to us because of our cognitive and technological limitations. As a result, we could switch the two, and the evidence available to us would not be affected. Moreover, even if a description of an accessible causal role singled out an absolutely intrinsic property of a fundamental entity, this description alone would not directly specify the complete nature of this property.11 By analogy, while a description of the 6 John Hawthorne, “Causal Structuralism,” Philosophical Perspectives 15 (2001), pp. 361–78, at p. 366; Rae Langton, “Elusive Knowledge of Things in Themselves”; Jonathan Schaffer, “Quiddistic Knowledge,” pp. 225–28. See also Alyssa Ney, “Physicalism and Our Knowledge of Intrinsic Properties,” pp. 51–53. 7 David Lewis, “Elusive Knowledge,” Australasian Journal of Philosophy 74 (1996), pp. 549–67. 8 See Jonathan Schaffer, “Quiddistic Knowledge,” p. 210, for a discussion of this term. 9 Sydney Shoemaker, “Causality and Properties.” 10 John Hawthorne, “Causal Structuralism,” pp. 366–37; Troy Cross, The Nature of Fundamental Properties, Dissertation, Rutgers University, 2004. 11 See the discussion of this view in chapter 2; cf. Colin McGinn, “What Constitutes the MindBody Problem,” p. 11.

106 Consciousness and the Prospects of Physicalism

causal role of the ball’s particular roundness and rigidity might well single out these intrinsic properties, this description would not directly specify their complete nature. If we understood only the causal role of these intrinsic properties, we might fix reference to these properties, but we would not thereby have grasped the complete nature of these properties themselves. Plausibly, then, a significant aspect of the nature of an absolutely intrinsic property cannot be directly specified just by descriptions of causal relations. Consequently, even if causal role descriptions singled out absolutely intrinsic properties, significant ignorance of the nature of these properties would remain in play. But so far we have rejected arguments for ignorance that specify a deducibility requirement on knowledge, or those that demand ruling out of all alternative possibilities. What are the other options?

Ignorance from Lack of Acquaintance The Kantian argument for ignorance that assumes a deducibility requirement presupposes that knowledge of such properties would be inferential. But Kant also entertains the possibility that knowledge of such properties would be immediate. However, in his view we cannot have immediate knowledge—that is, intuition—of such properties, and this issues in a distinct argument for ignorance about which properties of this sort the world features. Here is one of Kant’s most informative expositions of the Humility thesis: If the complaints—that we have no insight whatsoever into the intrinsic [properties] of things (das Innere der Dinge)—are to mean that we do not conceive by pure understanding what the things that appear to us may be in themselves, they are entirely illegitimate and unreasonable. For what is demanded is that we should be able to know things, and therefore to intuit them, without senses, and therefore that we should have a faculty of knowledge altogether different from the human, and this not only in degree but also in intuition and kind— and thus that we should be not humans but beings of whom we are unable to say whether they are even possible, much less how they are constituted. Observation and analysis of appearances penetrate into what is intrinsic in nature (ins Innere der Natur), and no one can say how far this will go in time. But with all this knowledge, and even if the whole of nature were revealed to us, we should still never be able to answer those transcendental questions which go beyond nature, since it is not given to us to observe our own mind with any other intuition than that of inner sense. For in that lies the secret of the origin of our sensibility. Its relation to an object and what the transcendental ground of this unity may be undoubtedly lie too deeply hidden for us—who after all know even ourselves only through inner sense and therefore as appearance—to be able to use such an unsuitable instrument of investigation for discovering anything except always still more appearances, eager as we yet are to explore their non-sensible cause (nichtsinnliche Ursache). (A277–78/B333–34)

107 Russellian Monism II

This passage identifies lack of insight into things in themselves with lack of insight into the intrinsic [properties] of things. We can safely assume that Kant means these properties to be absolutely and not merely comparatively intrinsic properties. (We can know the comparatively intrinsic properties through observation and analysis of appearances.) The argument for ignorance Kant presents here hinges on the claim that we have no intuition, that is, immediate or direct representation of them (or perhaps of things in themselves as having them). Notice also that the last sentence of the passage suggests that these absolutely intrinsic properties are the nonsensible causes of appearances. In my interpretation, he is thinking of these properties as the fundamental, non-Humean causal powers of the universe, or as bestowing these fundamental causal powers on substances.12 Can the claim that we are ignorant of which absolutely intrinsic categorical properties are instantiated plausibly be sustained just by the fact that we lack the power to intuit such properties or, in more familiar terminology, that we cannot have Russellian acquaintance with them? Kant himself maintains that we can know unobservable features of material things despite the fact that we do not have immediate perception of these features: “from the perception of the attracted iron filings we know of the existence of a magnetic matter pervading all bodies, although the constitution of our organs cuts us off from all immediate perception of this medium.”13 These views of Kant’s might be reconciled, but let us set that issue aside. What this example brings to the fore is that we believe we can have relevantly complete knowledge of categorical and intrinsic physical properties, for example, of the intrinsic structural properties that water has, even though we lack acquaintance with them. But then, as Alyssa Ney asks, why couldn’t science discern the fundamental intrinsic properties that ground the physical world?14 If we are indeed ignorant of which absolutely intrinsic categorical properties are instantiated, there must be a sharp contrast between our access to these properties and our access to the intrinsic structural properties of water.

Ignorance from Failure of Abduction The H2O-structural property is an intrinsic property of water, and we arguably understand the complete nature of this property and that water has it. We have this knowledge despite lacking acquaintance with this property because we conceived a model of the unobserved basis of water dispositions that turned out to be a component of a best explanation. In principle, could we not do the same for absolutely intrinsic properties? We might imagine: physics provides a model for the fundamental particles in which their absolutely intrinsic property is perfect solidity. The model turns out to be so 12 This is one respect in which my interpretation of Kant’s position differs from Rae Langton’s in Kantian Humility. For a further discussion of my interpretation, see Derk Pereboom, “Is Kant’s Transcendental Philosophy Inconsistent?” esp. pp. 363–64. For a nuanced view of the philosophical issues involved, see Alyssa Ney, “Physicalism and Our Knowledge of Intrinsic Properties.” 13 Immanuel Kant, Critique of Pure Reason, A225–26/B273–74. 14 Alyssa Ney, “Physicalism and Our Knowledge of Intrinsic Properties,” p. 57.

108 Consciousness and the Prospects of Physicalism

explanatorily impressive that it yields abductive knowledge that perfect solidity is an instantiated absolutely intrinsic property. However, given this abductive model, it remains plausible that we are currently ignorant of which absolutely intrinsic properties are instantiated. Several distinct candidates for such properties have been conceived that are not abductively ruled out, and it is seriously open that we have not yet conceived all of the viable candidates. This will be true not only on Lewis’s quiddism, as Dustin Locke in effect convincingly argues,15 but also if we take Shoemaker’s causal structuralist view as the point of departure. Shoemaker’s causal structuralism does not preclude distinct absolutely intrinsic properties with causal profiles we are unable to distinguish from one another, either because the distinguishing elements of these causal profiles are uninstantiated or because we lack the ability to discern them. Moreover, even if we were able to individuate the instantiated absolutely intrinsic properties by a causal-role specification, we might yet be significantly ignorant of them because a causal role specification provides us with only limited knowledge of a property’s nature. Which candidates for absolutely intrinsic properties have we conceived? We’ve already discussed perfect solidity, and also Leibniz’s model in which the absolute intrinsic properties are mental properties of immaterial entities. Robert Adams has recently developed and defended a theistic variant on this exclusively mentalistic proposal.16 On a panpsychist version of this option, defended by Galen Strawson, the absolutely intrinsic properties are mental properties of physical entities.17 We’ve mentioned that Chalmers suggests a protophenomenalist alternative. Armstrong once proposed primitive color as the missing intrinsic physical property, and we could expand this idea to include primitive versions of the other secondary qualities.18 One might want to say that a number of these proposals can be ruled out as too wild to be in play. However, reflection on the strengths of the knowledge and conceivability arguments against physicalism suggests that possibilities that initially seem wild remain salient after all. Also, if introspection does accurately represent phenomenal properties as having qualitative natures that physical theory does not represent them as having, the plausibility of a number of these views is raised. In addition, it appears very far from certain that any proposed candidate that we understand is actually instantiated, and so

15 Dustin Locke, “A Partial Defense of Ramseyan Humility.” 16 Robert Adams, “Idealism Vindicated,” in Persons, Human and Divine, ed. Peter van Inwagen and Dean Zimmerman (Oxford: Oxford University Press, 2007), pp. 35–54. 17 Galen Strawson, “Realistic Monism,” in his Real Materialism and Other Essays (Oxford: Oxford University Press, 2008), pp. 54–74. 18 David Armstrong, Perception and the Physical World (London: Routledge, 1961); Armstrong rejects this proposal in A Materialist Theory of Mind (London: Routledge, 1968). He proposes color as the relevant property; I’m assuming that he had primitive or perfect color in mind (see chapter 2)—an intrinsic property of physical objects whose qualitative nature is accurately and completely represented by normal human sensory color perception. This proposal can’t be ruled out a priori. Still, it seems extremely unlikely on empirical and philosophical grounds that there is such a property in the physical world, even for medium-size objects.

109 Russellian Monism II

it may well be that there are possibilities for such properties that we do not comprehend that are also salient alternatives. So we have come to an argument for ignorance about which absolutely intrinsic properties are actually instantiated from the claim that there is a plurality of candidates for such properties, some of which are not currently understood, more than one of which is still in the running for yielding the best explanation of the relevant phenomena, and none of which now convincingly does so. Unlike Kant’s or Lewis’s arguments, the conclusion is not inevitable and permanent ignorance, but rather a sort that is potentially remediable. It is thus congenial to Chalmers’s proposal, which leaves it open that we will come to adequately understand the nature of the fundamentally intrinsic properties that ground physical dispositions, whether they be phenomenal or protophenomenal. On the other hand, perhaps the Intrinsicness Principle, that any mind-independently real substantial entity must have at least one absolutely intrinsic property, should not be allowed to stand no matter what. It has been challenged in recent philosophy of physics by structural realists, who claim that all physical properties reduce to purely extrinsic or structural properties, and yet physical things are mind-independently real.19 Kant would say that if there are no absolutely intrinsic physical properties, then physical things cannot be mind-independently real. But we might escape this conclusion if the Intrinsicness Principle is legitimately abandoned, as structural realists advocate. But they do so only controversially. The intuition that the Intrinsicness Principle is true is strong, and the prospect of rejecting it can’t be taken lightly.

RUSSELLIAN MONISM AGAIN We provisionally characterized Russellian monism as any view according to which physics or current physics leaves us ignorant of (certain) categorical properties that underlie physical dispositions, and these categorical properties have a crucial role in accounting for consciousness. Given Leibnizian intuitions, the unknown categorical properties underlying the known physical properties must include at least one absolutely intrinsic property. We now have in place an argument that we are currently ignorant of which properties of this type are actually instantiated (with the escape clause that there may be no such properties). This argument yields support for Chalmers’s position, according to which the absolutely intrinsic properties of fundamental physical entities are not available by way of a priori reflection on ‘PT,’ and a posteriori physical investigation has not yet revealed to us the nature of these properties. How might we assess the various proposals for absolutely intrinsic properties as ways of filling out Russellian monism? If we supplemented ‘P’ just with putative truths about perfect solidity, assuming that perfect solidity is the missing absolutely intrinsic property, the sense that ‘PT and ~ Q’ is ideally, primarily, positively conceivable is not 19 See James Ladyman’s exposition in his “Structural Realism.” Jennifer McKitrick, in “The Bare Metaphysical Possibility of Bare Dispositions,” might be viewed as providing a general philosophical challenge to the Intrinsicness Principle.

110 Consciousness and the Prospects of Physicalism

diminished (supposing that introspection accurately represents the qualitative natures of phenomenal properties). Imagine instead, inspired by Armstrong’s suggestion, that we supplemented ‘P’ just with putative truths about primitive colors or primitive versions of other secondary qualities. Aristotle conceived of such properties as physical, so maybe the result could be a variety of physicalism. But the idea that these are the missing absolute intrinsic properties does not seem especially plausible. At this point, we seem to have run out of candidates for the missing absolutely intrinsic physical properties that have been conceived. What remains are the mental candidates and, as Nagel, Chalmers, and McGinn suggest (or might be taken to suggest), possible candidates that we have not conceived. Chalmers takes the mental candidates seriously—the resulting view would qualify as panpsychism or as micropsychism. As Galen Strawson points out, one might hold, by contrast with the panpsychist, that some but not all microphysical entities have intrinsic mental or phenomenal properties, and he calls this less demanding view micropsychism.20 The most favorable prospect for physicalist Russellian monism would appear to lie in properties whose nature is currently unconceived. Protophenomenalism is a view of this sort. The kind of ignorance about the properties at issue that would be in place, together with the fact that the tradition in physics allows for entities not hitherto countenanced as physical (recently, quantum fields) to count as physical, would seem to make protophenomenalism the physicalist Russellian monist’s best hope. If there are currently unconceived possibilities for physical and protophenomenal absolutely intrinsic properties, they might remain unconceived. More optimistically, as physics develops, we may come to conceive them. Or as Chalmers suggests, phenomenology together with physics might arrive at such a conception.21 We will examine these proposals critically. But first, let’s turn to a recent challenge to Chalmers’s version of the Humility claim.

STOLJAR’S OBJECTION TO CHALMERS’S ARGUMENT FOR RUSSELLIAN MONISM On one version of Chalmers’s position, what underwrites the conceivability and knowledge arguments against ordinary physicalism is the following structure-anddynamics thesis: (SDT) There are experiential [or phenomenal] truths that cannot be deduced from truths solely about structure and dynamics.22 20 Galen Strawson, “Realistic Monism.” 21 In his presentation on structuralism in physics at the Australian National University, November 2005. 22 This formulation is from Torin Alter, “Does the Ignorance Hypothesis Undermine the Conceivability and Knowledge Arguments?” Philosophy and Phenomenological Research 79 (2009), pp. 756–65, at p. 760.

111 Russellian Monism II

When Chalmers writes as if he could be interpreted this way, structural and dynamical properties contrast with intrinsic properties.23 Since the properties that physical theory specifies (ordinary physical theory—not supplemented by protophenomenal concepts, for instance) are all structural and dynamical, and phenomenal properties are intrinsic properties of experiences, one can conclude that experiential truths about phenomenal properties cannot be deduced from (ordinary) physical theory. If SDT turns out to be mistaken, as Stoljar argues it is, then it may be that these experiential truths are derivable from the complete physical truth, P, after all. This, in turn, would threaten to undermine the motivation for a Russellian monist account of consciousness. As Torin Alter explains it, SDT is based on three claims; (1) There are experiential truths, (2) The from-structure-only-structure thesis, that is, from truths solely about structure and dynamics, one can deduce only truths solely about structure and dynamics, and (3) The experience-isn’t-just-structure thesis, that is, experiential truths are not solely about structure and dynamics.24 In his critical discussion of Chalmers’s view, Stoljar rejects the second thesis and raises issues for the third. About the second thesis, he contends: The simplest way to see that the from-structure-only-structure thesis is false is to note that one can derive the instantiation of an intrinsic property from a relational one just by shifting what thing you are talking about. For example, being a husband is a relational property of Jack Spratt, and being a wife is a relational property of his wife. But being married is an intrinsic property of the pair (or the sum) of Jack Spratt and his wife. To take a different example, it seems plausible to say that I have the property of having a hand intrinsically, but my having this property obviously follows from a relation between my hand and the rest of my body, and that the truth concerning this is a relational truth.25 Alter agrees that Stoljar has a point: if objects x and y compose object z, then it is possible to derive intrinsic properties of z from relational properties of x and y. But Alter thinks that this observation undermines the from-structure-only-structure thesis only if nonstructural/nondynamical properties are identified with intrinsic properties, and in his view that identification is mistaken, for “the property being married is purely structural/dynamic despite being intrinsic to the Spratts. Any structural/dynamical

23 David Chalmers, “Does Conceivability Entail Possibility?” p. 197. 24 Torin Alter, “Does the Ignorance Hypothesis Undermine the Conceivability and Knowledge Arguments?” pp. 761–63; cf. Daniel Stoljar, Ignorance and Imagination, pp. 147–53. 25 Daniel Stoljar, Ignorance and Imagination, p. 152.

112 Consciousness and the Prospects of Physicalism

duplicate of the actual world contains a corresponding married pair.”26 Alter thinks that such examples indicate not that we should reject the from-structure-only-structure thesis, but rather that we should not identify nonstructural/nondynamical properties with intrinsic properties. How do we evaluate the claim that being a married pair is a structural/dynamical property of the Spratts? Kant’s distinction between comparatively and absolutely intrinsic properties yields a diagnosis. Although the property of being a married pair is intrinsic to the Spratts, this reduces to, and can be derived a priori from, Jack’s purely extrinsic property of being married to Jill and Jill’s purely extrinsic property of being married to Jack. Thus being a married pair is merely a comparatively intrinsic property and not an absolutely intrinsic property of the Spratts. Perhaps, then, we should say that all nonstructural/nondynamic properties will be absolutely intrinsic properties. Stoljar’s counterexample would then fail against (2), the from-structure-only-structure thesis. The revised version of Chalmers’s proposal would then be that phenomenal properties of experiences are intrinsic to them, and not merely comparatively but absolutely intrinsic to them. The epistemic thesis about the physical would be that current physics is limited in what it directly specifies to extrinsic or comparatively intrinsic properties, while phenomenal properties are absolutely intrinsic properties of experience. The from-structure-only-structure thesis becomes the claim that from facts about extrinsic and merely comparatively intrinsic properties, no facts about absolutely intrinsic properties can be derived a priori.27 These ideas need to be refined. Stoljar argues that: When Chalmers says . . . that “truths about consciousness are not truths about structure and dynamics,” this is false . . . because consciousness itself has structural and dynamic features.28 To this, Alter responds: Although consciousness has structural and dynamic features, it does not follow that Chalmers’s claim is false, i.e., that truths about consciousness are truths about structure and dynamics. For comparison, consider the claim that truths about biology are not truths about geometry. The latter claim is not refuted by the observation that living things have geometrical properties. However,

26 Alter adds: “I assume instantiating being married in no way consists in having experiences. Otherwise instantiating the corresponding relational properties would also consist at least partly in having experiences, in which case those properties might not be purely structural/dynamic”; see Torin Alter, “Does the Ignorance Hypothesis Undermine the Conceivability and Knowledge Arguments?” p. 763, note 8. 27 On the derivability characterization of absolutely intrinsic properties, this is a definitional fact. 28 Daniel Stojlar, Ignorance and Imagination, p. 147.

113 Russellian Monism II

Chalmers’s formulation could mislead: some experiential truths are about (experiential) structure and dynamics, at least in part. He should reformulate his claim as follows: truths about consciousness are not truths solely about structure and dynamics. If the proposal is that truths about consciousness are not solely truths about structure and dynamics, preserving a nonderivability thesis requires that physics characterizes properties as solely structural and dynamical. One way to explicate this proposal is by way of the notion of a purely extrinsic property: physics characterizes properties either as purely extrinsic or as resolving into purely extrinsic properties, and while phenomenal properties may have extrinsic aspects, they are not purely extrinsic, and for this reason, all the facts about phenomenal properties cannot be a priori derived from the physical facts. With these clarifications, we can restate the from-structure-only-structure thesis in this way: (2*) From truths solely about purely extrinsic properties and comparatively intrinsic properties, we cannot a priori derive truths about absolutely intrinsic properties or absolutely intrinsic aspects of properties. In fact, this claim will be true by the a priori derivability versions of the definitions of comparatively and absolutely intrinsic properties. Correlatively, the structure-anddynamics thesis can be reformulated as follows: (SDT*) There are experiential truths that cannot be a priori derived from truths solely about purely extrinsic properties and comparatively intrinsic properties. One significant concern that remains is this. Suppose physics did feature characterizations of absolutely intrinsic properties, such as being perfectly solid or being prime matter, and a property of this sort is widely distributed, as Locke and Newton or the late-medieval Aristotelians maintained. Imagine that Mary has mastered the information about such a property when she is in the black-and-white room. The intuition that she learns something when she leaves the room and introspects her sensation of red is not diminished. This suggests that it is not the from-structure-only-structure thesis and the structure-and-dynamics thesis, at least under these interpretations, that accounts for the intuition that Mary learns something when she leaves the room. So either there is another reading of these theses on which being perfectly solid and being prime matter count as structural properties29 or else we are back to the more general claim made by advocates of the knowledge and conceivability arguments, formulated by Stoljar, that the theories of contemporary physics quantify only over entities with 29 Alter suggested this in response to this challenge at the Pasadena American Philosophical Association meeting, 2008.

114 Consciousness and the Prospects of Physicalism

some feature X, which is such that phenomenal truths cannot be derived solely from truths about entities with that feature.

MICROPSYCHISM AND PROTOPHENOMENALISM Chalmers’s Russellian monist thought is that one can primarily conceive ‘PT and ~ Q’ only because one is conceiving just structural and dynamical properties on the physical side, which, in my view, can be analyzed as purely extrinsic and merely comparatively intrinsic properties. If ‘P’ were replaced with an embellished ‘P*’ that included concepts that allowed for direct representation of the currently unknown absolutely intrinsic properties, ‘P*T and ~ Q’ would not be primarily conceivable. Instead, ‘Q’ would be derivable from ‘P*T’ a priori. The resulting Russellian monism has phenomenal-micropsychist and protophenomenal versions. On the phenomenal-micropsychist option, which Strawson would endorse, the absolutely intrinsic properties that account for our phenomenal consciousness are themselves phenomenal and irreducibly so, while on the protophenomenalist variant, they are not phenomenal but nonetheless account for phenomenal consciousness.30 Imagine first that ‘P*’ supplements ‘P’ by adding in the proposed micropsychist truths, statements about phenomenal absolutely intrinsic properties of fundamental physical entities that directly specify those properties. Suppose ‘Q’ to be a phenomenal truth about my current experience of blue. Would ‘P*T and ~ Q’ be ideally, positively, primarily conceivable? We might ask whether there is any less reason to believe that ‘P*T and ~ Q’ is ideally, positively, primarily conceivable than there is to believe this about ‘PT and ~ Q.’ Imagine that every fundamental particle has some absolutely intrinsic phenomenal property or other, and that ordinary introspectible phenomenal states are composed of many fundamental particles of this sort. It seems as easy for me to conceive of any such array of fundamental particles without my phenomenal blueness as it is to conceive of any conventional physical concatenation of fundamental particles without it. But in support of the micropsychist, we can invoke a misrepresentation thesis here as well, not of the Kantian variety, according to which introspection represents experience as having properties it actually lacks, but of the Leibnizian sort, by which introspection merely fails to represent experience as having properties it in fact has.31 While my experience of blue is represented introspectively to feature only phenomenal blueness, this phenomenal blueness is in fact composed of an unrepresented complex microphenomenal array. Here phenomenal-micropsychism might have an advantage over conventional physicalism, since it is perhaps more plausible that ordinary phenomenal blueness is composed of an unrepresented complex microphenomenal array than that it really is conventionally physically constituted. This advantage is due to the 30 See Galen Strawson, “Realistic Monism”; there Strawson also defends the stronger view, panpsychism; see also Thomas Nagel, “Panpsychism,” in his Mortal Questions (Oxford: Oxford University Press, 1979). 31 Thanks again to Nico Silins for this distinction.

115 Russellian Monism II

fact that phenomenal-micropsychism demands less by way of error for introspective representation than does conventional physicalism. Phenomenal-micropsychism requires only that introspection mistakenly represents properties like phenomenal blueness, and in particular its qualitative nature, not as having a complex phenomenal composition. Conventional physicalism requires in addition that the phenomenal blueness of my experience does not have any qualitative phenomenal nature of the general type that introspection represents it as having. Note that phenomenal-micropsychism would claim that there are laws governing how truths about microphenomenal properties yield truths about macrophenomenal properties such as my experience’s phenomenal blueness. These laws would have to be derivable from ‘P*T’ alone (P* adds in the micropsychist truths), for ‘Q’ must be derivable from ‘P*T’ alone. Perhaps this can be rendered credible by the analogy of the derivability of certain macrophenomenal properties from their known components, such as phenomenal tastes from the components of sweet, sour, salty, bitter, and umami.32 Introspectible phenomenal properties would be analogously derivable from currently unknown microphenomenal absolutely intrinsic properties and the rest of the base described by ‘P*T,’ and the relevant laws would then be similarly derivable from this base. However, building on a point made by Karen Bennett, the envisioned sort of phenomenal micropsychism would need to posit fundamental laws linking the micropsychist absolutely intrinsic properties with the microphysical properties that they underlie, without which the truths about the microphysical properties could not be derived from the micropsychist truths.33 For this reason, phenomenal micropsychism seems inadequately set up to yield a deeply illuminating explanation of the properties specified by current microphysics. And it would be theoretically advantageous if the absolutely intrinsic properties provided such explanations for both phenomenal properties and the properties specified by current microphysics. Chalmers’s protophenomenalist proposal appears better equipped for this dual task. It is much less specific about the nature of the absolutely intrinsic properties, and partly for this reason, it leaves open the possibility that these properties would count as physical. But then it would also be open that the protophenomenal properties yield explanations for the microphysical properties they underlie without fundamental laws linking the protophenomenal properties with the microphysical properties. This results in an advantage over phenomenal micropsychism. Now imagine that ‘P**’ supplements ‘P’ by adding the truths about protophenomenal absolutely intrinsic properties of fundamental physical entities by way of concepts that directly refer to such properties. Would ‘P**T and ~ Q’ be ideally, positively, primarily conceivable? It seems epistemically open that there are protophenomenal properties such that the phenomenal truths are derivable a priori from truths about them, together with the rest of what is included in ‘P**T,’ and this would undercut the ideal, positive, primary

32 Thanks to Louis deRosset for this suggestion. 33 Karen Bennett, “Why I Am Not a Dualist,” ms.

116 Consciousness and the Prospects of Physicalism

conceivability of ‘P**T and ~ Q.’ The resulting all-inclusive potential explanatory advantage of protophenomenalism over phenomenal micropsychism is offset to a certain degree by the liability that it proposes properties of which we currently have only a minimal conception. But given the available options, this downside is not decisive. Might we ever form concepts that directly specify such protophenomenal properties? Chalmers is cautiously optimistic. In a slightly different context, McGinn is thoroughly skeptical. Protophenomenalism is consistent with McGinn’s claims, but he would deny that we can form concepts that directly represent the natures of protophenomenal properties. For him, to solve the mind-body problem, we would need to form concepts that would bridge the gap between conscious properties as revealed by introspective acquaintance-concepts on the one hand, and neural and other physical properties on the other. By contrast with forming concepts that facilitated past major theoretical shifts in science, such as the advance to relativity theory, this we cannot achieve; “what we need is a perspective shift, not just a paradigm shift—a shift not merely of world view, but of ways of apprehending the world. We need to become another type of cognitive being altogether.”34 But Nagel and Chalmers think it is open that our cognitive and imaginative capacities are up to this sort of task. For example, Nagel says: The difference between the mental and the physical is far greater than the difference between electrical and mechanical. We need entirely new intellectual tools, and it is precisely by reflection on what appears impossible—like the generation of mind out of the recombination of matter—that we will be forced to create such tools. It may be that the eventual result of such exploration will be a new unity that is not reductionist. We and all other creatures with minds seem to be composed of the same materials as everything else in the universe. So any fundamental discoveries we make about how it is that we have minds, and what they actually are, will reveal something fundamental about the constituents of the universe as a whole. In other words, if a psychological Maxwell devises a general theory of mind, he may make it possible for a psychological Einstein to follow with a theory that the mental and the physical are really the same.35 What might well explain McGinn’s reluctance to take this route is that for him any concepts available to us will be sufficiently closely tied to acquaintance to foreclose the possibility of our acquiring concepts of the bridging sort specified, while for Nagel and Chalmers, it’s open that human imagination will be capable of venturing beyond these limits to form the kinds of concepts at issue. Kant, in The Critique of the Power of Judgment, potentially sides with Nagel and Chalmers against McGinn’s position on the limits of our abilities to form concepts.

34 Colin McGinn, “What Constitutes the Mind-Body Problem,” p. 24. 35 Thomas Nagel, The View from Nowhere, pp. 52–53.

117 Russellian Monism II

The power of judgment is the ability to think particulars under a universal—a universal rule, principle, or law. If the universal is given, Kant says, the judgment that subsumes particulars under it is determinative. For him, there are two sorts of given universals: the empirical concepts whose legitimate applicability to experience is secured because they have been derived from experience, and the a priori concepts that have their source in the subject and for which there is a transcendental deduction. But if the universal is not given and only particulars are, the judgment must find a universal concept for the particular, and the judgment is reflective.36 Finding a universal is a creative act that requires our imaginative capacities to surpass what is merely given. Perhaps McGinn is right to argue that the sorts of concepts at issue cannot be constructed from given empirical concepts, and the kinds of given a priori concepts Kant has in mind wouldn’t seem to help. For us to be able to form the concepts at issue, we might well need the power of reflection. Whether we have it is in contention. If we do have it, what we can currently think and understand would then not preclude that we could form protophenomenal concepts and come to know whether protophenomenal properties are among the fundamental properties of the universe.

T WO NOTIONS OF OBJECTIVITY, T WO NOTIONS OF THE PHYSICAL We’ve seen that a range of positions pivot around the Intrinsicness Principle—that any mind-independently real substantial entity must have at least one absolutely intrinsic property, and whether we can know which absolutely intrinsic properties are actually instantiated. Leibniz accepts the Intrinsicness Principle and believes he can establish a mentalistic proposal about the nature of those properties. Kant also endorses this principle but argues that we cannot know which absolutely intrinsic properties the world features. Kant thus accepts both the intrinsicness and ignorance provisions of Russellian monism. Lewis concurs with Kant’s position, and McGinn may as well. Chalmers is situated between Leibniz and Kant, since he accepts the Intrinsicness Principle but contends that while we are currently ignorant about which absolutely intrinsic properties are instantiated, this ignorance may not be permanent. There are also those who reject the Intrinsicness Principle, such as James Ladyman, Stephen French, and Don Ross in the philosophy of physics, and Jennifer McKitrick cautiously suggests that it may be dispensable.37 At the same time, the legacy of this principle and the claim of ignorance is complicated by the repercussions of an antimetaphysical,

36 Immanuel Kant, Critique of the Power of Judgment, tr. Paul Guyer and Eric Matthews (Cambridge: Cambridge University Press, 2000), Ak V 179. 37 Steven French and James Ladyman, “Remodeling Structural Realism: Quantum Physics and the Metaphysics of Structure,” Synthèse 136 (2003), pp. 31–56; James Ladyman and Don Ross, Every Thing Must Go (Oxford: Oxford University Press, 2007), chapters 2 and 3, pp. 66–189; for an overview, see James Ladyman, “Structural Realism,” especially section 4, “Ontic Structural Realism”; Jennifer McKitrick, “The Bare Metaphysical Possibility of Bare Dispositions.” Thanks to Alyssa Ney for discussion of these issues.

118 Consciousness and the Prospects of Physicalism

antirealist dismissal of absolutely intrinsic properties, which was extensively developed by Rudolf Carnap, most notably in his The Logical Structure of the World (Der Logische Aufbau der Welt).38 Broadly Leibnizian reflections give rise to two distinct notions of the physical. One of these is characterized by the structural and dynamical properties over which physics quantifies, while the other includes physical substantival absolutely intrinsic properties. In the twentieth century, logical positivists, most notably Carnap, focused solely on the former, as yielding a paradigmatic notion of objectivity. In his antimetaphysical conception, any terms for absolutely intrinsic properties are to be eliminated in a physical language. But despite the rehabilitation of metaphysics in the second half of the twentieth century, and with it a metaphysical notion of the physical, the fact that something was eliminated by Carnap’s conception has often remained forgotten. In the Aufbau, what makes the physical paradigmatically objective is that the physical is expressible in a purely relational logical language. This sort of logical expressibility makes for a kind of objectivity that is in turn analyzed as a kind of rational intersubjective accessibility. Herbert Feigl also thinks of the physical as objective in this way.39 Here is a characterization of rational intersubjective accessibility based on Feigl’s: X is rationally intersubjectively accessible just in case X’s existence and defining properties (or essence) can be known either directly through observations that any subject with a reasonably powerful sensory apparatus could make, or indirectly through deduction, induction, or abduction from such observations and background conditions.40 The first phase of Carnap’s project in the Aufbau aims to show that the phenomenal language we ordinarily use to describe the world of experience can be translated into a phenomenal language that employs no monadic predicates and just one basic relation, that of recollected phenomenal similarity in some respect.41 This canonical

38 Rudolf Carnap, The Logical Structure of the World, tr. Rolf George (Berkeley and Los Angeles: University of California Press, 1967). 39 Herbert Feigl, The ‘Mental’ and the ‘Physical’ (Minneapolis: University of Minnesota Press, 1967), pp. 66–68. 40 Here is an alternative characterization that accommodates a rationalist perspective: X is rationally intersubjectively accessible just in case X’s existence and defining properties (or essence) can be known either directly through rational intuitions or else by observations available to any subject with a reasonably powerful cognitive and sensory apparatus, or indirectly through deduction, induction, or abduction from such intuitions and observations, together with background conditions. 41 Chalmers discussed Carnap’s project in his presentation on structuralism at the Australian National University, November 2005.

119 Russellian Monism II

phenomenal language in turn can be translated into a physical language that features no monadic and only relational predicates.42 In the final phase, the conjunction of all the sentences in the relational physical language is converted into a Ramsey sentence, in which the referring terms are replaced by existentially quantified variables, and the relational predicates are also replaced by variables. The specific content of the physical relational predicates disappears, leaving only descriptions that abstract from this content. The result is a complete description of the world in an abstract logical language of relations—a purely structural language.43 The aspiration is that this description will command general rational agreement, irrespective of initial philosophical predilections. In Carnap’s scheme, what has become of the absolutely intrinsic properties, about which metaphysicians conflict? Sentences with terms for such intrinsic properties are transformed into purely structural sentences, purged of features that impede rational agreement. Some contemporary functionalist projects are similar. The content of phenomenal experience is expressed in functional terms, as relations among sensory inputs, behavioral outputs, and other mental states; the outcome is then Ramsified; and the hope is that there will be nothing about the resulting sentences that would thwart rational agreement. Carnap’s position arguably has its roots in Hume’s treatment of the idea of causal power; the sorts of absolutely intrinsic properties at issue are often held to be fundamental causal powers.44 Hume’s argument is that since we have no sensory impression corresponding to the idea of causal power, this idea is fictional or else meaningless. After showing some idea to be fictional or meaningless, he typically specifies a way to reformulate thoughts employing the defective idea in epistemically acceptable terms. Those that would seem to be about causal powers are recast as thoughts about regularities or counterfactual dependencies among epistemically transparent perceptions. In the same spirit, Carnap has us translate statements that include terms for entities that warn of epistemic opacity into transparent substitutes. For our case in point, statements with terms for intrinsic properties that threaten to be opaque are translated into

42 For Carnap, it turns out that the phenomenal language of experience translates equally well into distinct, nonequivalent structural physical languages, and which language one chooses is a matter of convention and not of one being superior to another. Each of these physical languages expresses the entire content of the phenomenal language. 43 In 1928, Maxwell Newman argued that structure conceived in this way is not sufficient to uniquely pick out any relations in the world; see M. H. A. Newman, “Mr. Russell’s Causal Theory of Perception,” Mind 37 (1928), pp. 137–48. As Ladyman puts the problem, “Suppose that the world consists of a set of objects whose structure is W with respect to some relation R, about which nothing else is known. Any collection of things can be regarded as having structure W provided there is the right number of them. This is because according to the extensional characterisation of relations defined on a domain of individuals, every relation is identified with some set of subsets of the domain. The power set axiom entails the existence of every such subset and hence every such relation” ( James Ladyman, “Structural Realism”). 44 David Hume, An Enquiry Concerning Human Understanding (Oxford: Oxford University Press, 1999), §VII.

120 Consciousness and the Prospects of Physicalism

statements without terms for such properties, featuring by way of replacement only terms for epistemically transparent relations. The project is to secure objectivity by reformulation of any content that hinders rational intersubjective agreement.45 A critical question for Carnap is whether the ordinary phenomenal language of experience is genuinely translatable without loss of content into physical-structural or logical-structural sentences. This question has been a focus of controversy. But a further issue is whether such structural sentences can adequately describe things that are objective in another way, in the sense of being metaphysically objectively real: X is metaphysically objectively real just in case X’s existence and X’s essential nature are independent of how X is perceived or conceptualized. The notion of the metaphysical objective reality comes apart from the notion of objectivity as intersubjectivity.46 Something’s being objectively real in this sense does not require that it be intersubjectively accessible. Nagel plausibly contends that there may be metaphysically objectively real things that we cannot even conceive.47 Also, our

45 In Carnap’s view, structuralism extends to all of the sciences by virtue of the doctrine of the unity of science. Claude Lévi-Strauss advocates a structuralist view of sociology and anthropology, for example, in Les structures élementaires de la parenté, 1949, translated as The Elementary Structures of Kinship by James Harle Bell, John Richard Von Sturmer, and Rodney Needham (Boston: Beacon, 1969). It is interesting to note that Jacques Derrida’s central criticism of Lévi-Strauss’s structuralism is that structure presupposes what he calls a center that is not itself structural but is rather inner or intrinsic. In his view, even though structuralists tried to do without a center, the sense that there is such a center is inescapable. Here Derrida is arguably expressing the Leibnizian-Kantian intuition that we have been exploring. See Jacques Derrida, “Structure, Sign, and Play in the Discourse of the Human Sciences,” in his Writing and Difference, tr. Alan Bass (Chicago: University of Chicago Press, 1978), pp. 278–93; originally L’écriture and la différence (Paris: Editions du Seuil, 1967). A defining feature of Derrida’s view is a denial of Kant’s optimism about the possibility of achieving scientific knowledge, given ignorance of the absolutely intrinsic properties. Kant endorses the view that the sciences can be successful and purely structural, while Derrida claims that our ignorance of intrinsic properties will constrain the success of any attempt at systematic knowledge. Derrida’s position is illuminated by Maxwell Newman’s criticism of pure structuralism (note 43, this chapter). For any purely structural system—a text, in Derrida’s vocabulary—any attempt to determine an interpretation of it, which in the last analysis involves consideration only of structural relations (les différences), will result in deferring such a determination to consideration of yet further relations. But even the entire system of structural relations will not fix a particular interpretation. As a result, interpretation—and scientific endeavor more generally—faces a serious limitation, which gives rise to Derrida’s deconstructive proposal, according to which any claim to a single privileged interpretation of a text can be undermined, often in an illuminating way. See Jacques Derrida, Of Grammatology, tr. Gayatri C. Spivak (Baltimore: Johns Hopkins University Press, 1976, 1998); cf. Jonathan Culler, On Deconstruction: Theory and Criticism after Structuralism (Ithaca, NY: Cornell University Press, 1982, 2007). 46 Andrew Chignell points out that Kant distinguishes these two notions of objectivity in the Critique of Pure Reason, in the section titled “The Canon of Pure Reason,” A795–831/B823–59; see “Belief in Kant,” Philosophical Review 116 (2007), pp. 323–61, at pp. 336–37. 47 Thomas Nagel, The View from Nowhere, pp. 13–27.

121 Russellian Monism II

notion of intersubjective accessibility does not require that intersubjectively accessible entities be metaphysically objectively real. On a widespread interpretation, Kant held that physical things are intersubjectively accessible but not objectively real in the metaphysical sense just defined.48 Carnap aimed to set metaphysics aside permanently, and with it the metaphysical notion of the objectively real. For him, objectivity is solely a kind of intersubjective accessibility, and insofar as the physical is a notion of objectivity, it is characterized by intersubjective accessibility and not by metaphysical objective reality. Here we find a Kantian theme continued in Carnap. Kant held that since the physical features no absolutely intrinsic properties, it cannot be mind-independently objectively real. Yet it can be objective in the sense that our theorizing about it potentially secures rational agreement. Since the recent renaissance of metaphysics, philosophers have generally assumed that physicalism is a thesis about what is metaphysically objectively real, not merely a claim about what is objective in the sense of rational intersubjective accessibility. However, then the question about the reality and nature of absolutely intrinsic properties arises again and can’t be dismissed in the way it was by the logical positivists.

48 Robert Howell develops a view he calls subjective physicalism (in “The Knowledge Argument and Objectivity,” Philosophical Studies 135 [2007], pp. 135–77; “The Ontology of Subjective Physicalism,” Noûs 43 [2009], pp. 315–45), according to which everything—things, facts, properties—is physical, but no objective theory can be a complete description or account of reality. For Howell, ‘objective’ is not a synonym of ‘real.’ Rather, following Carnap, Feigl, and Nagel (“What Is It Like to Be a Bat,” The View from Nowhere), he characterizes objectivity in terms of accessibility by way of multiple possible routes. More precisely, for Howell, “an objective theory cannot require that a subject enter any token state of determinate type T in order to fully understand states of type T” (“The Knowledge Argument and Objectivity,” p. 149). A subjective feature of the world, by contrast, is one such that the subject must be in a token of one determinate type of state in order to fully understand it. With this understanding of these notions, Howell argues that when Mary is in the room, she has acquired all of the objective knowledge, but it remains open that she does not yet possess all of the physical knowledge. He then contends that when she learns something upon leaving the room, what she learns is subjective but nonetheless physical. (A quibble about Howell’s characterization of objectivity is that while mathematical facts are paradigmatically objective, understanding some mathematical facts may require occupying a state of one determinate type. Perhaps for some complex mathematical theorems, there is only one type of state by means of which it can be understood.) I discuss such a proposal in “Bats, Brain Scientists, and the Limitations of Introspection,” pp. 319–21. In supplement to what I say there, the idea of subjective physical features of the world is fairly unfamiliar, and as a result, perhaps there is a burden on Howell to explain how such features are really possible. The paradigm cases of physical entities are objective in the sense of rational intersubjective accessibility, and if Howell wants to propose that some physical entities are subjective, he owes us an account of how this could be. In particular, given that understanding the nature of all currently paradigmatic cases of physical entities does not require being in a state of a particular determinate type, one would like to know what it is about the subjective physical features that explains why understanding them requires being in a state of one particular determinate type, and how it might be that such features should nevertheless count as physical.

122 Consciousness and the Prospects of Physicalism

T WO VARIETIES OF PHYSICALISM According to the protophenomenalist option for physicalism, the underlying absolutely intrinsic properties are nonmental and similar enough to paradigmatic properties of current physics to count as physical, while at the same time they have a crucial role in grounding phenomenal properties of conscious states. An important advantage of this conception is that it can endorse the qualitative accuracy thesis about introspective phenomenal property representation that provides the knowledge and conceivability arguments with their considerable intuitive force. This view is in fact partly motivated by such an endorsement. But such protophenomenal properties are currently unconceived. A physicalist position that does not have this sort of disadvantage accepts the qualitative inaccuracy thesis. On this option, the knowledge and conceivability arguments appear to refute physicalism only because we are not taking into consideration the open possibility that introspection misrepresents phenomenal properties by attributing to them qualitative features they do not really have. This open possibility is a serious one, and the physicalist position that ensues is a rival to protophenomenalism, micropsychism, and primitivism in its dualist and idealist varieties as an account of consciousness.

7 ROBUST NONREDUCTIVE PHYSICALISM

INTRODUCTION In this chapter, I defend a nonreductive version of the physicalist position against its opponents. The contemporary debate between reductionism and nonreductivism has its roots in a concern about whether humanistic theories, such as history and beliefdesire psychology, are importantly distinct from the natural sciences, paradigmatically physics, and also chemistry and biology, together with its subfields, such as neurophysiology. In the late nineteenth and early twentieth centuries, Wilhelm Dilthey and neo-Kantians such as Wilhelm Windelband, Heinrich Rickert, and R. G. Collingwood argued for an affirmative answer1—in Dilthey’s case, on the ground that the method of the human sciences requires an empathetic understanding of the minds (Verstehen) of others while the method of the natural sciences does not; for Windelband and Rickert, for the reason that the aim of the human sciences is to determine individual events by contrast with general laws, and our understanding of individual events will of necessity be value-dependent in a particular way;2 and for Collingwood, in addition to grounds

1 Wilhelm Dilthey, The Formation of the Historical World in the Human Sciences (1910; Princeton, NJ: Princeton University Press, 2002); Rudolf Makkreel, Dilthey: Philosopher of the Human Studies (Princeton, NJ: Princeton University Press, 1975); Wilhelm Windelband, “History and Natural Science” (1924), tr. G. Oakes, History and Theory 19 (1980), pp. 169–85; Heinrich Rickert, The Limits of Concept Formation in the Natural Sciences, tr. G. Oakes (1929; Cambridge: Cambridge University Press, 1986); R. G. Collingwood, The Idea of History (Oxford: Oxford University Press, 1946); Guiseppina D’Oro, “Collingwood, Metaphysics, and Historicism,” Dialogue 41 (2002), pp. 1–20. 2 In Rickert’s conception, an individual object is always indefinitely complex in its features, and since one’s conception of it must be finitely complex, one must be selective in which features one conceptualizes it as having. This selection process will be mediated by one’s values. See Heinrich Rickert, The Limits of Concept Formation in the Natural Sciences. 123

124 Consciousness and the Prospects of Physicalism

similar to Dilthey’s, because the human sciences provide reasons-explanations, and not the causal explanations characteristic of the natural sciences. Logical positivists such as Carnap opposed these claims with the hypothesis of the unity of science.3 All of the sciences aim to determine general causal laws and to predict and explain particular events by means of such laws. Furthermore, on Carnap’s version of the unity doctrine, all the genuine claims of the human sciences can be translated without loss of content into sentences of natural science, of physics in particular. In the late nineteenth and early twentieth centuries, the contentions of the nonreductivists were largely methodological as opposed to metaphysical, in the sense that they did not argue for the existence of distinct metaphysical levels of reality corresponding to different sorts of sciences, but rather for diverse methods of thinking or explanation. However, nonreductivism is well-advised to adopt a metaphysical and not just a methodological focus. One reason is that there is a reasonable presumption that a nonreductive methodology requires a dualistic metaphysics or else an idealistic metaphysics in which, for example, the differing domains of the natural sciences and the human sciences result from the imposition of distinct conceptual schemes. It might then be objected that such views are untenable. In response, the nonreductivist might well want to establish that her position is also compatible with physicalism. Moreover, it is independently valuable to discern the metaphysical commitments of a nonreductive position and whether it can accommodate physicalism. A metaphysical type of nonreductivism was proposed in Amsterdam in the 1920s and 1930s by Herman Dooyeweerd, who rejected the explicit or implicit antimetaphysical stance of the neo-Kantians and the logical positivists.4 More exactly—and this is a general model I endorse—his fundamental conception is metaphysical, while at the same time it has significant consequences for the methodologies of the special sciences. In this conception, reality is structured by an integrated hierarchy of distinct law-spheres or sets of laws, with the quantitative the most basic, through the physical and biological, and on to the law-spheres of the human and social sciences. No one of these sets of laws is reducible to a more basic one, yet each presupposes those that are more basic than it. The task of each special science is to discover and systematize its set of laws, under the constraint of these metaphysical assumptions. Dooyeweerd casts his view as neither physicalist nor idealist, but as pluralist in a contrasting sense, for the reason that the ontology of any law-sphere is not reducible to that of any other. Thus he does not assume the burden of attempting to show that nonreductivism is compatible with physicalism; in fact, on his nonreductive conception, it is not. The renewed debate about reductionism in the 1960s and 1970s did feature a nonreductivism that is at the same time physicalist in its metaphysics, and this is the position I defend. Still, in these decades the debate retained a pragmatic and nonmetaphysical slant. When Hilary Putnam first argued for nonreductive physicalism, he cited the 3 Rudolf Carnap, “Psychology in Physical Language,” Erkenntnis 3 (1932–33), reprinted in Logical Positivism, ed. A. J. Ayer (New York: Macmillan, 1959), pp. 165–98. 4 Herman Dooyeweerd, A New Critique of Theoretical Thought, tr. David Freeman (Philadelphia: Presbyterian and Reformed Publishers, 1953–58).

125 Robust Nonreductive Physicalism

phenomenon of multiple realizability as its main ground.5 Since mental states can be realized by indefinitely many kinds of neurophysiological states, and perhaps by many kinds of nonneurophysiological states, mental states are not reducible to neurophysiological states. While multiple realizability is a metaphysical notion, the most prominent argument for nonreductivism in which it played a role featured a significant pragmatic element. Consider Jerry Fodor’s statement of this argument in his influential article “Special Sciences.”6 We begin by imagining a law in some special science: S1x causes S2x where S1 and S2 are natural kind predicates in that science. The most appropriate model of reduction requires that every kind that appears in this law be identified with a kind in the reducing science, in virtue of bridge principles. Bridge principles specify an appropriate metaphysical relation, typically identity, between the kinds of one science and those of the reducing science. But in some cases, the sort of bridge principle required for reducibility will not be available. If kinds in psychology are multiply realizable in an indefinite number of ways at the neurophysiological level, purported bridge principles for relating psychological to neurophysiological kinds will involve open-ended disjunctions. Such purported bridge principles typically have the form: P1 = N1 v N2 v N3 . . ., which says that a certain psychological state, P1, is identical to an open-ended disjunction of neurophysiological states, N1 v N2 v N3. . . . Fodor argues that since open-ended disjunctions of kinds in neurophysiology are not natural neurophysiological kinds, psychological kinds cannot be reduced to neurophysiological kinds. Why are such disjunctions not natural kinds? Fodor’s reason is that they are not natural kinds because they cannot appear in laws. They cannot appear in laws because “laws” involving such disjunctions are not explanatory. He then argues that such laws are not explanatory because they do not meet our interests in explanation. Fodor’s argument for irreducibility, then, appeals to the fact that purported explanations for psychological phenomena fail to satisfy our interests when couched in terms of open-ended disjunctions. If a nonmetaphysical, pragmatically driven reductionism were the target of this argument, then invoking these interests concerning explanation might well be pertinent. But the metaphysical reductionist could claim that this argument has little force against her view. Suppose we accept multiple realizability, she might say. Even then, open-ended disjunctive explanations could be true, despite failing to meet certain pragmatic requirements, for it might be that our failing to find explanations

5 Hilary Putnam, “The Nature of Mental States.” 6 Jerry Fodor, “Special Sciences.”

126 Consciousness and the Prospects of Physicalism

satisfying when they contain open-ended disjunctions reflects a legitimate pragmatic concern, but no metaphysical facts. It’s not obvious to me that this is correct.7 But since the nonreductivism I propose is primarily metaphysical and derivatively methodological, it would be advantageous to adduce clearly nonpragmatic reasons to accept it. With the metaphysical focus on the ascendancy in recent decades, nonreductive physicalism about the mental has been put on the defensive by a series of welldeveloped arguments against its central claims. Four of these challenges, each of which has been advanced by Jaegwon Kim, are especially prominent: the argument from causal exclusion against irreducibly mental causal powers; the contention that the nonreductive view is indistinguishable from the British emergentism of the late nineteenth and early twentieth centuries,8 a position widely held to be metaphysically extravagant; the claim that the functionalism typically endorsed by nonreductive physicalists is incompatible with irreducibly mental causal powers; and the argument that if mental state types are multiply realizable, they cannot be genuinely scientific kinds, for then terms that refer to these kinds will be only as weakly projectible as the description of the wild disjunction of their possible realizations. Here is the plan for this chapter and the next. I first examine whether nonreductive physicalism might finesse the causal exclusion problem. I claim that it can and that one way to do so is by grounding the mental in the neural and the microphysical by a constitution relation.9 I then set out my version of this position and differentiate it from other nonreductive physicalist views. Subsequently, I argue that there are significant differences between the controversial sort of emergentism and a plausible sort of nonreductive physicalism and that a nonreductive physicalist need not be emergentist in this sense. In the next chapter, I contend that a position according to which mental states instantiate irreducibly mental causal powers—the key feature of a robust version of nonreductive physicalism—cannot be functionalist in the standard sense, according to which the essence of mental states and properties consists in their causal relations to sensory inputs, behavioral outputs, and other mental states. I then argue there is a nonfunctionalist conception of mental states and properties to which the nonreductive physicalist can turn. In this account, mental properties are compositional: that is, properties something has solely by virtue of intrinsic features of its parts 7 Hilary Kornblith and I have suggested that such a pragmatic concern might yield evidence that open-ended disjunctions do not reveal mind-independently real psychological laws. Since concerns of this sort might plausibly be instrumental in generating theories that are instead predictively successful, it could be reasonable to conclude that our failing to find explanations involving open-ended disjunctions satisfying indicates that they do not reflect real laws; see Derk Pereboom and Hilary Kornblith, “The Metaphysics of Irreducibility,” Philosophical Studies 63 (1991), pp. 125–45, at pp. 127–28. 8 Brian McLaughlin, “The Rise and Fall of British Emergentism,” in Emergence or Reduction? Essays on the Prospects of Nonreductive Physicalism, ed. A. Beckermann, H. Flohr, and J. Kim (New York: Walter de Gruyter, 1992), pp. 49–93. 9 Richard Boyd’s position in “Materialism without Reductionism: What Physicalism Does Not Entail” is an important precursor to this view.

127 Robust Nonreductive Physicalism

and relations these parts have to one another.10 I close that chapter by evaluating concerns that have recently been raised for arguments from multiple realizability, and I conclude that the nonreductive view I set out can withstand these challenges. This nonreductive position identifies mental properties with (broadly) physical compositional properties at a level higher or more abstract than neural compositional properties, and my sense is that it accommodates many of the reasons type-type identity theorists have had for preferring their stance. By way of completing the two physicalist accounts of phenomenal properties I’ve developed, the one that builds in qualitative inaccuracy and the Russellian monist option, for each I endorse interpretations on which phenomenal properties are compositional. In addition, this theory of mental properties can profitably be viewed as an instance of Stephen Yablo’s proposal that the relation between mental properties and their underlying physical properties is that of determinable to determinate.11 Finally, it is significant that in this position the physicality of mental state types is twice grounded, first by way of constitution of each token of the type in the microphysical, and second by way of identity to sufficiently abstract physical compositional properties.

CAUSAL EXCLUSION Robust nonreductive physicalism, as I conceive it, is a view about specifically psychological entities and explanations, although it easily generalizes to other levels of reality. In this discussion, I will use the term ‘causal power’ to refer to that by virtue of which an effect is produced.12 I’m attracted to the view that property instances are the entities by virtue of which effects are produced, as L. A. Paul advocates, and thus causal powers would be property instances. In the ensuing discussion, it can be assumed that I have this view in mind, although my account does not depend on this particular conception. Causal powers would then be, in the first place, token entities rather than types, as is sometimes supposed.13 A mere relation of a property instance to an effect will not count as a causal power, since such a relation is implausibly that by virtue of which an 10 David Armstrong classifies such properties as structural in Universals and Scientific Realism (Cambridge: Cambridge University Press, 1978). This use of that term conflicts with the one at play in the discussion of Russellian monism in chapters 5 and 6. In “Robust Nonreductive Materialism,” I used Armstrong’s term. 11 Stephen Yablo, “Mental Causation,” Philosophical Review 101 (1992), pp. 245–80; for a careful critical discussion of this proposal, see Eric Funkhouser, “The Determinable-Determinate Relation,” Noûs 40 (2006), pp. 548–69. 12 Lynne Baker, The Metaphysics of Everyday Life (Cambridge: Cambridge University Press, 2007), p. 98. 13 L. A. Paul, “Aspect Causation,” Journal of Philosophy 97 (2000), pp. 235–56; Paul argues for this view partly on the ground that aspects or property instances are sufficiently fine-grained, by contrast with events, at least on some conceptions. I prefer to think of property instances as ways particular things are, as in David Robb’s characterization in his “Power Essentialism”, (manuscript) contrasting with abstracta. Token causal powers will then also be ways particular things are, and not abstracta, as some might conceive causal power types.

128 Consciousness and the Prospects of Physicalism

effect is produced.14 I am partial to the non-Humean idea that the paradigm case of a causal power is an instance of an intrinsic property that is dispositional in the sense that it is an aptness to produce effects, as Shoemaker specifies, where that aptness is not reducible to a feature of the world that is not itself an aptness of this kind, such as a regularity or counterfactual dependency among nondispositional property instances.15 According to robust nonreductive physicalism, an act such as Anne’s deciding to catch the next train to Grand Central Station—let’s call it M2—will be explained psychologically by some mental state M1, which consists in certain mental properties being instantiated at some time. Each of M1 and M2 will be microphysically realized—assuming no specific theory of realization—where realization is just, following Shoemaker, to make real or to implement.16 In particular, there will be a microphysical realization P2 of M2 and a microphysical realization P1 of M1, such that P2 is microphysically explained by P1. But the psychological explanation of M2 by M1 will not reduce to the microphysical explanation of P2 by P1 (and, likewise, mutatis mutandis for states at various other levels of description, such as the neural). Underlying the irreducibility of the psychological explanation is the fact that it appeals to the irreducibly mental causal powers of M1 to account for M2, while the microphysical explanation appeals to microphysical causal powers of P1 to account for P2. Accordingly, the causal powers of M1 will not be identical with those of P1, and those of M2 will not be identical with those of P2. If these identities did hold, then the causal powers to which the psychological explanation refers would in the last analysis be microphysical. Psychological explanations might then presume a classification that clusters microphysical causal powers in a way distinct from how microphysics sorts them, but this would not compromise the microphysical status of those causal powers.17 But furthermore, in this robust nonreductive conception, there will be a microphysical explanation for P2 that appeals to the microphysical causal powers of P1, and at the same time P2 (perhaps together with certain relational features) will be (noncausally, 14 Hence the relations in terms of which external-relations functionalism characterizes mental states are not causal powers. (See chapter 8 for a further discussion of this sort of functionalism.) 15 Sydney Shoemaker, “Causality and Properties” and in conversation. For a comprehensive discussion of Shoemaker’s views on these issues, see Dean Zimmerman, “Properties, Minds, and Bodies: An Examination of Sydney Shoemaker’s Metaphysics,” Philosophy and Phenomenological Research 79 (2009), pp. 673–738. Rom Harré and E. H. Madden develop a view of this type in Causal Powers. Another way of specifying this non-Humean notion of a causal power is as an instance of an intrinsic qualitative property that is a vehicle of dispositionality, in accord with Heil’s suggestion, where its being such a vehicle is not reducible to a feature of the world that is not itself a vehicle of this sort, such as a regularity or counterfactual dependency among nondispositional property instances; see John Heil, “Dispositions,” p. 352; see also David Robb, “Power Essentialism.” 16 Sydney Shoemaker, Physical Realization (Oxford: Oxford University Press, 2007), pp. 1–2. When I specify the relation I have in mind more precisely, I will opt for a variety of constitution. 17 Jaegwon Kim discusses several nonreductive views that are not robust in this sense in Mind in a Physical World (Cambridge, MA: MIT Press, 1998), pp. 67–87.

129 Robust Nonreductive Physicalism

synchronically) sufficient for M2. Consequently, there will be a microphysical causal explanation for M2 that appeals to the microphysical causal powers of P1. Since one standard way of explaining an event causally is to cite causal powers whose exercise is a sufficient cause of this event, citing the causal powers of P1 yields a causal explanation of M2. In a distinct nonreductive model, deriving from Putnam, which I do not endorse, there is no genuine microphysical or even a neural explanation for the action, only a psychological one.18 Familiarly, the position under scrutiny gives rise to pressing questions. What is the relationship between the microphysical and psychological explanations for M2? And given that both sorts of explanation refer to causal powers, what is the relationship between the causal powers to which the microphysical explanation appeals and those to which the psychological explanation appeals? This is the point at which Kim’s challenge from causal exclusion enters in.19 We just saw that since P1 yields a causal explanation of microphysical realization P2 of M2, it will provide a causal explanation of M2 itself. What room is then left for a distinct psychological causal explanation of M2 by M1?20 Kim argues that it is implausible that the psychological explanation appeals to causal powers whose exercise is sufficient for the event to occur, and at the same time the microphysical explanation appeals to distinct causal powers whose exercise is also sufficient for the event to occur, and that as a result the event is causally overdetermined. The concern is that on this picture every event that is mentally caused will be overdetermined in the way that someone who is fatally hit simultaneously by each of two bullets shot by two assassins is overdetermined, a result that is implausible. Let us call this phenomenon redundant overdetermination (a necessary condition of which we will examine in section 5; it’s open that there be nonproblematic nonredundant overdetermination). It is also implausible that each of these distinct groups of causal powers yields merely a partial cause of the event and that each would be insufficient for the event to occur. According to the solution to this problem Kim develops, real causal powers exist at the microphysical level, and so microphysical explanations will refer to real 18 Hilary Putnam, “Language and Reality,” in Philosophical Papers, vol. 2, pp. 272–90, at p. 278; see also Stephen Yablo, “Mental Causation.” Elliott Sober argues against the viability of this alternative in “The Multiple Realizability Argument against Reductionism,” Philosophy of Science 66 (1999), pp. 542–64. In addition, suppose neural property instance N1 causes an instance of depression, D1, but without any rival mental cause, but in a different case, another neural property instance N2 is a candidate for a cause of an instance of the same sort of depression, D2, but with a rival mental cause M2. It would seem odd for the depression to have a neural cause in the first but not in the second case. 19 See, for example, Jaegwon Kim, “The Myth of Nonreductive Materialism,” in his Supervenience and Mind (Cambridge: Cambridge University Press, 1993), pp. 265–84, at pp. 280–82; This article was first published in Proceedings and Addresses of the American Philosophical Association 63 (1989): 31–47. Yablo provides a historical bibliography of the causal-explanatory exclusion argument in “Mental Causation,” p. 247, note 5. 20 Note that the crucial element of how I set out the exclusion problem here is the upward causation of M2 by P1, by contrast with the downward causation of P2 by M1, as we at times see in Kim’s formulations.

130 Consciousness and the Prospects of Physicalism

microphysical causal powers. Only if mental causal powers are identical with microphysical causal powers does it turn out that there is genuine mental causation, but mental causal powers will then ultimately be microphysical. Psychological explanations that do not reduce to microphysical explanations will fail to refer to causal powers and thus will have some lesser status; such explanations might express regularities without at the same time referring to causal powers. On Kim’s view, this account solves the exclusion problem because if the causal powers to which a psychological explanation refers are identical with those to which the underlying microphysical explanation refers, then there will be no exclusion-generating competition between mental and microphysical causes, and if a psychological explanation does not refer to causal powers at all, there will be no such competition either. But this account, which Kim believes is the only viable solution to the problem he raises, would rule out any robust nonreductive view about mental causal powers. Kim concludes that any token causal powers of a higher-level property instance at a time will be identical with token microphysical causal powers of its microphysical realization at that time. He applies this view to mental properties by invoking the following principle: [The Causal Inheritance Principle] If mental property M is realized in a system at t in virtue of physical realization base P, the causal powers of this instance of M are identical with the causal powers of P.21 Kim contends that rejecting this principle would be tantamount to accepting “causal powers that magically emerge at a higher level and of which there is no accounting in terms of lower-level causal powers and nomic connections.”22 If the causal inheritance principle is true, there would be no token causal powers distinct from token microphysical causal powers, and this would rule out robust nonreductive physicalism.23

AGAINST IDENTITY AND THE SUBSET VIEW What sort of response might the advocate of robust nonreductive physicalism provide? Various proposals have been advanced under the name “nonreductive physicalism,” according to which mental properties are causally relevant or explanatory without being causally efficacious qua mental properties. For a mental property to be causally efficacious qua mental, its causal efficacy cannot consist only in its microphysical, chemical, or neural realizers’ being causally efficacious. Such purportedly nonreductive physicalist views, just like Kim’s, 21 Jaegwon Kim, “Multiple Realization and the Metaphysics of Reduction.” 22 Jaegwon Kim, “Multiple Realization and the Metaphysics of Reduction,” pp. 326–27. 23 Terence Horgan, “Kim on Mental Causation and Causal Exclusion,” Philosophical Perspectives 11 (Oxford: Blackwell, 1997), pp. 165–84, at p. 179. See also Kim’s discussion of this move in Mind in a Physical World, pp. 67–72.

131 Robust Nonreductive Physicalism

claim that all causal efficacy is nonmental in this way.24 As Kim points out, these proposals do not amount to a robust sort of nonreductive physicalism, one that would preserve the claim that mental properties or events, qua mental properties or events, are causally efficacious.25 To advance a robust nonreductive physicalist response, let’s begin by assessing the causal inheritance principle just mentioned. In Kim’s conception, any token causal power of a higher-level property instance at a time will be identical with a token microphysical causal power. However, first of all, a strong case can be made that higher-level token entities are typically not identical with lower-level token entities that realize them. The ship of Theseus is not identical with a token current microphysical realizer because, for example, it would have been the same token ship had the token microphysical realizer been microphysically exactly qualitatively similar with yet distinct from the actual one. It would also remain the same ship if the token microphysical realizer actually changed just enough to become a different microphysical token. The ship is in this sense token multiply realizable. Mark Moyer and Lynne Baker are among those who concur that the ship is not identical to its current plank-realization or its current microphysical realization but argue that we have a notion of sameness by which the ship is the same thing as the current token microphysical realizer, one that abstracts from any temporally extrinsic or modal properties.26 All of this seems plausible to me. Moyer and Baker argue that the intuition that ship and its current plank-realization are identical can be explained or explained away by this notion, thus mitigating a challenge to the view that they are not absolutely identical by virtue of differing in their modal or temporal properties. For Baker, this relation of sameness has a positive role in her account of material constitution, since it is part of what comprises the relation of unity between a constitutor and a constituted entity. Contemporary nonreductive physicalism about the mental is grounded in modal and temporal arguments against identity,27 and if one is persuaded by arguments of this sort against a type-identity thesis, one should at least be motivated to accept similar arguments against any token-identity thesis about the mental. If one is inclined to dissent from such a pluralist view, according to which the ship and its realizer are 24 See, for example, Frank Jackson and Philip Pettit, “Program Explanation: A General Perspective,” Analysis 50 (1990), pp. 107–17. 25 Jaegwon Kim, Mind in a Physical World. 26 Mark Moyer, “Statues and Lumps: A Strange Coincidence?” Synthèse 148 (2006), pp. 401–23; Lynne Baker, The Metaphysics of Everyday Life, pp. 40–41. Gareth Matthews points out that Aristotle endorsed such a notion of sameness alongside absolute identity; see Gareth Matthews, “Aristotle’s Theory of Kooky Objects” (ms); Baker also argues that constitution is not identity in “Why Constitution Is Not Identity,” Journal of Philosophy 94 (1997), pp. 599–621, and in Persons and Bodies (Cambridge: Cambridge University Press, 2000). 27 Saul Kripke, Naming and Necessity, pp. 144–48; Derk Pereboom and Hilary Kornblith, “The Metaphysics of Irreducibility,” pp. 131–32; Lynne Baker, Explaining Attitudes: A Practical Approach to the Mind (Cambridge: Cambridge University Press, 1995), pp. 9–10; Stephen Yablo, “Mental Causation.”

132 Consciousness and the Prospects of Physicalism

not identical, and to accept a monist view on which they are, one might take my argument for robust nonreductive physicalism to be conditional: if modal and temporal arguments taken to support nonidentity claims are sound, then the arguments for robust nonreductive physicalism will have significant force.28 So first, is token mental state M identical with P, an actual token microphysical realizer? Suppose that M is realized by a complex neural state. It is possible for M to be realized differently only in that a few neural pathways are used that are distinct from but perhaps exactly qualitatively similar to those actually engaged in M. We need not rule at this point on whether the actual neural realization is identical with this alternative; for all that’s been said, it might be. ( Just as the ship of Theseus would retain its identity supposing the replacement of just a few of its planks, so it would seem that a token neural state would retain its identity given the replacement of just a few of its neural pathways—more on this in the next chapter.) But it is evident that this alternative neural realization is itself realized by some microphysical state P* distinct from P. It is therefore possible for M to be realized by a microphysical state not identical with P, and thus M is not identical with P. Furthermore, this reflection also suggests a challenge to a token-identity claim for mental causal powers—should they exist—and their underlying neural and microphysical causal powers. Sydney Shoemaker and Jessica Wilson have endorsed a token-identity thesis for mental and lower-level causal powers, although they oppose reductive type- and token-identity claims for mental states. On their view, the mental is realized by and grounded in the neural and in the microphysical because the (forward-looking) causal powers of a mental state are a proper subset of the (forward-looking) causal powers of the lower-level state. Here is Shoemaker’s statement of his position: It is compatible with the claim that the instance of the higher-order property and that of its realizer are not identical that the former is part of the latter. And that seems the right conclusion to draw from the fact that the causal powers of the former are a proper subset of those of the latter. And then it is open to us to say that while it is true that the instance of the realizer property causes the 28 This pluralist view is fairly standard, but there are a number of dissenting views according to which it is false; see Ryan Wasserman, “Material Constitution,” in the Stanford Encyclopedia of Philosophy, for a fine account of the various positions on this issue. Mark Moyer, in “Statues and Lumps: A Strange Coincidence?” provides a recent defense of the pluralist view I endorse; see also Mark Johnston, “Constitution Is Not Identity,” Mind 101 (1992), pp. 89–105. Pluralists argue that the absolute numerical identity of x and y requires that they share all of their modal and temporal extrinsic properties, and since the statue and the lump do not, they are not absolutely numerically identical. Extreme monists maintain that the statue is absolutely numerically identical with the lump that now constitutes it even if there is another time during which the lump exists but the statue does not. Moderate monists contend that the statue is absolutely numerically identical with the lump only if they exist at the same times; Harold Noonan, in “Constitution Is Identity,” Mind 102 (1993), pp. 133–46, champions moderate monism; cf., “Moderate Monism, Sortal Concepts, and Relative Identity,’ Monist, forthcoming.

133 Robust Nonreductive Physicalism

various effects we attribute to the realized property, it does so because it includes as a part the instance of the realized property.29 On Shoemaker’s account, it turns out that because the forward-looking causal powers of a mental property instance are a subset of the forward-looking causal powers of the realizing physical property instance, each forward-looking causal power of the mental property instance will be identical with a forward-looking causal power of the physical property instance. Wilson makes this feature of the view explicit: What it is for a higher-level property to be realized by a lower-level property is for the set of forward-looking conditional causal powers associated with the higher-level property to be a subset of the set of forward-looking conditional causal powers of the lower-level property. Given this account of realization, it will never be the case that a given higher-level property has a conditional causal power different from any of those of its realizer base property.30 However, if modal multiple realizability arguments count against token identity claims for mental states and properties, they should for mental causal powers as well, no matter what one thinks causal powers are. So let me propose a modal multiple realizability argument that targets mental causal powers directly. Consider Anne’s belief at some particular time that her parents live in Manhattan—a mental token, an instance of a mental property—and the token causal power that it has or (as I prefer) with 29 Sydney Shoemaker, “Physical Realization and Mental Causation,” in The New Ontology of the Mental Causation Debate, ed. S. C. Gibb and Jonathan Lowe (Oxford: Oxford University Press, forthcoming). In this article, Shoemaker withdraws the claim he made in Physical Realization that the definition of realization requires reference to backward-looking causal features. The quoted passage continues as follows: For example, the instance of C-fiber firing causes, or contributes to causing, the moaning, groaning, and calls to the doctor, but it does so because it includes the instance of pain. It includes the instance of pain because its instantiation guarantees, constitutively, the instantiation of a property having the causal profile of pain—this because of the subset relations between the causal profiles of the two properties. The part of the causal profile of C-fiber firing that is exercised here is precisely the part it shares with the causal profile of pain in virtue of having the forward-looking causal features of pain as a subset. So while it is true that the instance of C-fiber firing “does the causal work,” it does not do so in a way that leaves the instance of pain with no work to do; on the contrary, it does the causal work because it includes as a part the instance of pain. 30 Jessica Wilson, “How Superduper Does a Physicalist Supervenience Need to Be?” Philosophical Quarterly 49 (1999), pp. 33–52, at p. 50. L. A. Paul also advocates an overlap account as a response to threat of redundant overdetermination, but in her view the overlap is between property instances; see L. A. Paul, “Constitutive Overdetermination,” in Topics in Contemporary Philosophy, Volume 4: Causation and Explanation (Cambridge, MA: MIT Press, 2005), pp. 265–90.

134 Consciousness and the Prospects of Physicalism

which it is identical. Suppose Anne is threatened with an illness that would damage a small part of her brain that has a crucial role in realizing this belief (but other parts of her brain have important roles as well). Before this part is damaged by the illness, a neurosurgeon could remove it and replace it with a sophisticated electronic microprocessor—let’s call it a silicon prosthesis. Imagine that the illness never actually materializes, and Anne does not undergo the operation. Still, this token belief would have retained its token mental causal power had she undergone the operation, and had it thus at that time been realized by the token neural-and-silicon causal power instead of the token neural power that actually realizes it. Consequently, the token mental causal power cannot be identified with a token neural causal power, specifically not with the token neural causal power that actually realizes it. We can construct a temporal variant of this argument if we are allowed the supposition that belief tokens can persist over a significant span of time. Imagine that the illness does continue to threaten, and before the part of the brain is damaged, the neurosurgeon removes it and replaces it with the silicon prosthesis. After the operation, Anne retains her token belief about where her parents live, and it possesses the token mental causal power it had prior to the operation. But this mental token causal power is no longer realized by the pre-operation neural token causal power but rather by a neural-and-silicon token causal power. Thus the token mental causal power cannot be identified with any neural token causal power, in particular, not with the token neural causal power that realized it just before the operation. Furthermore, since the token neural-and-silicon causal power, on the one hand, and the token neural causal power, on the other, are themselves realized by different types and thus different tokens of microphysical causal powers, we can conclude that the token mental causal power cannot be identified with any token microphysical causal power. We can now draw the general conclusion that token mental causal powers are not identical with token microphysical causal powers, and for a reason that motivates nonreductive physicalism generally: multiple realizability at the microphysical level.31 More generally, according to robust nonreductive physicalism, mental types are not identical with either neural or microphysical types, because any token of any mental type is realizable by tokens of distinct neural or neural-and-silicon types and by tokens of distinct microphysical types. Moreover, mental tokens are not identical with neural or microphysical tokens, since mental tokens can be realized by distinct neural or neural-and-silicon tokens and by distinct microphysical tokens. Nonreductive physicalists have often grounded their physicalism in a mental/physical token identity thesis.32 But any such token identity claim is vulnerable to a multiple realizability argument, an 31 But as I indicate in the next chapter, I would claim that a mental property instance is identical to a sufficiently abstract physical compositional property instance, and, correlatively, a mental causal power is identical to a sufficiently abstract physical causal power, where the level of abstraction must be higher than the neural, as this last argument shows. 32 Jerry Fodor, “Special Sciences”; Donald Davidson, “Mental Events,” in his Essays on Actions and Events (Oxford: Clarendon, 1980), pp. 207–25.

135 Robust Nonreductive Physicalism

argument of the very kind that typically motivates a nonreductive variety of physicalism in the first place.

FOR CONSTITUTION Still, given the truth of physicalism, there must be a sense in which the token causal power of mental state M would be “nothing over and above” the token causal power of microphysical state P—M’s causal powers would nevertheless be “absorbed” or “swallowed up” by P’s causal powers.33 But there are importantly distinct modes of this sort of absorption: identity is one, and realization or constitution without identity is another. If there are essentially mental causal powers that are physically realized, the relation of any one such token to its microphysical realization base would not be identity, but rather realization or constitution. So far, I have been using both realization talk and constitution talk for the relation I think holds between mental entities and underlying neural and microphysical entities. With Baker, I prefer to use ‘constitution.’ I will now more precisely specify what I mean by ‘constitution,’ which will also explain why I prefer this term to ‘realization.’ Let me say at the outset that I am confining my attention to material constitution, and I don’t aspire to providing an account of constitution more broadly construed. Given this restriction, one might say that (1) a statue is constituted by a lump, in which case the constitutor does not necessitate the entity that is constituted: the lump can exist without the statue existing. Or one might say instead that (2) the statue is constituted by, for example, particles arranged statue-wise, where this constitution does necessitate the higher-level entity. I aspire to option (2).34 Thus for P to constitute M in the sense I favor, the existence of P will necessitate the existence of M. But I may be forced to agree with Baker that the existence of P may necessitate the existence of M only in a certain relational context, or at least that this pattern holds for some cases of constitution. I will have more to say on this shortly. Carl Gillett differentiates between a flat conception of realization, in which properties of a thing are realized by properties of that same thing, and a dimensioned conception, in which properties of a thing are realized by properties of a distinct thing from which it is constituted.35 Shoemaker employs a similar distinction 33 Cf. John Heil, “Multiple Realizability,” American Philosophical Quarterly 36 (1999), pp. 189–208. 34 Karen Bennett, in “Why the Exclusion Problem Seems Intractable, and How, Just Maybe, to Tract It,” Noûs 37 (2003), pp. 471–97, at p. 495, note 2, points out, by way of criticizing the constitution thesis Kornblith and I proposed in “The Metaphysics of Irreducibility,” that it is inadequate to physicalism because it does not involve the necessitation of the mental by the physical. However, a constitution relation can be defined that stipulates this necessitation, and this is the sort of relation I have in mind. Baker remarks that the constitution relation between the mental and more basic properties that she defines differs from mine in this respect. Baker says “constitutors are not sufficient for constituted properties;” in The Metaphysics of Everyday Life, p. 114, note 54. But see p. 142 below. 35 Carl Gillett, “The Dimensions of Multiple Realization: A Critique of the Standard View,” Analysis 62 (2002), pp. 316–23, and “The Metaphysics of Realization, Multiple Realizability, and the Special Sciences,” Journal of Philosophy 100 (2003), pp. 591–603.

136 Consciousness and the Prospects of Physicalism

between a kind of property realization in which property F of X is realized by property G of the same thing X, and microrealization, in which property F of X is realized by a microphysical state of affairs, which consists in microphysical entities having certain properties.36 My notion of constitution allows for Gillett’s dimensioned view and Shoemaker’s notion of microrealization. If M is constituted by P, P might be a state that consists in certain microphysical entities’ having certain properties. In my conception, there are also states at a level higher than the most fundamental microphysical level—the chemical and neural levels, for example—that constitute M. Shoemaker argues for a related view, that S’s being in pain, for instance, is microrealized by distinct microphysical states of affairs at various levels of abstractness. Thus the constitution-based view I endorse shares some structural features with the notions of realization as explained by Shoemaker and Gillett. How, then, does it differ? Consider an instructive challenge Andrew Melnyk issues to my position: Pereboom explicitly denies that mental event-types are one and the same as microphysical event-types, and that mental event-tokens are one and the same as microphysical event-tokens. To articulate the idea that mental phenomena are nothing over and above physical phenomena, he appeals instead to constitution, claiming that every mental event is constituted of some or other microphysical event. Unfortunately, however, Pereboom offers no account of constitution. All he says is that, if a physical event-token constitutes a mental event-token, then the physical event-token, “together with any requisite relational features,” will be “sufficient” for the mental event-token. But, for all that Pereboom says, this sufficiency might be sufficiency in accordance with a fundamental law of physical-to-mental emergence whereby, if an event of p’s physical type occurs, then an event of m’s mental type occurs; and if it is, then (mental) m’s being constituted by (physical) p won’t entail that m is nothing over and above the physical. So Pereboom needs to say more about constitution.37 This is a fair challenge. In response, the sufficiency I have in mind is necessitation of the existence of M by the existence of P without the supplementation of P by a fundamental law of physical-to-mental emergence.38 We can grant that if this necessitation

36 Sydney Shoemaker, “Realization, Micro-Realization, and Coincidence,” Philosophy and Phenomenological Research,” 67 (2003), pp. 1–23; Physical Realization; “Physical Realization and Mental Causation.” 37 Andrew Melnyk, “Can Physicalism Be Non-Reductive?” Philosophy Compass 3, no. 6 (2008), pp. 1281–96, at p. 1295. 38 In “Robust Nonreductive Materialism,” I argue that the nonreductive view is not committed to emergentism, but I don’t specify nonemergence as a condition on constitution. For another thorough discussion of the need for formulations of physicalism to preclude emergence, see Jessica Wilson, “Supervenience-Based Formulations of Physicalism,” Noûs 39 (2005), pp. 426–59; cf. Colin McGinn, “What Constitutes the Mind-Body Problem,” p. 12.

137 Robust Nonreductive Physicalism

requires supplementation by such a fundamental emergence law, genuine physicalism is precluded (more on emergence later). But I prefer not to build an anti-emergence condition into the definition of what I think to be the crucial relation, which I will argue is best characterized as a type of constitution. So ruling out emergence would require a separate condition on physicalism. In this respect, my characterization of physicalism does not differ in substance from Melnyk’s, or anyone else’s, since no one, to my knowledge, knows how to rule out emergence by way of a more fundamental condition on physicalism.39 With multiple realizability and upward necessitation as accepted constraints, what is the remaining element of the relation between the mental and entities at more basic levels? We are now in a position to rule out identity, in view of the temporal and modal arguments against any claims for identities between mental and lower-level entities. Identity is too strong. An alternative is spatial coincidence, as in Baker’s account.40 Against this, Ted Sider argues that it is conceivable for two spaceships to be made of such extraordinary material that they can fly through each other, for a moment wholly coinciding spatially, without one constituting or realizing the other at all. If there is a corresponding metaphysical possibility, spatial coincidence is too weak.41 A third option, the one I endorse, is that the remaining element is the made up of relation (or equivalently, the wholly made up of relation), conceived as basic in the sense that it cannot be fully analyzed as consisting in more fundamental relations. In particular, it has no analysis into

39 This concern is expressed by Jessica Wilson in “Supervenience-Based Formulations of Physicalism.” Melnyk rules out emergence in his characterization of realization by specifying that propositions expressing the higher-level or nonfundamental facts be derivable from propositions expressing the fundamental physical facts alone, where fundamental emergence laws are not included from the fundamental physical facts; see A Physicalist Manifesto, pp. 20–32, 88–110. In his view, “The necessitation of the nonphysical [i.e., physical in the broad sense] by the physical [i.e., physical in the narrow or fundamental sense] that is entailed by realization requires no fundamental physical-to-nonphysical bridge laws. By contrast, of course, the strong emergence of the nonphysical from the physical would require precisely that the nonphysical be derivable from the physical only via physical-to-nonphysical bridge laws that are fundamental” (A Physicalist Manifesto, p. 32). The reason a proposition about an emergent property won’t be derivable from the appropriate base is just that this base does not include, by initial specification, propositions expressing fundamental emergence laws. 40 Lynne Baker, Persons and Bodies, pp. 39–42. 41 Ted Sider, “Review of Lynne Baker’s Persons and Bodies,” Journal of Philosophy 106 (2002), pp. 45–48; for the analogous point about composition, see Peter van Inwagen, Material Beings (Ithaca, NY: Cornell University Press, 1990), pp. 52–53. Perhaps it’s not clear that Sider’s spatially coinciding spaceships are metaphysically possible. If the absolutely intrinsic property of matter is perfect solidity, as Locke and Newton proposed (see chapter 5), and perfect solidity is also essential to matter, then it might well not be. Dean Zimmerman, in “The Constitution of Persons by Bodies: A Critique of Lynne Rudder Baker’s Theory of Material Constitution,” Philosophical Topics 30 (2002), pp. 295– 338, develops an extended critique of Baker’s claims for the role spatial coincidence can have in an account of constitution.

138 Consciousness and the Prospects of Physicalism

more fundamental mereological relations. The made up of relation is asymmetric and irreflexive: the lattice is not made up of the diamond, and the diamond is not made up of itself.42 It has a specific direction: the less fundamental made up of the more fundamental.43 But the core of the made up of relation is unanalyzable and thus primitive. I resist the claim that this proposal amounts to obscurantism. It is sufficiently clear what we mean when we say that the diamond is made up of a lattice of carbon atoms, and that the brain is made up of a configuration of various kinds of neurons, even if no reductive analysis is provided for this relation. Does another condition need to be added to preclude the whole lump from constituting not only the statue but also the head of the statue? It’s not natural to say that the head is made up of that whole lump. I will make the feature clear by adding to the account that for x to constitute y, x and y must be materially coincident. In turn, inspired by a recommendation from Dean Zimmerman, we might define material coincidence mereologically: x and y are materially coincident just in case they, at some level, are made out of the same parts.44 An alternative is to add the requirement that x and y be spatially coincident, but it will be controversial that there is no possible

42 Baker’s constitution relation is also irreflexive and asymmetric (Persons and Bodies, pp. 39– 42). One difference from my proposal is that Baker attempts to secure asymmetry by way of conditions that indirectly imply it. I criticized her strategy in “On Baker’s Persons and Bodies,” Philosophy and Phenomenological Research 64 (2002), pp. 616–23; Baker replies in “Replies to Derk Pereboom, Michael Rea, and Dean Zimmerman,” Philosophy and Phenomenological Research 64 (2002), pp. 623– 35, and in The Metaphysics of Everyday Life, pp. 163–64. 43 Karen Bennett discusses the general notion of building relations in “Construction Area (No Hard Hat Required),” manuscript, especially section 6. On her account, all such relations are irreflexive, asymmetric, and such that their input is more fundamental than their output. A third element is some sort of intimate connection; spatiotemporal coincidence won’t do, and Bennett does not offer an alternative informative analysis. In my account of constitution, this intimate connection is the made up of relation, understood to involve or be supplemented by material coincidence. For a relevant general discussion of the grounding relation with some similar themes, see Kit Fine, “The Question of Realism,” Philosophical Imprint (1) 2001, pp. 1–30; Jonathan Schaffer, “On What Grounds What,” in Metametaphysics, New Essays on the Foundations of Ontology, ed. David J. Chalmers, David Manley, and Ryan Wasserman (Oxford: Oxford University Press, 2009), pp. 345–83, esp. 375–77. 44 Zimmerman provides an attractive precise characterization of this condition (i.e., 6*). First, a definition: S is a complete decomposition of x =df. Every member of S is a part of x, no members of S have any parts in common, and every part of x not in S has a part in common with some member of S. Here is the condition: (6*) x and y share at least one complete decomposition. Zimmerman points out that (6*) is equivalent to the claim that, at some level, x and y are made out of the same parts; see Dean W. Zimmerman, “The Constitution of Persons by Bodies: A Critique of Lynne Rudder Baker’s Theory of Material Constitution,” p. 297.

139 Robust Nonreductive Physicalism

mismatch between the material structure of an object and the spatial structure of its location.45 One might now ask whether material coincidence all by itself wouldn’t suffice for what I’ve called the remaining element, supplanting the made up of relation. But constitution is intuitively irreflexive, asymmetric, and directed from the more fundamental to the less fundamental, and the made up of relation secures these characteristics, and does so directly, while material coincidence is reflexive and symmetrical. What should we call this fundamental made up of relation, together with material coincidence, upward necessitation, and multiple realizability? Should it be ‘constitution’ or ‘realization’? The former seems better because of the affinity between the concepts ‘constitution’ and ‘being made up of.’ This position differs significantly from the realization theories of Shoemaker, Gillett, and Melnyk, for example, since their views do not appeal to a fundamental made up of relation.46 Let me characterize this notion more formally. Constitution is a relation between concrete physical entities; they might be states, events, property instances, or causal powers.47 Suppose x and y are concrete physical entities. The made up of relation is asymmetric, irreflexive, and directed so that the less fundamental is made up of the more fundamental, while its core is primitive. Entities x and y are materially coincident just in case they, at some level, are made out of the same parts. Then, (C1) x materially constitutes y at t if and only if (a) y is made up of and materially coincident with x at t; (b) necessarily, if x exists at t, then y exists at t and is made up of and materially coincident with x at t; and (c) possibly, y exists at t and it is not the case that y is made up of and materially coincident with x at t.

To avoid such a mereological characterization issuing in identity (we’re assuming the pluralist position on material constitution), following Judith Thomson, I would deny the mereological principle: Extensionality:
∀x

∀y
[x = y
↔

∀
z(Pzx
↔
Pzy)]; (‘Pxy’ stands for ‘x is a part of y’). Judith Thomson, “Parthood and Identity across Time,” Journal of Philosophy 80 (1983), pp. 201–20; Ryan Wasserman, “Material Constitution.” The statue and the lump share all of their parts, but they are not identical since they differ in modal properties. 45 Raul Saucedo, in “Parthood and Location,” forthcoming, Oxford Studies in Metaphysics 6 (2010), argues for the possibility of mismatches of this sort. 46 Melnyk sets out his theory of realization in A Physicalist Manifesto, pp. 1–122. 47 Lynne Baker, Persons and Bodies, pp. 39–42. In Persons and Bodies, the account of constitution is specified for concrete individuals such as statues and pieces of marble. I assumed in my “On Baker’s Persons and Bodies,” Philosophy and Phenomenological Research 64 (2002), pp. 616–23, that it also applied to token beliefs, but in her reply, Philosophy and Phenomenological Research 64 (2002), p. 631, Baker dissented. Later, in The Metaphysics of Everyday Life, pp. 167–68, Baker specifies a notion of constitution for property instances. I agree that there is such a notion, and that it applies to instances of belief properties.

140 Consciousness and the Prospects of Physicalism

The last clause (c) precludes the identity of x and y (on the assumption of the necessity of identity), as does clause (a), since the made up of relation is irreflexive. Baker’s discussion of constitution features a number of counterexamples that would pose a threat to clause (b) of this characterization, the necessitation of the constituted entity by its constitutor.48 Here is one: the existence of the driver’s license in my wallet is not necessitated by the existence of its piece-of-laminated-paper constitutor, for without the legal context that makes it the case that this laminated piece of paper that features the photo and the relevant writing functions as a driver’s license, there would be no driver’s license. Here is another: the existence of the flag on the flagpole is not necessitated by its cloth and dye constitutor, for there would be no flag without a nation and its laws. For a mental example, on an externalist view about psychological content of the kind developed by Tyler Burge, the existence of a token belief with some specific content will not be necessitated by the existence of its neural or microphysical constitutor, for in an alternative physical and social environment, this same neural or microphysical constitutor would not yield a belief with that content.49 Phenomena of these kinds can be accommodated by a characterization very close to (C1), in which, on the recommendation of Baker’s account, (b) is revised to specify that the existence of y is necessitated by the existence of x in an appropriate relational context, and (c) is similarly altered. Suppose ‘D’ designates the y-favorable circumstances—the relational context required for something to be y. Then: (C2) x materially constitutes y at t if and only if (a) y is made up of and materially coincident with x at t; (b) necessarily, if x exists and is in D at t, then y exists at t and is made up of and materially coincident with x at t; and (c) possibly, y exists at t and it is not the case that y is made up of and materially coincident with x in D at t.

48 Lynne Baker, The Metaphysics of Everyday Life, pp. 11–13, 106–10. Here is the condition that corresponds to (b) in Baker’s account: F and G are primary kind properties, properties things have without which they would not exist; x and y are concrete individuals; F and G are distinct primary kind properties; ‘F*’ designates the property of having F as one’s primary-kind; x has F* and y has G*; and ‘D’ designates G-favorable circumstances, that is, the relational context required for something to be a G. Then: it is necessary that:
∀z[(F
*z at t & z is in D at t)
→

∃
u(G*u at t & u is spatially coincident with z at t)]; Persons and Bodies, pp. 41–43. 49 Tyler Burge argues for externalism about psychological content in “Individualism and the Mental,” Midwest Studies in Philosophy 6 (1978), pp. 73–121, and in “Other Bodies,” in Thought and Content, ed. A. Woodfield (Oxford: Oxford University Press, 1979), pp. 97–120. In “Individualism and the Mental,” p. 111, Burge presents a closely related argument against token identities for mental and physical entities. Paraphrasing, if a mental token has essential extrinsic properties that invoke relations to the environment, while the underlying neural and microphysical tokens do not, the mental token will not be identical to the neural token or to the microphysical token.

141 Robust Nonreductive Physicalism

For mental entities, I aspire to C1 being adequate, but I will not argue that point here.50 Constitution is a relation that is most naturally conceived as holding between objects, and between states or states of affairs. Plausibly, a brain is constituted by a complex microphysical object, and a neural state, or state of affairs, is constituted by a microphysical state, or state of affairs. But can constitution relations hold between causal powers? In particular, can one causal power be made up of and materially coincident with another? If causal powers were abstract entities, then it might be difficult to see how this could be. However, if causal powers are conceived as property instances, and more specifically as concrete ways particular things are, then the prospects are good. A diamond’s property instance of being hard, a causal power, is plausibly made up of and materially coincident with an instance of a compositional property featuring bonds among carbon atoms, also a causal power. And it makes sense to say that any of Anne’s particular mental causal powers is made up of and materially coincident with one of her particular neural causal powers. In accord with this conception, the nonreductive position I prefer endorses a weaker variant of Kim’s causal inheritance principle: [The Weaker Causal Inheritance Principle] If mental property instance M is realized in a system at t by realization base (property instance) P, then M, a causal power, is constituted by this instance P, also a causal power. Kim’s original principle is: [The Causal Inheritance Principle] If mental property M is realized in a system at t in virtue of physical realization base P, the causal powers of this instance of M are identical with the causal powers of P.51 The noncosmetic alteration is the substitution of constitution for identity.

AVOIDING REDUNDANT OVERDETERMINATION So now, given that one of the specified constitution relations, C1 or C2, holds between mental entities on the one hand and, for instance, neural and microphysical entities on the other, precisely how might redundant overdetermination (as opposed to

50 I make a case for internalism about psychological content, and against its having essential relational properties, in Derk Pereboom, “Conceptual Structure and the Individuation of Content,” Philosophical Perspectives 9 (1995), pp. 401–26. 51 Jaegwon Kim, “Multiple Realization and the Metaphysics of Reduction.”

142 Consciousness and the Prospects of Physicalism

unobjectionable overdetermination, if there is such a thing)52 be avoided? (Note that the nonreductivist can accept overdetermination that is not redundant.) Karen Bennett provides a necessary condition for redundant overdetermination that, should it be satisfied in instances of mental causation, would pose a problem for the nonreductivist:53 Event E is redundantly overdetermined by M and P only if (O1) if M had happened without P, E would still have happened; and (O2) if P had happened without M, E would still have happened. If M and P redundantly overdetermined E, then each of these counterfactuals would be nonvacuously true. If two assassins’ shots S1 and S2 redundantly overdetermined event D, then it would be nonvacuously true that if S1 had happened without S2, D would have occurred, and that if S2 had happened without S1, D would have occurred. Bennett then argues convincingly that on a nonreductive physicalist view, O2 is either false or vacuously true, depending on how one individuates P and M, and this is indeed the case on the position I have just developed. On my first constitution schema, C1, P would necessitate M, and then it would be impossible that P happen without M, whereupon O2 would be vacuously true. If we required the second schema, C2, and P necessitated M only in a certain relational context, then O2 would be false if not vacuously true. Bennett’s reasoning provides a strong case that a nonreductive position can avoid redundant overdetermination. Very plausibly, she contends that the key ingredient in such an account is the tightness of the connection between the mental and the physical.54 But now a further question arises: what is it about the relation between M and P that makes it tight enough to explain why they are not redundantly overdetermining despite their distinctness? The position of Shoemaker and Wilson features an account of the tightness of this relation. The central nonreductive claim of their view is that mental property instances are not identical with neural or microphysical property instances for the reason that the forward-looking causal powers of the mental property instance are only a proper subset of the forward-looking causal powers of the lower-level property instance.55 52 I’m using ‘redundant overdetermination’ to indicate a kind of overdetermination that would be objectionable in this mental/physical context. For a critical discussion of the claim that any overdetermination would be objectionable here, see Ted Sider, “What’s So Bad about Overdetermination?” Philosophy and Phenomenological Research 67 (2003), pp. 719–26. 53 Karen Bennett, “Why the Exclusion Problem Seems Intractable, and How, Just Maybe, to Tract It,” Noûs 37 (2003), pp. 471–97; “Exclusion Again,” in Being Reduced, ed. J. Kallestrup and J. Hohwy (Oxford: Oxford University Press, 2009). 54 Barry Loewer argues that Kim’s reasons for concern about overdetermination apply only to multiple determination by independent causes (like two assassins acting independently), but not to cases of overdetermination in which the causes are metaphysically connected (in his review of Kim’s Mind in a Physical World, Journal of Philosophy 108 [2001], pp. 315–24). I endorse the substance of Loewer’s criticism. Thomas Crisp and Ted Warfield also propose that Kim too hastily dismisses the overdetermination solution in their “Kim’s Master Argument,” Noûs 35 (2001), pp. 304–16. 55 Sydney Shoemaker, Physical Realization; “Physical Realization and Mental Causation.”

143 Robust Nonreductive Physicalism

However, as we’ve seen, because these forward-looking causal powers of the mental property instance are a subset of such causal powers of the lower-level property instance, each forward-looking causal power of the mental property instance is identical with a forward-looking causal power of the physical property instance. Consequently, the feature of the relation between M and P that precludes redundant overdetermination is in fact the identity of causal powers.56 I have argued that token mental causal powers are not identical with token lower-level causal powers for the reason that motivates nonreductive physicalism generally: multiple realizability of the mental at the lower level. If this reason counts against token identity for mental states and properties, it should for mental causal powers as well. Thus by the sorts of multiple realizability arguments I’ve canvassed, any mental entity that causes effects will not be identical with any underlying microphysical entity that causes effects, whether it be classified as a token state, state of affairs, property instance, causal power, or components or aspects of any of these sorts of entities. But then the tightness of the relation between the mental and the microphysical that would preclude redundant overdetermination cannot result from identity. My alternative proposal is that the tightness consists in the material constitution relation as I’ve defined it. Against this, one might raise the basic objection that if causes M and P are identical, there is a single cause of E, while if M is materially constituted by P, there are (at least) two such causes, and this makes the relation insufficiently tight to bar redundant overdetermination. But this concern is not decisive. As Bennett’s reasoning indicates, it is far from clear that such mere duality introduces redundant overdetermination and that it thus precludes the requisite tightness. And if x constitutes y, x and y are far more intimately related than are the shots of the two assassins. On C1, x’s constituting y is similar to identity in that the existence of x necessitates the existence of y, and on both C1 and C2, x’s constituting y requires that, as in the case of identity, x and y are materially coincident. Moreover, the requirement that y is made up of x significantly enhances the tightness of their relation.57 Furthermore, if identity and not just constitution were required to preclude the sort of causal competition that generates redundant overdetermination, there would be a feature required for avoiding redundant overdetermination that identity has and current constitution does not. Here one might persist and claim that the distinguishing feature of identity is that it secures a single cause. I doubt that there are rational considerations that would decisively defeat this claim. But suppose that this feature of identity is set aside in the interest of a possible rapprochement. We can then point out that the candidate relevant features that identity has and current constitution lacks are constitution at all other times and at all other possible worlds. But then it would have to be the actual or possible absence of constitution at some past time or at some future 56 Jessica Wilson makes this point in “Supervenience-Based Formulations of Physicalism,” Noûs 39 (2005), pp. 426–59, at p. 431. 57 In Gene Witmer’s terms, the resulting overdetermination would be dependent and not autonomous “two assassins” overdetermination; see his discussion in “Functionalism and Causal Exclusion,” Pacific Philosophical Quarterly 84 (2003), pp. 198–214.

144 Consciousness and the Prospects of Physicalism

time, or the possible current absence of constitution that generates the sort of causal competition issuing in redundant overdetermination. It’s difficult to see how this could be so. Suppose that my current token mental state M1 is actually constituted by token microphysical state P1. Agreed, if M1 were identical with P1, and if their causal powers were identical, there would be no causal competition of the kind that generates redundant overdetermination. If substituting constitution for identity did result in such causal competition, this would have to be because at some time in the past or in the future or at any time at some other possible world, M1 was not constituted by P1, or the causal powers of M1 were not constituted by the causal powers of P1. For instance, M1 would still exist even if a few neural pathways of its neural constitution had been token-distinct from what they actually are. These neural changes would make M1’s microphysical constitutor distinct from P1, and thus in some other possible world M1 is not currently constituted by P1, and similarly, mutatis mutandis, for their causal powers. Intuitively, a possibility of this sort could not introduce causal competition of the kind that results in redundant overdetermination. This argument does not involve the conjecture that M1’s features at other times or at other possible worlds are causally irrelevant. It also does not preclude a counterfactual dependence theory of causation. Rather, it makes the much more limited claim that certain modally and temporally extrinsic features of M1—that is, M1’s not being constituted by P1 at some other worlds and at some other times—do not plausibly give rise to M1’s competition with P1 for causing M2 in a way that issues in redundant overdetermination. More generally, the only credible candidates for relevant differences between identity and constitution (other than the singularity of the cause that identity secures) are absence of constitution at other times and other worlds, and it is implausible that redundant overdetermination could result by virtue of such differences.58 It’s thus reasonable to venture that constitution provides the nonreductivist with the requisite tightness of the relation between the mental and the physical, and a solution to the exclusion problem no less adequate than Kim’s own.59

58 It might be that a further source of relevant difference between identity and constitution derives from the possibility that if x merely constitutes y, there may be contingent categorical properties that x and y fail to share. Kit Fine discusses such failures of property sharing in “The NonIdentity of a Material Thing and Its Matter,” Mind 112 (2003), pp. 195–234. For instance, perhaps the statue is Romanesque, while the lump that constitutes it is not. How such possibilities might generate redundant overdetermination is not immediately clear, and my sense is that they wouldn’t. 59 Note that if one required identity for precluding redundant overdetermination and agreed that biological causes are not identical to their underlying constituting causes, one would face the prospect of redundant overdetermination in every case of biological causation. For instance, if one accepted Philip Kitcher’s argument from multiple realizability for the nonidentity of classical genetic causes and their constituting molecular causes, then one would encounter the challenge of genetic/ molecular overdetermination in any case of genetic causation; see Philip Kitcher, “1953 and That: A Tale of Two Sciences,” Philosophical Review 93 (1984), pp. 335–73. This concern might well extend to chemical causes and even to higher-level causes in physics.

145 Robust Nonreductive Physicalism

NONREDUCTIVE PHYSICALISM AND EMERGENCE In my view, no feature of nonreductive physicalism requires the truth of emergentism. By contrast, Kim contends that nonreductive physicalism is committed to this position: “The fading away of reductionism and the enthronement of nonreductive physicalism as the new orthodoxy simply amount to the resurgence of emergentism— not all of its sometimes quaint and quirky ideas but its core ontological and methodological doctrines.”60 A critical examination of Kim’s argument will indicate why the nonreductive view does not have this consequence. In Kim’s analysis, emergentism claims a distinction between two sorts of higher-level properties, resultant and emergent, that arise from the basal conditions of physical systems.61 The basal conditions of a physical system comprise (i) the basic particles that constitute the physical system, (ii) the intrinsic properties of these particles, and (iii) the relations that configure these particles into a structure. The higher-level properties that are merely resultant are simply and straightforwardly calculated and theoretically predictable from the facts about its basal conditions—which include the laws that govern the basal conditions—while those that are emergent cannot be calculated and predicted from those facts. The variety of “predictability” at issue is the possibility of deriving from an entity’s realization base its concurrent higher-level properties. It is in this sense synchronic and not diachronic predictability. Theoretical predictability contrasts with inductive predictability. Having regularly witnessed that an emergent property is realized by particular basal conditions, we would be able to predict this relationship, but this sort of inductive predictability is not at issue. Rather, emergentists maintain basal conditions alone, no matter how complete, will not suffice for derivation of an emergent property.62 Construing emergence in terms of prima facie epistemic notions such as predictability is not standard. Metaphysically, higher-level properties of a thing are emergent just in case they are not necessitated by its basal conditions, as Kim characterizes them, alone. Plausibly, what would typically explain emergence characterized epistemically is emergence in the metaphysical sense. I contend that nonreductivism is not committed to emergentism on either an epistemic or on the more fundamental metaphysical characterization.63 60 Jaegwon Kim, “Making Sense of Emergence,” Philosophical Studies 95 (1999), pp. 3–36, at p. 5. 61 Jaegwon Kim, “Making Sense of Emergence,” pp. 6–7. 62 Jaegwon Kim, “Making Sense of Emergence,” p. 8. 63 Randolph Clarke provides a fine statement of emergentism and argues that the nonreductivist can avoid it in “Nonreductive Physicalism and the Causal Powers of the Mental,” Erkenntnis 51 (1999), pp. 295–322. For skeptical concerns about emergentism, see Brian McLaughlin, “The Rise and Fall of British Emergentism”; Mark Bedau, “Weak Emergence,” Philosophical Perspectives 11 (Oxford: Blackwell, 1997), pp. 375–99, at pp. 376–77; and Alexander Rueger, “Physical Emergence, Diachronic and Synchronic,” Synthese 124 (2000), pp. 297–322, at pp. 317–18. For a defense of emergentism, see Timothy O’Connor, “Emergent Properties,” American Philosophical Quarterly 31 (1994), pp. 91–104, and John Dupré, “The Solution to the Problem of the Freedom of the Will,” Philosophical Perspectives 10 (Oxford: Blackwell, 1996), pp. 385–402. Jonathan Schaffer argues that there are emergent quantum phenomena in “Monism: The Priority of the Whole,” Philosophical Review 119 (2010), pp. 31–76. Carl Gillett develops a notion of strong emergence and provides possible illustrations in biology in “‘A Whole Lot More from Nothing But’: Scientific Composition and the Possibility of Strong Emergence,” delivered at the American Philosophical Association Meetings, Central Division, February 2009.

146 Consciousness and the Prospects of Physicalism

In Kim’s analysis, a further characteristic of emergent properties is that they cannot be reductively explained in terms of the physical basal conditions. He rejects Ernest Nagel’s “bridge law” conception of reductive explanation in favor of a model that first functionalizes the higher-level property to be reduced and then identifies that property with the realizer of that functionalization in the physical base. In this process, one might be interested in finding a specific realizer for a particular instance of the higherlevel property, or else in determining the general realizer for some species or structure type—what then results is a species-specific or a structure-specific reductionism. Finally, one finds a theory at the level of the physical base that explains how such realizers can instantiate the functional characterization of the higher-level property.64 In this conception, a property is emergent if it cannot be functionalized or if no realizer in the physical base can be found for particular instances of the property or for species or structure types—no entity in the physical base for which there is a theory that can explain how it can realize that property’s functional characterization. One might note that if there is to be a debate about whether nonreductivists might avoid a controversial sort of emergentism, this irreducibility condition should at least initially be viewed as necessary but not sufficient for a property’s being emergent. By Kim’s characterization, emergentism also endorses downward causation; that is, it claims that higher level property instances can have lower-level effects. (Kim raises a serious difficulty for the synchronic reflexive version of downward causation; here we will assume the diachronic variety.)65 Applied to the issue at hand, emergentism asserts that a mental property instance can cause a microphysical property instance. According to Kim, one problem with this claim again derives from causal exclusionary considerations. Suppose mental property instance M1 causes microphysical property instance P2. Then M1 will be realized by some microphysical property instance P1, M1 and P1 will compete as the cause of P2, and P1 will ultimately win out. Only by identifying M1 and P1 can M1’s status as cause be salvaged. Nonreductive physicalism might indeed countenance downward causation of this sort. Given that M1 causes M2, one might want to agree that M1 also causes M2’s realizer P2 (but note that my statement of the exclusion problem earlier crucially features the upward causation of M2 by P1 rather than the downward causation of P2 by M1). The nonreductivist can legitimately do so partly because one can reasonably hold that if M1 is constituted by P1, M1 and P1 will not compete as causes of P2. However, this is not by itself sufficient to render the nonreductive position radical in the sense that it is incompatible with the necessitation of mental property instances by basal conditions as Kim characterizes them. Downward causation would be radical in this way if it specified that higher-level property instances are (or have) causal powers that might result in contraventions of microphysical laws that can ideally be discovered without taking into account any higher-level properties—henceforth, ordinary microphysical laws. Timothy O’Connor

64 Jaegwon Kim, “Making Sense of Emergence,” pp. 10–11. 65 Jaegwon Kim, “Making Sense of Emergence,” pp. 28–31.

147 Robust Nonreductive Physicalism

provides an illustration of this idea; “If, for example, the multiple powers of a particular protein molecule were emergent, then the unfolding dynamics of that molecule at the microscopic level would diverge in specifiable ways from what an ideal particle physicist .  .  . would expect by extrapolating from a complete understanding of the dynamics of small-scale particle systems.”66 On this picture, if M1 were an instance of an emergent property, M1 could then cause P2 with the result that ordinary microphysical laws would be contravened. And as Randolph Clarke explains, such a capacity of an emergent property instance to contravene the ordinary microphysical laws would not be necessitated by a base that includes ordinary laws alone.67 The base would require, in addition, a fundamental emergence law. Suppose, for example, that the capacity for contravening the ordinary laws in a particular way is part of an emergent property’s essential nature. An instance of such a property would then not be necessitated by a base that includes only the ordinary laws. But as Clarke also points out, the nonreductivist is no more beholden to some factor that threatens to inhibit necessitation from ordinary basal conditions, such as the capacity of higher-level property instances to contravene the ordinary microphysical laws, than is the reductionist. An important difference between the views is that the nonreductivist contends that higher-level properties are often multiply realizable, and because of this, instances of higher-level properties will often not necessitate their actual basal conditions. But this fact does not undermine the necessitation of higherlevel property instances from ordinary basal conditions. More generally, I can find no feature of nonreductive physicalism per se that commits it to emergentism, or at least to the sort of emergentism that has generated controversy in the history of science and philosophy.

66 Timothy O’Connor, “Agent-Causal Power,” in Dispositions and Causes, ed. Toby Handfield (Oxford: Oxford University Press, 2008), pp. 189–214, at p. 195. 67 Randolph Clarke, “Nonreductive Physicalism and the Causal Powers of the Mental,” p. 309.

8 MENTAL COMPOSITIONAL PROPERTIES

In this chapter, I set out a model of the mental that is not functional in the standard sense, that is, a model in which the essences of types of mental properties do not consist in their causal relations to sensory inputs, behavioral outputs, and other mental states. Instead, mental properties—and this includes phenomenal properties—are identical to broadly physical compositional properties, properties things have solely by virtue of intrinsic features of their parts, either proper or improper, and relations these parts have to one another. This model would secure the causal efficacy of the mental qua mental in a way that the standard sort of functionalism cannot. It would preserve nonreductivism, since multiple realizability arguments indicate that mental compositional properties would not be essentially neural or microphysical. At the same time, given the identities that it affirms, in a significant respect the position espoused amounts to a compromise with the type-type reductionist views of U. T. Place and J. J. C. Smart.1 I close by considering several objections that have been raised against nonreductive views generally, arguing that in each case the model yields an adequate response.

COMPOSITIONAL, NOT FUNCTIONAL To qualify as robust, a nonreductive physicalism must satisfy two requirements. First, neither types nor tokens of mental causal powers can be identical to types or tokens of causal powers at a more basic level. Second, the mental must be causally efficacious qua mental. If, for example, mental causal powers are property instances, these property instances cannot be causally efficacious just by virtue of being realized by microphysical property instances that are causally efficacious. Rather, it must be that as mental that these property instances are causally efficacious. The model I propose meets each of these criteria. 1 U. T. Place, “Is Consciousness a Brain Process?” British Journal of Psychology 47 (1956), pp. 44–50; J. J. C. Smart, “Sensations and Brain Processes,” Philosophical Review 68 (1959), pp. 141–56. 148

149 Mental Compositional Properties

In the last chapter, we examined reasons to deny reductionism; let’s now turn to the issue of mental causal efficacy. Kim points out that nonreductive physicalists have typically endorsed a functionalist perspective, and he contends that this commitment undermines any claim for genuine mental causal efficacy, and I agree. Let’s begin with some preliminaries. First of all, functionalists of the sort Kim has in mind characterize mental states solely in terms of their causal roles, which are exclusively relational or extrinsic features of those states. More precisely, they are to be analyzed, metaphysically, solely in terms of causal relations to perceptual inputs, behavioral outputs, and other mental states, with no residue of intrinsic features. This version of functionalism is not the only one. To distinguish it from the others, Melnyk calls it external-relations functionalism; let it be ER-functionalism for short.2 Shoemaker’s functionalism, for example, differs significantly, and he, too, espouses a nonreductive position. In his view, mental properties—in fact, all properties—are individuated by their causal roles, but these properties confer causal powers on their bearers that are intrinsic to them, or, alternatively, these properties are identified with these intrinsic causal powers.3 Mental properties are also dispositions, and while dispositions are characterized extrinsically by their relations to manifestations and circumstances of manifestation, Shoemaker conceives them as intrinsic properties of their bearers. It’s not clear that the metaphysics of the position I am about to set out conflicts with Shoemaker’s, although, as will become clear, in my view, mental properties are not most directly specified by their causal roles, but instead as compositional properties. By way of raising a challenge to Kim’s species- or structure-specific reductionism, Ned Block once asked: “What is common to the pains of dogs and people (and all other species) in virtue of which they are pains?”4 According to ER-functionalism, what all pains would have in common, that by virtue of which they are all pains, is a pattern of such relations described by a causal-role specification—call it H. Given this understanding, Kim points out that in providing an answer to Block’s question, the local reductionist—the one who opts for species- or structure-specific reductionism— is in one crucial respect no worse off than this functionalist. Both are committed to the claim that there is no intrinsic property that all pains have in common, and both can specify only shared external-relational properties: The local reductionist must grant that on his view there is nothing intrinsic that all pains have in common in virtue of which they are all pains (assuming that Nh v Nr v Nm [i.e., various neural realizations of pain] have nothing in common). But that is also precisely the consequence of the functionalist view. That, one might say, is the whole point of functionalism: the functionalist, especially one who believes in MR [multiple realizability], would not, and should not, look 2 Andrew Melnyk, “Can Physicalism Be Non-reductive?” pp. 1281–96. 3 Sydney Shoemaker, “Causality and Properties” and in conversation. 4 Ned Block, “Introduction: What Is Functionalism?” in Readings in the Philosophy of Psychology, ed. Ned Block (Cambridge, MA: Harvard University Press, 1980), pp. 171–84, at pp. 178–79.

150 Consciousness and the Prospects of Physicalism

for something common to all pains over and above H (the heart of functionalism, one might say, is the belief that mental states have no “intrinsic essence”).5 The nonreductivist who opts for ER-functionalism will contend, however, that none of this entails a reduction of mental property M to a property in a realizing base, since M can be retained as an irreducible ER-functional mental property. But at this point Kim advances a highly significant objection to this functionalist move as a way of preserving nonreductive physicalism about the mental: How should we counter this line of argument? I think it will be helpful to consider the causal picture, and ask: What are the causal powers of this instance of M, namely [a system] S’s having M on this occasion? If S has M in virtue of M’s realizer Q, it is difficult to see how we could avoid saying this: the causal powers of this instance of M are exactly the causal powers of this instance of Q.6 Here Kim cites what we might call the Causal Inheritance Principle for functional properties: (CIP - FP) If a functional property M is instantiated on a given occasion in virtue of one of its realizers, Q, being instantiated, then the causal powers of this instance of M are identical with the causal powers of this instance of Q.7 In Kim’s view, the problem with the ER-functionalist’s proposal is that the causal powers of any instance of M will be causal powers in a nonmental realizing base. As a result, ER-functionalism cannot successfully defend the claim that there exist causal powers that are in the last analysis mental. Furthermore, Kim argues that given the multiple realizability of property M, the causal powers of the realizers of M will exhibit significant causal and nomological diversity, and for this reason the causal powers of M will feature such diversity. In his estimation, M will therefore be “unfit to figure in laws, and is thereby disqualified as a useful scientific property.” He concludes that the ERfunctionalist model cannot protect M as a property with a role in scientific laws and explanations.8

5 Jaegwon Kim, “Multiple Realization and the Metaphysics of Reduction,” p. 332. 6 Jaegwon Kim, “Making Sense of Emergence,” p. 16; I’ve substituted ‘M’ for Kim’s ‘E’ in this and the subsequent quotation. 7 Jaegwon Kim, “Making Sense of Emergence,” p. 16. I deny (CIP - FP) for the following reason. Suppose M is a mental functional property realized at the neural level by N, and N is actually realized at the microphysical level by P1 but is multiply realizable at the microphysical level by P1 and P2. (CIP - FP) would have it that the causal powers of this instance of M are identical with the causal powers of N and also with the causal powers of P1. By transitivity, the causal powers of N would be identical with the causal powers of P1, but by my argument in chapter 7, they will not be. 8 Jaegwon Kim, “Making Sense of Emergence,” pp. 17–18.

151 Mental Compositional Properties

To my mind, Kim’s argument against the compatibility of ER-functionalism and genuinely mental causal powers is compelling.9 The underlying general intuition is that a causal power, that by which an effect is produced, cannot be a causal relation but rather is in central cases an instance of an intrinsic property. I propose instead a strategy for securing mental causal powers that is motivated by this intuition, which therefore rejects ER-functionalism.10 A salient line of thought with implications that oppose ER-functionalism can be found in the deepest criticism of behaviorism that Putnam advances in his classic essay, “Brains and Behavior.”11 There he argues that we should characterize mental states in a way that conforms to our characterization of kinds in the natural sciences. In the case of the kind polio, for instance, we have found a biological explanation for the symptoms of the disease, which we might call its forward-looking causal relata, and we identify the disease with the underlying biological properties that provide this explanation. By contrast, certain types of behaviorists identify mental states with forward-looking causal relata, with observable behavior in particular, and not with underlying mental properties that explain observable behavior. Putnam recommends that we abandon such a behaviorism in favor of a conception that would characterize mental states on analogy with the biological example. But soon thereafter Putnam developed and endorsed a variety of ER-functionalism, and he came to expect that functional properties would yield explanations for the forward-looking causal relata of mental states much in the way that a viral infection provides an explanation for the symptoms of polio: My own view is that psychological predicates correspond to functional properties of human beings and other sentient beings. The presence of these properties explains the clustering of what some have called the ‘symptoms’ and ‘criteria’ of the various psychological states and conditions.12 Putnam’s idea is that the forward-looking causal relata of types of mental states have psychological explanations, and the explananda to be invoked are properties whose 9 Paul Churchland argues that the functionalist version of nonreductivism can preserve theories that should be eliminated, such as alchemy and phlogiston theory, as easily as it can preserve the mental states of commonsense psychology; see Paul Churchland, “Eliminative Materialism and the Propositional Attitudes,” Journal of Philosophy 73 (1981), pp. 67–90, at pp. 78–81. On a nonreductive physicalist view according to which mental states are irreducibly mental intrinsic properties, this type of concern will not arise. See also Thomas Polger’s critical discussion of functionalism in Natural Minds (Cambridge, MA: MIT Press, 2004), pp. 71–212. 10 Derk Pereboom, “Why a Scientific Realist Cannot Be a Functionalist” and “Robust Nonreductive Materialism.” 11 Hilary Putnam, “Brains and Behavior,” in his Philosophical Papers, vol. 2 (Cambridge: Cambridge University Press, 1975), pp. 325–41, first published in Analytical Philosophy, second series (Oxford: Blackwell, 1965), pp. 1–19. 12 Hilary Putnam, “Language and Reality,” p. 278. Putnam endorsed functionalism in “The Nature of Mental States.”

152 Consciousness and the Prospects of Physicalism

essences are exclusively causal relations to backward- and forward-looking causal relata. The problem is that this proposal would yield a variety of psychological explanation that is inadequate by the compelling scientific realist standards Putnam himself has espoused.13 What might it be for an instance of a property whose essence is a set of relations to backward- and forward-looking causal relata to explain one such forward-looking causal relatum? To oversimplify, the phenomenal property of being in pain might be functionally characterized as follows: Pain is the property whose instances are caused by pinpricks and pinches, cause the thought “I should avoid those stimuli from now on,” and given the belief that it’s OK to express one’s pain, cause winces and utterances of “ouch.” What happens when we now try to explain a wince, for example, by way of the essential features of pain as specified by this ER-functionalist characterization? One would be explaining the wince solely by relations to backward- and forward-looking causal relata, and in particular by a set that features the type of forward-looking causal relatum to be explained as one of its components. Such an explanation would be of a weak sort, and it would differ markedly from paradigmatic causal explanations by scientific realist criteria. To the extent that such ER-functional explanations are genuinely causal, Kim is exactly right to claim that they appeal only to nonmental causal powers.14 The realist model inherited from other sciences, to which Putnam appeals in “Brains and Behavior,” is best interpreted as explaining the forward-looking causal relata of kinds not simply by way of functional relations, but rather in central cases by properties intrinsic to those kinds—that is, properties intrinsic to every possible instance of the kind—and indeed, by intrinsic properties at the same level as the kinds themselves. In chemistry, the forward-looking causal relata of compounds are explained in part by their compositional properties—chemical properties intrinsic to those kinds of compounds, which each molecule of the compound has by virtue of the intrinsic properties of its component parts and the relations these parts have to one another. In biology, polio symptoms are explained partly by an intrinsic biological property of that kind of disease, being a particular viral infection. By analogy, the nonreductivist might advance the proposal that there are properties intrinsic to types of mental states that explain their forward-looking causal relata. Richard Boyd develops and defends a theory of natural kinds that is consonant with Putnam’s scientific realist claims. Boyd notes that on various antirealist views, natural kinds are characterized solely in terms of observable features. He then argues that a position of this sort fails to account for successful inductions based on natural kinds: it does not allow for an explanation for the high degree of projectibility of terms for

13 Derk Pereboom, “Why a Scientific Realist Cannot Be a Functionalist,” pp. 348–50. 14 Ned Block argues that (ER-) functional properties cannot be causally efficacious in standard cases (cases in which no intelligent being recognizes them); see “Can the Mind Change the World,” in Meaning and Method, ed. Ned Block (Cambridge: Cambridge University Press, 1990), pp. 137–70.

153 Mental Compositional Properties

these kinds. The remedy is to characterize natural kinds by underlying causal powers that serve to explain their observable features and thereby to build into the nature of these kinds grounds for the success of inductions that appeal to them. Boyd points out that in mature sciences, natural kinds are in fact typically characterized by such explanatory essences: Kinds characterized by “explanatory essences” are also kinds from the point of view of inductive generalization: indeed, in mature sciences, kinds which are explicitly characterized in terms of explanatory essences are the overwhelmingly typical cases of inductively natural kinds. Kinds natural from the point of view of successful induction need not always be explanatorily natural kinds, but they must correspond in relevant respects to the (perhaps unobservable) properties and mechanisms which causally determine the observable properties of the subjects of empirical generalizations.15 Given the paradigms we have for kinds in the natural sciences, explanatory essences would feature prominently properties that are intrinsic to kinds.16 In accord with such paradigms, the nonreductivist might endorse intrinsic mental properties, instances of which serve as specifically mental causal powers.17 What would such mental properties be like? First of all, despite the prevalence of ER-functionalism, it is quite natural to suppose that phenomenal properties are intrinsic to the states that have them, and also that the contents of propositional attitude states such as beliefs are intrinsic to those states. Even if belief-contents are partially extrinsically individuated, it remains natural to suppose that they are at least partially intrinsic features of beliefs. Moreover, we readily assume that behavior is causally explainable by way of intrinsic mental properties of these kinds. This model might be elaborated with the analogy of specific artifacts of certain internally complex types, for instance, a ball piston engine, a recent version of the rotary internal combustion engine. Characteristic of this engine is having parts with particular shapes and rigidities, and these parts being arranged in a particular way. These features are manifestly not external functional relations that such an engine stands in. Rather, they comprise a compositional property intrinsic to such an engine. Notice also that this property is multiply realizable. The parts of the engine can be made of materials of different sorts, as long as these materials can yield, for example, the required shapes and rigidities. A ball piston engine, then, has a distinctive

15 Richard Boyd, “Scientific Realism and Naturalistic Epistemology,” in Proceedings of the Philosophy of Science Association 2 (1980), pp. 613–62, at p. 642. 16 See Boyd’s “Kinds, Complexity, and Multiple Realizations,” Philosophical Studies 95 (1999), pp. 67–98, for his view of the prospects for psychological kinds fitting this model. 17 Robert Van Gulick’s conception of higher order patterns that “have a degree of independence from their underlying physical realizations” also inspires the sort of view I am developing here; cf. “Who’s in Charge Here? And Who’s Doing All the Work?” in Mental Causation, ed. John Heil and Alfred Mele (Oxford: Oxford University Press, 1993), pp. 233–56, at pp. 249–56.

154 Consciousness and the Prospects of Physicalism

non-ER-functionalist intrinsic compositional property, which nevertheless allows for different realizations.18 Similarly, it might be that the heterogeneous physical realizations of the dog’s and the human’s belief that this fire is dangerous exhibit a compositional property of a single type that is intrinsic to this kind of mental state, a compositional property instances of which are the causal powers of this belief.19 This property might well be more abstract than any specific sort of neural compositional property, given that it can be realized in distinct sorts of neural systems.20 Perhaps this same compositional property can also be realized in a silicon-based electronic system, and such a system could then have the belief about danger. Suppose one built a silicon-based system that replicated the capacities of and interconnections among neurons in a human brain as much as is physically possible and then excited it to mimic as closely as possible what happens when a human being has this belief about danger. It is a serious empirical possibility that this silicon-based state would realize the same belief and have an internal structure that, conceived at a certain level of abstractness, is similar enough to the internal structure of the ordinary neural system for both to count as instantiations of the same compositional property. It would seem unlikely that nothing of relevance would be alike in these systems other than relations to perceptual inputs, behavioral outputs, and other mental states. At least, in this case and more generally, we shouldn’t retreat to mere ER-functional resemblance prior to investigating whether the relevant similarities extend to intrinsic compositional properties. This position can be viewed as an instance of Yablo’s proposal that the relation between mental properties and their underlying physical properties is that of determinable to determinate.21 In the analogy, we can think of the distinctive compositional property of the ball piston engine as a determinable relative to which steel and titanium realizations of this property are determinates. Similarly, a mental compositional property can be construed as a determinable relative to neural and silicon-based realizations of that property. Considering mental properties as sufficiently abstract compositional properties in fact makes it natural to regard the relation between them and their neural or silicon-based realizations as that of determinable to determinate. Yablo’s proposal also intuitively fits with the constitution of mental property instances and causal powers by lower-level property instances and causal powers. An instance of the engine’s distinctive compositional property, its characteristic causal power, will be made up of, materially coincident with, and necessitated by its steel-realization property instance and causal power, and at the same time it is intuitive that the engine’s 18 The same point can be made with the rolling ball example of chapter 6 and with Putnam’s peg-and-board illustration in “Language and Reality,” pp. 295–98. If Philip Kitcher is right about the irreducibility of certain biological properties, analogous claims would seem to hold for them; see “1953 and All That: A Tale of Two Sciences.” 19 Beliefs of this sort are, I think, the best candidates for mental state types that are multiply realizable in human and animal brains. Sensations are another candidate, but against this, Thomas Polger argues that sensations might well differ along with the neurophysiology; see Natural Minds, pp. 12–16. 20 On abstractness of this sort, see Richard Boyd, “Kinds, Complexity, and Multiple Realization,” pp. 91–96. 21 Stephen Yablo, “Mental Causation.”

155 Mental Compositional Properties

compositional property is the determinable of which the steel realization is a determinate. The suggestion is that this is analogous in all relevant respects to the relation between mental property instances or causal powers and their underlying neural and microphysical counterparts. The resulting position has the advantages that Yablo cites; most generally, it allows us to see how mental and lower-level properties can be distinct and yet intimately related. How does this position differ from Kim’s view? The common ground includes the contention that if instances of mental properties are mental causal powers, these mental properties cannot be ER-functionally defined. But in addition, Kim allows for neural structure-specific reduction. For example, in his view there may be neural structures common to more than one species to which some class of instantiations of a mental state type can be reduced. I suggest we identify a mental state type with a compositional state type (a state type whose distinctive property is compositional) more abstract than any neural compositional type, one that can potentially be realized by a silicon-based system.22 This proposal is in fact an identity theory, but it is not reductionist, since it does not identify mental state types with a type of compositional state at a level of classification more basic than the mental itself. Kim, by contrast, envisions the reducing compositional types to be neural, or physical at a lower level yet. An important concern for my proposal is that the requisite compositional types may not exist—that in general, no significantly homogeneous compositional types correlate with what are intuitively the mental state types. For instance, it may be that higher-level compositional types that realize the belief that there is danger nearby differ on the order of the way in which a cat and an ordinary mousetrap differ as instantiations of being a mouse-catcher.23 This would constitute a serious challenge to the robust nonreductive physicalist view, but it would not yet be decisively undermined. It may be that these compositional types, although they fail to correlate neatly with our ordinary mental state categories, are not specifically neural compositional types either. In that case, one might take advantage of what room there is for altering the ordinary system for classifying mental states, at least for the purpose of scientific psychology, and identify the distinct compositional types with distinct mental state types. Scientific reclassifications relative to ordinary categories are, after all, not unusual. Still, it may also turn out that in general, the only significantly homogeneous compositional types to be found are essentially neural compositional types. In that case, it would be hard to see how there could yet be irreducibly mental causal powers, and my sense is that Kim would then be right. However, that result does not appear especially likely, given the thought that the internal structure of a

22 A more general allied claim is advanced by Frederick Adams in “Properties, Functionalism, and the Identity Theory,” Eidos 1 (1979), pp. 153–79: “Just because two systems are different kinds of stuff does not mean that they do not share some identical property-kinds” (p. 158); see Thomas Polger’s discussion in Natural Minds, pp. 2–12. 23 See also Richard Boyd, “Kinds, Complexity, and Multiple Realization,” pp. 71–72.

156 Consciousness and the Prospects of Physicalism

state of a silicon-based system, conceived at some suitably high level of abstractness, might well be similar enough to the internal structure of a neural realization of a mental state for both the silicon-based and neural states to count as instances of the same (mental) kind. A further challenge to this proposal derives from arguments for the externalist individuation of mental states. If externalism is true, then properties such as being the belief that there is water in the glass will not be intrinsic properties of mental states. My own predilection is to resist arguments for externalism, although I will not take on this issue here.24 But even if mental states, due to externalist individuation, have essential extrinsic properties that involve relations to entities external to those states, it remains open that the core of the mental state is an abstract compositional property. If I became convinced of a variety of externalism, this is the sort of position I would defend.25 Finally, one might object that it is metaphysically possible that nonphysical beings have mental states, and the model I’ve presented precludes this possibility, since nonphysical beings cannot have physical compositional properties. Since we don’t have a clear conception of a nonphysical substance, perhaps we can’t rule out that such substances can instantiate the proposed compositional properties, whereupon those properties would be so abstract that they are not even essentially physical. But if, as seems much more likely to me, those compositional properties are essentially physical, our states and those of a nonphysical being might still be functionally similar, even if the nonphysical being’s states are not genuinely of the same kind as ours. This response accords with the widespread theological tradition, according to which God’s states are significantly analogous to ours but not really of the same type. It’s puzzling to me that realists about the mental have so readily endorsed ER-functionalism, a model for the nature of mental kinds so closely tied to a general antirealist point of view. Positivist antirealists advocated an operationalist characterization of natural kinds, defining them in terms of their observable backward- and forward-looking causal relata. Logical behaviorism provides a good example of this practice. However, scientific realism rejects such an operationalist conception on the ground that it has instances of natural kinds entering into causal relations without those kinds being defined by those relations. Although ER-functionalist characterizations of mental kinds are more sophisticated than those of its behaviorist progenitor, ER-functionalism nevertheless fits squarely within this antirealist tradition. My alternative proposal is not novel in spirit. It simply recommends for mental kinds what realists typically advocate for natural kinds generally. 24 I argue for internalism about psychological content, and thus against extrinsic individuation, in Derk Pereboom, “Conceptual Structure and the Individuation of Content.” Tyler Burge develops the case for externalism in “Individualism and the Mental” and in “Other Bodies.” 25 I suggest a position of this sort in Derk Pereboom, “Why a Scientific Realist Cannot Be a Functionalist,” Synthese 88 (1991), pp. 341–58, at p. 351.

157 Mental Compositional Properties

MENTAL CAUSAL EFFICACY DEFENDED A view of this type allows instances of mental properties to be causally efficacious as mental properties, and not just by virtue of having lower-level realizers that are causally efficacious. Alyssa Ney raises a challenge for nonreductive physicalisms that aspire to this sort of mental causation. She contends that on any nonreductive account in which the relation between the mental and the physical is tight enough to avoid redundant overdetermination, mental properties instances qua mental will fail to do causal work, since all of it will be accomplished by the microphysical properties. I maintain that conceiving of mental properties as multiply realizable abstract compositional properties provides a model for answering this concern.26 Ney develops this challenge in response to L. A. Paul’s nonreductive account of the relations between the mental and the microphysical. In this account, mental token state M and microphysical token state P, each of which is a bundle of property instances, are distinct due to having diverse modal properties, but they nonetheless share a core of property instances, such as having a mass of a certain magnitude and being made up of particles arranged in a particular way. Here is an example Paul cites: Consider protein Pro, constituted by sum of molecules Mol. The nonreductionist should hold that the property instances of being protein Pro is a complex property instance that is really just a conjunction of many more fundamental property instances such as having shape s, having mass m, . . . etc. Now consider the property instance of being sum Mol: it is a complex property instance that is just a conjunction of many more fundamental property instances such as having shape s, having mass m, . . . etc. . . . the property instance of being Pro and the property instance of being Mol share some of their conjuncts. And . . . it is the shared (instances of) conjuncts that are responsible for the problematic cases of putative constitutive overdetermination.27 Ney notes that in Paul’s view, the property instances that are shared by the mental and physical events are the causally efficacious ones, while those that do not overlap are causally irrelevant modal property instances, such as being possibly constituted by different neural property instances or being essentially intentional. But now, she argues, the property instances that make the mental token mental—the ones without which it would just be a physical token—are causally inefficacious. Ney contends that on Paul’s account, it’s not by virtue of the mental token’s being mental that anything gets done and that the property instances that are causally efficacious are shared by the microphysical token. When one looks closely at the relation between the mental and the microphysical, it seems clear that all of the causal powers are really microphysical after all, since all of the causally efficacious property instances are shared by the microphysical token. 26 Alyssa Ney, “Can an Appeal to Constitution Solve the Exclusion Problem?” Pacific Philosophical Quarterly 88 (2007), pp. 486–506. 27 L. A. Paul, “Constitutive Overdetermination.”

158 Consciousness and the Prospects of Physicalism

The challenge for me is to show that if the relation between mental and microphysical tokens is constitution, mental entities qua mental can nevertheless be causally efficacious. Suppose that M1 and M2 are mental property instances and M1 causes M2. M1 is constituted by microphysical property instance P1, and M2 is constituted by microphysical property instance P2, such that P1 causes P2. M1 is not an ER-functional property instance but rather a compositional property instance, and so there will be no objection based on the sort of property M1 is to its being causally efficacious. The account provides a positive model of a property instance that is causally efficacious qua mental: an instance of a requisitely abstract compositional property. Multiple realizability arguments show that M1 is not identical to P1 or to any component of P1, and no component of M1 is identical to a component of P1. Instead, the relation between these mental and microphysical entities is best understood as constitution as I’ve defined it. Since constitution does not involve the identity of any mental and any microphysical entity, Ney’s criticism of Paul’s view won’t transfer. It won’t be the case that all of the causally efficacious property instances or causal powers are really microphysical.

MULTIPLE REALIZABILITY AND PROJECTIBILITY It was once commonly supposed that nonreductive views about the special sciences are grounded most fundamentally in the phenomenon of multiple realizability by way of a fairly formal argument. Kinds in the special sciences can be realized in different ways from the perspective of lower-level sciences, and thus an attempt to reduce higher-level kinds, laws, and explanations to those at a lower-level will involve replacement by disjunctive properties—properties that are perhaps even wildly disjunctive in the sense that the disjuncts have at best little in common. Moreover, the disjunctions that these properties feature might even be open-ended or infinite. The received wisdom was that such disjunctive properties are not kinds, for the reason that statements of regularities involving such disjunctive properties fail to be laws and, perhaps most fundamentally, because “explanations” involving such disjunctive properties are not genuine explanations. This standard argument for nonreductive physicalism appears to rely on a certain formal prescription for laws and explanations, that they cannot contain disjunctive properties or at least not wildly disjunctive properties. In turn, as I noted in the previous chapter, this formal prescription is at least sometimes rooted in subjective interests or else in human limitations.28 Kim argues, however, that a term for a higher-level property is precisely as projectible as the disjunction that expresses its multiply realizable character at a more basic level, and thus a generalization involving such a disjunction is exactly as lawlike as the higher-level generalization that it was meant to reduce.29 The underlying 28 Jerry Fodor, “Special Sciences.” Hilary Kornblith and I endorse a version of this argument in “The Metaphysics of Irreducibility,” pp. 126–28, against which William Jaworski advances a counterargument in “Multiple Realizability, Explanation, and the Disjunctive Move,” Philosophical Studies 108 (2002), pp. 298–308, especially pp. 301–3. 29 Jaegwon Kim, “Multiple Realization and the Metaphysics of Reduction,” pp. 319–25.

159 Mental Compositional Properties

reason is that a higher-level property is nomically equivalent to a disjunctive property. Nomic equivalence might be defined in this way: Properties F and G are nomically equivalent if they are coextensive in all possible worlds compatible with the laws of nature.30 If Kim is right, then the formal argument does not appear to go through, for it relies on the possibility that generalizations involving a higher-level property are lawlike, while those involving the corresponding disjunctive property are not. But furthermore, Kim contends that wildly disjunctive predicates are not projectible, and hence terms for higher-level properties that are nomically equivalent to corresponding disjunctive properties are not projectible either. As a result, such higher-level terms cannot figure into laws, and they do not pick out genuinely scientific kinds. The example of a disjunctive property Kim adduces to make this last point is jade. ‘Jade’ is a category that comprises two mineralogical kinds, jadeite and nephrite, and hence jade is the same property as the disjunction jadeite or nephrite. From this, we can conclude that the term ‘jade’ will not be projectible. Suppose, Kim argues, that we’re trying to confirm the generalization ‘jade is green.’ We might check many instances of jade and find that they are all green. But it could be that the entire sample consists of jadeite and no nephrite. We must conclude that the generalization is not confirmed, and thus ‘jade’ is not projectible.31 To clarify his claim, Kim considers the objection that we can think of genuinely projectible kind terms as picking out disjunctive properties. ‘Emerald,’ for example, can be thought of as picking out being an African emerald or non-African emerald. But, he says, this possibility fails to undermine the projectibility of ‘emerald’; for example, it doesn’t show that there is anything wrong with the lawlikeness of “All emeralds are green.” However, this analogy does not serve to reinstate the projectibility of ‘jade,’ for, by contrast with ‘jadeite or nephrite,’ the disjunction, “being an African emerald or non-African emerald,” does not denote some heterogeneously disjunctive, nonnomic kind; it denotes a perfectly well-behaved nomic kind, that of being an emerald! There is nothing wrong with disjunctive predicates as such; the trouble arises when the kinds denoted by the disjoined predicates are heterogeneous, “wildly disjunctive,” so that instances falling under them do not show the kind of “similarity,” or unity, that we expect of instances falling under a single kind.32

30 Ned Block, “Anti-Reductionism Slaps Back,” in Philosophical Perspectives 11 (Oxford: Blackwell, 1997), pp. 107–32, at p. 109. 31 Jerry Fodor argues that Kim’s example merely involves a sampling error; see “Special Sciences: Still Autonomous after All These Years,” Philosophical Perspectives 11 (Oxford: Blackwell, 1997), pp. 149–63, at pp. 151–52. 32 Jaegwon Kim, “Multiple Realization and the Metaphysics of Reduction,” p. 321; cf. Lenny Clapp, “Disjunctive Properties: Multiple Realizations,” Journal of Philosophy 109 (2001), pp. 111–36, at pp. 120–21.

160 Consciousness and the Prospects of Physicalism

But notice that given these reflections even jade might turn out to be a fairly wellbehaved nomic kind after all. As Block points out, all samples of jade share certain appearance properties, similarities that give rise to a certain degree of projectibility.33 In Block’s view, more generally, predicates that pick out properties that are multiply realizable can yet be projectible with respect to “properties of channeled selection, learning, and design.”34 Because there are typically only a few ways in which entities of a particular higher-level type can be designed and produced, we can expect relatively broad similarities among these things that would render terms for corresponding higher level properties significantly projectible.35 The point to extract from this debate is that the considerable heterogeneity of the possible realizations of a property is compatible with their nevertheless having significant features in common, features that will undergird the significant projectibility of a term that refers to the property.36 This point is consistent with Kim’s claim that a term that refers to a higher-level property is precisely as projectible as the disjunctive predicate that stands for all of its possible realizations. One should not conclude from the heterogeneity of the possible realizations of a higher-level property that there is no feature that can sustain the projectibility of a term that picks it out—in fact, of terms for both the higher-level property and for the disjunctive property that comprises all of its possible realizations. The projectibility-sustaining feature of a kind could be a compositional property that is significantly homogeneous across its heterogeneous realizations, a property that might be a unitary type of causal power at the level of description of the kind. Note that disjunctive terms will typically fail to express or will at least mask any such homogeneous compositional properties and unitary types of causal powers to which they might correspond. In the case of the kind ball piston engine, for example, a disjunctive term that details its possible realizations would fail to express or would at least mask the distinctive compositional property that sustains its projectibility. By contrast, the term ‘ball piston engine’ itself can serve to express this compositional property without obscuring it. But note that one cannot conclude merely from the fact that a term for a property is projectible that it is an intrinsic compositional property that is a unitary type of causal power at the level of description of the property. Terms that pick out ER-functional properties, for example, may be projectible, while these properties are neither intrinsic to their bearers nor unitary types of causal powers. ‘Being soluble’ is projectible, yet although for any instance of solubility there will be a causal power that helps account for the projectibility of this term, solubility itself is not a unitary type of causal power.

33 Ned Block, “Anti-Reductionism Slaps Back,” pp. 126–27. Louise Antony and Joseph Levine make a similar point in “Reduction with Autonomy,” Philosophical Perspectives 11 (Oxford: Blackwell, 1997), pp. 83–105, at pp. 90–91. 34 Ned Block, “Anti-Reductionism Slaps Back,” pp. 120–29. 35 Louise Antony and Joseph Levine, “Reduction with Autonomy,” pp. 92–94. 36 See also Lenny Clapp, “Disjunctive Properties: Multiple Realizations,” pp. 123–32. Clapp explains how, on a causal-powers notion of properties, a disjunctive predicate might indicate a nondisjunctive property.

161 Mental Compositional Properties

The tie between projectibility and unitary types of causal powers is therefore not tight. If a term for a property is strongly projectible, it may yet refer to a functional property with no intrinsic features that can result in a unitary type of causal power, let alone a unitary type of causal power at the level of description of the property. But nevertheless, the weakness of a term’s projectibility counts as evidence that the referent of the term is not a unitary type of causal power. Consider an instructive example of Kornblith’s. In 1869, the term ‘neuraesthenia’ was introduced to designate a nervous disease that results in severe fatigue—a characterization in terms of its symptoms. The term was soon established worldwide, but “like most descriptive terms, where basic organic or psychological understanding was lacking, it tended to be overinclusive and a receptacle for many diverse conditions.”37 But when cures for neuraesthenia were sought, it was found that different sorts of causes had to be treated. Several distinct sorts of underlying causes were discovered for its symptoms. As a result, the term ‘neuraesthenia’ became obsolete by about 1930. What makes us think that neuraesthenia does not correspond to a unitary type of causal power at the level of description of this kind? First, there is the evidence that the term ‘neuraesthenia’ is not projectible to a high degree. In addition, researchers discovered a disjunction of properties coextensive with being neuraesthenia, terms for each of which is more strongly projectible. Explanations involving these properties effectively replaced those involving neuraesthenia. Moreover, the characterization of neuraesthenia was forced to remain functional because no homogeneous underlying properties were discovered across its instances that could explain its symptoms. Accordingly, I would suggest that whether there is good evidence that mental states instantiate unitary and specifically mental types of causal powers depends on whether terms for mental state types are projectible to a high degree; on the failure of a search for coextensive sets of properties, terms for which are more strongly projectible; and on whether specifically mental explanatory essences—intrinsic mental properties in central cases—can be found. In short, whether there exists good evidence of this sort depends on whether there are powerful, resilient, and thoroughly realist psychological explanations in which mental state types play a part.

THE SIGNIFICANCE OF MULTIPLE REALIZABILITY We can conclude that a nonreductivist would be ill-advised to suggest that the multiple realizability of a mental type indicates that a disjunctive term for realization base is only weakly projectible, by contrast with the term for the mental type itself, and that for this reason mental types do not reduce to their realization bases. Kim is right to claim that the strength of the projectibility of the terms for the mental type and the realization base would not differ. But one should not draw the more general conclusion that the nonreductivist’s case cannot be supported by the phenomenon of multiple 37 Hilary Kornblith, Knowledge without Foundations: A Causal Theory, Dissertation, Cornell University, 1979.

162 Consciousness and the Prospects of Physicalism

realizability. As I argued in the last chapter, multiple realizability serves to preclude various identity claims of both the type and token varieties. A critical examination of further objections against the value of multiple realizability for nonreductive physicalism will clarify this role. We’ll first consider the multiple realizability of mental types and then turn to mental tokens.

Types Lawrence Shapiro is skeptical about any such significance for multiple realizability of mental state types: Take what appears to be a legitimate case of multiple realization. . . . Either the realizing kinds truly differ in their causally relevant properties, or they do not. If they do not, then we do not have a legitimate case of multiple realizability. . . . If the realizing kinds do genuinely differ in their causally relevant properties, then they are different kinds . . . and so we do not have a case in which a single kind has multiple realizations.38 To illustrate the notion of a causally irrelevant property, Shapiro points to the color of a corkscrew. Corkscrews can be gray or black, for example, but the color of a corkscrew is causally irrelevant to its nature—in this case, to what it does, and is designed to do. He argues on this basis that differences in color among corkscrews do not amount to a legitimate case of multiple realization. Shapiro extends the claim of causal irrelevance to an example involving neural and silicon-based realizations of a mind. “If each neuron’s contribution to psychological capacities is solely its transmission of an electrical signal, and if silicon chips contribute to psychological capacities in precisely the same way, then the silicon brain and the neural brain are not distinct realizations of a mind.”39 His thought is that the sameness in contribution to psychological capacities screens off the difference between neurons and silicon chips and makes it the case that they are not distinct realizations of these capacities. Genuine cases of multiple realization of E would feature realizations of E that differ in their causally relevant properties, and realizations differ in their causally relevant properties only when they make distinct contributions to the nature of E. But first of all, would it be implausible to claim that the identical contributions to psychological capacities made by neural and silicon-based systems do amount to a genuine case of multiple realizability while denying this of the gray and the black corkscrews? No. The color of a corkscrew is causally irrelevant in a starker sense than the one Shapiro has in mind, since its color makes no positive causal contribution whatever to its nature—in this case, to what it does, never mind different colors not making distinct causal contributions to what it does. Shapiro’s characterization extends causal irrelevance to pairs of realizers, each of which in fact makes a causal contribution to the 38 Lawrence Shapiro, “Multiple Realizations,” Journal of Philosophy 97 (2000), pp. 635–54, at p. 647. 39 Lawrence Shapiro, “Multiple Realizations,” p. 645.

163 Mental Compositional Properties

nature of the thing, whenever each makes the same causal contribution. This characterization fails to count as different realizations pairs of distinct realizers, each of which does in fact make a causal contribution to the nature of the thing that has it. So he says: “Steel and aluminum are not different realizations of the waiter’s corkscrew because, relative to the properties that make them suitable for removing corks, they are identical.”40 But unlike color, being made of steel or of aluminum does make a causal contribution to what a corkscrew does; in this respect, these properties are causally relevant in a way in which colors are not. Suppose that an effective corkscrew can only be made either of steel or of aluminum, because only these materials have the right kind of rigidity, yet each material makes exactly the same contribution to what it does. The fact remains that these materials, as opposed to any others, have the right kind of rigidity. Thus making a causal contribution to the nature of the thing that has it might be the notion of causal relevance that is pertinent to a condition on multiple realization.41 This alternative conception would license steel and aluminum but not distinct colors as genuine multiple realizations of a corkscrew, and for silicon-based and neural systems to count as genuine multiple realizations of psychological features. Why adopt this alternative conception of multiple realizability? Perhaps it is enough to point out that distinct realizers can make the same positive causal contribution to the nature of a thing, and this is just what we mean when, in debates about reductionism, we talk about multiple realizability. But in addition, the fact that the neural system and the silicon-based system make identical contributions to psychological capacities seems to force us to say that the features thus contributed are neither essentially neural nor essentially silicon-based. Here there is significant work for multiple realizability to do: because some one type of thing can have realizations of distinct types F and G, it can be characterized neither as essentially F nor as essentially G. Shapiro provides no good reason to deny any of this. But given his conception, the realizations in this example will not really be multiple, and thus his conception fails to allow multiple realizability to do this work—which it can in fact do. For this reason, his conception of multiple realizability is best rejected in favor of the proposed alternative. What then is the legitimate role for multiple realizability in supporting nonreductive physicalism? The answer is implicit in the previous discussion. Whether a property is multiply realizable can indicate the level at which it should be classified. Is the kind corkscrew a kind of steel thing? No, for it also has a possible aluminum realization. Is the kind mind a neural kind of thing? If mental states are also realizable in silicon, then it isn’t. Multiple realizability provides the key to precluding classification of mental states as essentially neural or as classified essentially at some lower level yet. A number of reductionists have provided appreciable reason to think that realizability in different kinds of neural systems alone need not advance the cause of robust nonreductive physicalism. Suppose that what would seem to be a single mental state type were realizable only neurally, albeit in neural systems that differ, such as a human’s, 40 Lawrence Shapiro, “Multiple Realizations,” p. 644. 41 Carl Gillett makes a similar point in “The Metaphysics of Realization, Multiple Realizability, and the Special Sciences,” pp. 596–600.

164 Consciousness and the Prospects of Physicalism

a dog’s, and an octopus’s. Imagine first that a single compositional property was found that is intrinsic to this mental state type. This property might well count as neural and not as irreducibly mental, for as William Bechtel and Jennifer Mundale point out, within the realm of the neural itself, there are possibilities for classification at different levels of abstractness or coarse-grainedness, and the property might well be characterizable as neural at some sufficiently high level of abstractness.42 Then we would not have multiple realizability at the neural level after all. In addition, as Oron Shagrir argues, if thoroughly distinct neural realizations were discovered for what initially appears to be a single mental state type, then it might be that these realizations correspond to what are in fact different mental state types43 The stronger argument against reductionism invokes the nonneural realizability of mental types. If a single mental type is realizable both in neural and in silicon-based systems, then they are not essentially neural types at any level or, for that matter, essentially microphysical types, since the neural and the silicon-based realizations will differ microphysically. Patricia Churchland and Paul Churchland have argued that the multiple realizability of psychological types by different neural types and by nonneural types does not undermine reductionism, for the reason that reductionism might be “domain-specific”: . . . visual experience may count as one thing in a mammal, and a slightly different thing in an octopus, and a substantially different thing in some possible metaland-semiconductor android. But they will all count as visual experiences because they share some set of abstract features at a higher level of description. That neurobiology should prove capable of explaining all psychological phenomena in humans is not threatened by the possibility that some other theory, say, semiconductor electronics, should serve to explain psychological phenomena in robots. The two reductions would not conflict. They would complement each other.44 If indeed visual experience in humans and in mammals had only functional characteristics in common, then the claim that they have distinct reductions would be 42 William Bechtel and Jennifer Mundale, “Multiple Realizability Revisited: Linking Cognitive and Neural States,” Philosophy of Science 66 (1999), pp. 175–207, at pp. 201–4; cf. Jaegwon Kim, “Phenomenal Properties, Psychophysical Laws, and the Identity Theory,” Monist 56 (1972), pp. 190–91, at p. 190; Laurence Mucciolo, “The Identity Thesis and Neurophysiology,” Noûs 8 (1974), pp. 327–42. 43 Oron Shagrir, “Multiple Realization, Computation, and the Taxonomy of Psychological States,” Synthèse 114 (1998), pp. 445–61, at pp. 451–52. 44 Paul M. Churchland and Patricia S. Churchland, “Intertheoretic Reduction,” in their On the Contrary (Cambridge, MA: MIT Press, 1998), pp. 65–79, at p. 78. They cite temperature as an example of a property that has been domain-specifically reduced; their claim is that temperature is reduced differently in a gas, a solid, and in a vacuum. In my view, the differences among temperatures in a gas, a solid, and a vacuum indicate that temperature is multiply realizable but not that it has distinct reductions. As Kornblith and I have argued, there is a unitary property that these cases of temperature have in common, with which temperature should be identified; see Derk Pereboom and Hilary Kornblith, “The Metaphysics of Irreducibility,” pp. 138–39.

165 Mental Compositional Properties

plausible, for these distinct varieties of visual experience would not share causal powers. However, if these varieties of visual experience were to share a compositional property, then they might well share causal powers that are essentially neither neural nor electronic but rather psychological or mental. Visual state types could then be mental state types that conform to the robust nonreductive physicalist conception. A key motivation for the Churchlands’ remarks is their sense that it is neuroscientific research that will reveal the nature of human psychology. But there is a natural way of understanding this motivation so that it is consistent with the robust nonreductivist position. If there are irreducibly mental compositional properties, then a very likely avenue for discovering them would indeed be research in neuroscience. Even if such compositional properties could also be realized in silicon-based electronic devices, it is highly plausible that we would first learn about their nature through neuroscientific research. But the upshot is not metaphysical neural reductionism, but rather an epistemic claim about how the nature of the mental is likely to be ascertained. In Paul Churchland’s most recent exploration of these issues, he proposes a model that in its metaphysical essentials I can fully endorse, with only verbal differences (if any) remaining.45 Minds are engines of a specific type, a sort that exploits entropy-increasing energy flow to produce information.46 The specifics of the model for such an engine “can be realized in a wide variety of material substrates: in mammalian brains, in octopus brains, in extraterrestrial brains, in electronic chips, in optical systems, and so forth.”47 Even though our discovery and initial articulation of these specifics will be through neuroscience, “largescale electronic realizations . . . will explore entirely new horizons for information processing and world representation.”48 All of this is consonant with the model I propose. Churchland would describe his model as reductionistic, since it reduces mental types to physical information-producing engine types. But I welcome this sort of reduction. However, it is not a reduction to a physical level more basic than the mental, such as the neural or microphysical. Such reduction instead amounts to the identification of a mental property with a physical property more abstract than the neural, which most nonreductive physicalists have always allowed. Indeed, almost all have endorsed the identification of mental properties with ER-functional properties at a level more abstract than the neural. True, some have thought of ER-functional properties as not essentially physical, and so identification with such a functional property would not be reduction to a physical property. Still, this is not where nonreductivists have drawn the dividing line. Instead, it’s been between those who identify mental properties with physical properties at the neural, biological, chemical, or microphysical levels, and 45 Paul Churchland, “Functionalism at Forty: A Critical Retrospective,” Journal of Philosophy 102 (2005), pp. 33–50. 46 Paul Churchland, “Functionalism at Forty: A Critical Retrospective,” pp. 43–50. 47 Paul Churchland, “Functionalism at Forty: A Critical Retrospective,” p. 46. 48 Paul Churchland, “Functionalism at Forty: A Critical Retrospective,” p. 49.

166 Consciousness and the Prospects of Physicalism

those who claim that mental properties cannot be identified with those at such more basic levels.49

Tokens The identity of mental and neural types yet leaves open the identity of mental and neural tokens. This raises a critical issue for my position, for if token mental causal powers in particular were identical with token neural causal powers, a robust version of nonreductive physicalism about the mental would be precluded. I set out my argument against identity of such token causal powers in the previous chapter, and I will now modify it to draw a more general conclusion. The claim that mental tokens are identical with neural tokens can resist certain kinds of arguments from multiple realizability. Stable tokens often retain their identity over changes in their constitutions and configurations, and, significantly, they enjoy a degree of resilience in the production of their characteristic effects under these changes. My decision to ring the doorbell can survive small changes in a microphysical token that constitutes it, and nature has probably endowed it with resilience in producing its characteristic effects under these changes. But this is also the case for a neural token that realizes it—it, too, can survive small changes in the microphysical token that constitutes it. Establishing that a mental token M is multiply realizable at the microphysical level will therefore not suffice to show that it is distinct from a neural token N that constitutes it, for it is open that N is similarly multiply realizable at the microphysical level, whereupon it could yet be identical with M. One might argue instead that M, which is in fact currently realized by N, could have been realized by a token of a distinct neural type from N’s, and given that N would not be identical with a token of such a distinct neural type, M would not be identical with N. But here it might be difficult to defeat the counterclaim that M is identical with a neural token at a level sufficiently abstract to accommodate each of the neural types invoked. The more convincing argument invokes the thought, encountered in the preceding chapter, that M could have been realized by a token that is partly neural and partly silicon-based. Since a neural-and-silicon token would not qualify as a neural token, it could not be identical with N, and M could then also not be identical with N. So suppose that at some time in the future we are capable of fitting patients who are, say, threatened by impending brain damage with silicon-based prostheses. Consider Anne’s token belief at a particular time t that her parents live a short subway ride from Grand Central Station, a mental token actually realized by a neural token. Her belief could plausibly have been the same mental token at t had she undergone the brain surgery the week before and had her belief been realized at t by a neural-and-silicon token instead. It’s also credible that she could retain this token belief after it comes to be realized by a neural-andsilicon-based replacement, whereupon one and the same mental token would initially be realized by a neural token and later by a neural-and-silicon token. Thus Anne’s mental token is distinct from any neural token. Since this argument generalizes to mental tokens 49 Thanks to Jonathan Cohen for discussion of these issues.

167 Mental Compositional Properties

of any sort, we can conclude that all mental-neural token identity claims are false. Again, a nonidentity claim would be established by an argument from multiple realizability.

NONREDUCTIVE PHYSICALISM AND EXPLANATION While the focus in this discussion has been on the metaphysical issues, robust nonreductivism has consequences for methodology and explanation in psychology. To provide a causal explanation of an event is to cite its causes, and so if an event has psychological causes that are not identical with physiological causes, its psychological causal explanation will not reduce to a physicological explanation. This claim is true for both type-level and token-level explanations. Patricia Churchland and Paul Churchland suggest, however, that nonreductive physicalism is mistaken because psychological phenomena often demand physiological explanations.50 The psychological effects of drugs, brain lesions, sleep, and fainting are compelling examples that should motivate us to accept some form of reductionism. In response, as Kornblith and I have argued, nonreductive physicalism is consistent with three conciliatory claims about the relation between psychology and more basic sciences. 51 First, nonreductive physicalism accommodates the claim that certain psychological effects are best causally explained by causes in a more basic science, and not by psychological causes. By analogy, even if classical genetics is not reducible to molecular biology, certain kinds of mutations in genes are plausibly best explained by changes in molecular structure, and not by genetic factors. Indeed, when the feeling of pain is caused by a pinprick, it’s a nonpsychological event that explains a psychological event, the feeling of pain. The loss of psychological functioning by someone who has undergone a lobotomy is best explained in terms of neural damage, and not by a psychological cause. Hallucinating upon ingesting LSD is also best explained by a mechanism more basic than the psychological. In general, tokens of a single psychological type may regularly be caused by tokens of a single type at the physiological level, while there is no psychological type that corresponds to the physiological type. At the same time, the type of psychological effect may not reduce to a physiological type, due to multiply realizability at this lower level. This situation would give rise to an interlevel law that does not reduce to a physiological law, which the nonreductivist can admit. Second, nonreductive physicalism is compatible with the reducibility of some apparently psychological kinds to kinds in a more basic science. For example, it could turn out that tiredness is nothing more than a single type of physiological phenomenon, and perhaps psychological explanations involving tiredness will be illuminated and deepened when recast as involving this physiological phenomenon. But the reducibility of some apparently psychological kinds to those in a more basic science

50 Patricia Churchland, Neurophilosophy:Toward a Unified Science of the Mind-Brain (Cambridge, MA: Bradford Books/MIT Press, 1986). 51 Derk Pereboom and Hilary Kornblith, “The Metaphysics of Irreducibility.”

168 Consciousness and the Prospects of Physicalism

does not all by itself threaten the irreducibility of most psychological kinds, such as the various sorts of beliefs and desires. Third, nonreductivism is consistent with the admission that some apparently psychological laws are reducible to laws at a more basic level. For example Tiredness typically causes sleep is prima facie a psychological law. But perhaps both tiredness and sleep are really physiological kinds, as a result of which this law will be reducible to one at a more basic level. Furthermore, it may be that when in theories the terms ‘tiredness’ and ‘sleep’ are replaced by physiological descriptions, explanations are illuminated and deepened rather than obscured. However, even if this turns out to be correct for some such laws, no argument is in place for the reducibility of all or most apparently psychological laws to laws at a more basic level.

PROSPECTS FOR ROBUST NONREDUCTIVE PHYSICALISM I have proposed that mental properties are compositional properties, that is, properties things have solely by virtue of intrinsic features of their parts and relations these parts have to one another. These compositional properties would be intrinsic to the states that have them, and they are in this respect importantly different from ER-functionalist properties. In addition, I have argued that mental compositional properties are distinct from any nonmental properties, and this makes the view nonreductive. On this proposal, phenomenal properties are also compositional properties. Thus I do not advocate an ER-functionalist version of either the qualitative inaccuracy account of phenomenal properties developed in chapters 1 through 4 or the Russellian monist alternative examined in chapters 5 and 6. In the case of Russellian monism, the nature of absolutely intrinsic features will be crucial to the compositional properties with which phenomenal properties are to be identified. Kim and others have raised a number of strong challenges to the nonreductive physicalist position. The most significant of these are the argument from explanatory exclusion and the contention that the ER-functionalism that many nonreductive physicalists espouse cannot accommodate irreducibly mental causal powers. I have attempted to answer these challenges, but one aspect of my anti-ER-functionalist response bears highlighting. Commonsense ER-functionalist characterizations of mental states need not await the results of scientific investigation. Hence, if such functionalist characterizations did capture the nature of mental states, and some version of nonreductive physicalism could accommodate ER-functionalism, then in an important respect nonreductive physicalism would not depend on scientific investigation for its validation. However, Kim is right to claim that ER-functionalism cannot countenance genuinely mental causal powers, and for this reason, ER-functionalism is incompatible with a robust version of nonreductive physicalism. I’ve instead proposed

169 Mental Compositional Properties

that mental properties are sufficiently abstract compositional properties, which allows mental properties to be genuinely mental causal powers. Whether there exist compositional properties of the requisite sort is a matter for empirical investigation, which is currently incomplete, and so confidence that this position is true would need to be moderated. At the same time, if my responses to the objections we’ve considered are plausible, this kind of robust nonreductive physicalism about mental states should count as a serious option.

CONCLUSION

What are the prospects of physicalism? I’ve argued that given the assumption that we introspect phenomenal properties as featuring qualitative natures distinct from any that physical theories represent them as having, two physicalist strategies are salient. The first exploits the serious open possibility that introspection misrepresents phenomenal properties as having qualitative natures that they in fact lack. Then the appearance that Mary learns something new upon leaving her black-and-white room may be due to our ignorance of this misrepresentation. When Mary leaves the room and sees the red tomato, we imagine her having the belief of the form: (A) Seeing red has R. On a first proposal, on which ‘R’ is taken to refer to a property with a qualitative nature accurately represented by the introspective what-it-is-like-to-sense-red mode of presentation, because phenomenal redness has no such nature, what Mary believes would be false, and thus she would not acquire a new true belief. On the other hand, if ‘R’ is conceived instead to refer to a property with a qualitative nature that appears to Mary in the what-it-is-like-to-sense-red way, but is misrepresented by this mode of presentation and is straightforwardly physical, we can suppose that her belief is true. But since in the room she already believed the truth expressed by (A), she also would not acquire a new true belief. So on these options either the belief of form (A) is false, or it is true but Mary already had it when she was in the room. On neither option is the knowledge argument sound. Inspired by Chalmers’s account of secondary-quality perceptual content, I’ve endorsed a dual-content view for phenomenal representation that accommodates each of these alternatives. This strategy also poses a challenge to Premise (1) of Chalmers’s zombie argument, that ‘P,’ the complete physical truth (together with ‘T,’ a ‘that’s all’ provision), conjoined with the negation of ‘Q,’ an arbitrary phenomenal truth, is ideally, positively, 170

171 Conclusion

primarily conceivable. Suppose ‘Q’ is a truth about phenomenal property R. Premise (1) requires that if a property does not have a qualitative nature accurately represented by phenomenal concept ‘R,’ then the property represented is not R. For if ‘R’ (say, by way of ordinary content) represented property R specified as lacking this qualitative nature, then the fact that a proposition about R as it is represented introspectively is not a priori derivable from ‘P’ fails to show that the truth about the real nature of R, or about R as it really is, is not a priori derivable from ‘P.’ Thus it also fails to show that ‘Q,’ our selected truth about R, is not a priori derivable from ‘P.’ If it’s open that ‘Q’ is a priori derivable from ‘P,’ it will also be open that ‘P and ~ Q’ isn’t ideally, positively, primarily conceivable. Then what we initially thought we were conceiving as a zombie world might not really be one in which ‘Q’ is false after all. The second physicalist proposal allows that introspection accurately represents the qualitative nature of phenomenal properties. On this proposal, we are currently significantly ignorant of the absolutely intrinsic properties—those that don’t reduce to purely extrinsic properties—that underlie the physical properties our best science reveals. Perhaps such properties not only serve as the categorical bases of physical dispositions but also explain phenomenal properties as they are introspectively represented, without being phenomenal themselves. They would then be protophenomenal properties. We currently have incomplete understanding of what such protophenomenal properties would be like or how they would combine to result in the phenomenal properties, and this is the main drawback of the proposal. But it might be that protophenomenal properties are instantiated that are similar enough to paradigmatic physical properties to count as physical. There would then be principles on the basis of which truths about protophenomenal and other physical properties combine to yield truths about phenomenal properties so that ‘P and ~Q’ (or ‘P** and ~Q’) would no longer be ideally, positively, primarily conceivable. At least some of what dualists and idealists have found attractive about a nonphysicalist position can be affirmed by a nonreductive version of physicalism. On the nonreductive view I’ve defended, there are no identities, at the type or token level, between mental entities of any sort and their microphysical, chemical, biological, or neural counterparts, and as a result, mental causal explanations do not reduce to lower-level explanations. Multiple realizability arguments undercut all such identifications. The mental is physical rather because each mental entity is constituted by—that is, made up of, materially coincident with, and necessitated by, but not identical with—some microphysical entity. This view can finesse the causal and explanatory exclusion problem because there is no less reason to think that constitution precludes redundant overdetermination than that identity does so. By contrast with other versions of nonreductive physicalism, this position is not external-relations- (ER-)functionalist, but rather identifies mental state types with compositional properties—properties something has solely by virtue of intrinsic features of its parts and relations these parts have to one another—at a fairly abstract level. These compositional properties are sufficiently abstract to be multiply realizable at the neural, biological, chemical, and microphysical levels. Unlike ER-functionalist properties, such compositional properties clearly can be causally efficacious. The

172 Consciousness and the Prospects of Physicalism

mental would thus be twice grounded in the physical: by way of constitution in the microphysical and intervening levels, and by way of identity with compositional properties that are sufficiently abstract to preclude classification at any level more basic than the mental. Is physicalism about the mental true? I have explored two ways in which it might be: one that requires that we misrepresent phenomenal properties in a fairly radical way but allows for an ordinary sort of physicalism, and another that claims that we don’t misrepresent phenomenal properties in this way but requires an extraordinary sort of physicalism. To my mind, it would also be extraordinary if dualism or idealism of some sort were true, and thus all these possibilities would be surprising in some significant respect, given fairly standard contemporary presuppositions. David Papineau argues that mid-twentieth-century developments in chemistry and biology swayed the prevailing opinion toward affirming the causal closure of the physical, on which every physical event has in some sense a complete physical causal explanation.1 But from the current perspective, the issues most pertinent to the truth of physicalism are not chemical or biological but psychological, and phenomenal in particular. Moreover, the inductive argument from the steadily increasing success of physicalistic explanation is not decisive. Tyler Burge contends that no feature of current scientific practice in psychology tells in favor of physicalism, and that psychological explanation does not require physicalist validation.2 About this he might well be right. The metaphysical considerations with which we’ve been occupied, not provided directly by scientific results, are unlikely to settle the issue. We formulate various coherent metaphysical positions, and then develop and critically examine arguments for and against that are most often inconclusive. This philosophical practice accords with the widespread contemporary rejection of the Kantian vision of philosophical reflection that rises to the level of a true science, one that features broad rational agreement in methodology and results.3 My aim has been to develop two coherent physicalist standpoints, to consider arguments for and against, and to formulate and defend a nonreductive view about the relation between psychology and more basic sciences, given the supposition of physicalism. I think that my arguments don’t come close to showing that these physicalist proposals are true, but only that they are reasonable options in the ongoing debate.

1 David Papineau, Thinking about Consciousness, Appendix, pp. 232–56. 2 Tyler Burge, “Mind-Body Causation and Explanatory Practice,” in Mental Causation, pp. 97–120; reprinted in Tyler Burge, Foundations of Mind (Oxford: Oxford University Press, 2007), pp. 344–62; “Postscript to Mind-Body Causation,” Foundations of Mind, pp. 363–82; Bernard Kobes, “Burge’s Dualism,” in The Waning of Materialism, ed. George Bealer and Robert C. Koons (New York: Oxford University Press, 2009). 3 Immanuel Kant, Critique of Pure Reason, Bvii–xv.

BIBLIOGR APHY

Adams, Frederick. “Properties, Functionalism, and the Identity Theory,” Eidos 1 (1979), pp. 153–79. Adams, Robert. “Flavors, Colors and God,” in his The Virtue of Faith, Oxford: Oxford University Press, 1987, pp. 243–62. Adams, Robert. “Idealism Vindicated,” in Persons, Human and Divine, ed. Peter van Inwagen and Dean Zimmerman, Oxford: Oxford University Press, 2007, pp. 35–54. Almog, Joseph. What Am I? Descartes and the Mind-Body Problem, New York: Oxford University Press, 2002. Alston, William. “Mystical and Perceptual Awareness of God,” in The Blackwell Guide to Philosophy of Religion, ed. William E. Mann, Oxford: Blackwell, 2004, pp. 198–219. Alter, Torin. “Mary’s New Perspective,” Australasian Journal of Philosophy 73 (1995), pp. 582–84. Alter, Torin. “On the Conditional Analysis of Phenomenal Concepts,” Philosophical Studies 134 (2007), pp. 235–53. Alter, Torin. “Does the Ignorance Hypothesis Undermine the Conceivability and Knowledge Arguments?” Philosophy and Phenomenological Research 79 (2009), pp. 756–65. Antony, Louise, and Joseph Levine. “Reduction with Autonomy,” in Philosophical Perspectives 11, Oxford: Blackwell, 1997, pp. 83–105. Armstrong, David. Perception and the Physical World, London: Routledge, 1961. Armstrong, David. A Materialist Theory of Mind, London: Routledge, 1968. Armstrong, David. “The Nature of Mental States,” in Readings in the Philosophy of Psychology, vol. 1, ed. Ned Block, Cambridge: Harvard University Press, 1980, pp. 191–99; originally published in The Mind/Brain Identity Theory, ed. C. V. Borst, London: Macmillan, 1970, pp. 67–79. Armstrong, David. Universals and Scientific Realism, Cambridge: Cambridge University Press, 1978. Armstrong, David. “Four Disputes about Properties,” Synthèse 144 (2005), pp. 309–20. Arnauld, Antoine. Fourth Set of Objections (to René Descartes’s Meditations on First Philosophy), in The Philosophical Writings of Descartes, vol. 2, tr. and ed. John Cottingham, Robert Stoothoff, and Dugald Murdoch, Cambridge: Cambridge University Press, 1984. Ayers, Michael R. Locke, 2 vols., London: Routledge, 1991. Baker, Lynne R. Explaining Attitudes: A Practical Approach to the Mind, Cambridge: Cambridge University Press, 1995. Baker, Lynne R. “Why Constitution Is Not Identity,” Journal of Philosophy 94 (1997), pp. 599–621. Baker, Lynne R. Persons and Bodies, Cambridge: Cambridge University Press, 2000. Baker, Lynne R. “Replies to Derk Pereboom, Michael Rea, and Dean Zimmerman,” Philosophy and Phenomenological Research 64 (2002), pp. 623–35. 173

174 Bibliography

Baker, Lynne R. The Metaphysics of Everyday Life, Cambridge: Cambridge University Press, 2007. Bealer, George. “Modal Epistemology and the Rationalist Renaissance,” in Conceivability and Possibility, ed. Tamar Szabó Gendler and John Hawthorne, Oxford: Oxford University Press, 2002, pp. 77–125. Bechtel, William, and Jennifer Mundale. “Multiple Realizability Revisited: Linking Cognitive and Neural States,” Philosophy of Science 66 (1999), pp. 175–207. Bedau, Mark. “Weak Emergence,” Philosophical Perspectives 11, Oxford: Blackwell, 1997, pp. 375–99. Bennett, Karen. “Why the Exclusion Problem Seems Intractable, and How, Just Maybe, to Tract It,” Noûs 37 (2003), pp. 471–97. Bennett, Karen. “Mental Causation,” Philosophy Compass 2 (2007), pp. 316–37. Bennett, Karen. “Exclusion Again,” in Being Reduced, ed. J. Kallestrup and J. Hohwy, Oxford: Oxford University Press, 2009. Bennett, Karen. “What You Don’t Know Can Hurt You,” Philosophy and Phenomenological Research 79 (2009), pp. 766–74. Bennett, Karen. “Construction Area (No Hard Hat Required),” ms. Bennett, Karen. “Why I Am Not a Dualist,” ms. Berkeley, George. Three Dialogues between Hylas and Philonous, ed. Jonathan Dancy, Oxford: Oxford University Press, 1998. Berkeley, George. A Treatise Concerning the Principles of Human Knowledge, ed. Jonathan Dancy, Oxford: Oxford University Press, 1998. Blackburn, Simon. “Filling in Space,” Analysis 50 (1990), pp. 62–65. Block, Ned. “Introduction: What Is Functionalism?” in Readings in the Philosophy of Psychology, ed. Ned Block, Cambridge, MA: Harvard University Press, 1980, pp. 171–84. Block, Ned. “Can the Mind Change the World,” in Meaning and Method, ed. Ned Block, Cambridge: Cambridge University Press, 1990, pp. 137–70. Block, Ned. “Anti-Reductionism Slaps Back,” Philosophical Perspectives 11, Oxford: Blackwell, 1997, pp. 107–32. Block, Ned, and Robert Stalnaker. “Conceptual Analysis, Dualism, and the Explanatory Gap,” Philosophical Review 108 (1999), pp 1–46. Boghossian, Paul, and David Velleman. “Colour as a Secondary Quality,” Mind 98 (1989), pp. 81–103. Boyd, Richard. “Materialism without Reductionism: What Physicalism Does Not Entail,” in Readings in the Philosophy of Psychology, ed. Ned Block, Cambridge, MA: Harvard University Press, 1980, pp. 67–106. Boyd, Richard. “Scientific Realism and Naturalistic Epistemology,” East Lansing, MI: Proceedings of the Philosophy of Science Association, vol. 2, 1980, pp. 613–62. Boyd, Richard. “Kinds, Complexity, and Multiple Realization,” Philosophical Studies 95 (1999), pp. 67–98. Braddon-Mitchell, David. “Qualia and Analytical Conditionals,” Journal of Philosophy 100 (2003), pp. 111–35. Brentano, Franz. Psychologie vom empirischen Standpunkt, Leipzig: Duncke & Humblot, 1874, translated as Psychology from an Empirical Standpoint by A. C. Rancurello, D. B. Terrell, and L. L. McAlister, London: Routledge and Kegan Paul, 1973. Broad, C. D. Mind and Its Place in Nature, London: Kegan Paul, 1925.

175 Bibliography

Brown, Robin, and James Ladyman. “Physicalism, Supervenience, and the Fundamental Level,” Philosophical Quarterly 59 (2009), pp. 20–38. Burge, Tyler. “Individualism and the Mental,” in Midwest Studies in Philosophy 6, ed. P. French, T. Uehling, and H. Wettstein, Minneapolis: University of Minnesota Press, 1978, pp. 73–121. Burge, Tyler. “Other Bodies,” in Thought and Content, ed. A. Woodfield, Oxford: Oxford University Press, 1979, pp. 97–120. Burge, Tyler. “Mind-Body Causation and Explanatory Practice,” in Mental Causation, ed. John Heil and Alfred Mele, Oxford: Oxford University Press, 1993, pp. 97–120; reprinted in his Foundations of Mind, Oxford: Oxford University Press, 2007, pp. 344–62. Burge, Tyler. “Postscript to Mind-Body Causation,” in his Foundations of Mind, Oxford: Oxford University Press, 2007, pp. 363–82. Byrne, Alex. “Review of Purple Haze,” Philosophical Review 111 (2002), pp. 594–97. Byrne, Alex. “Introspection,” Philosophical Topics 33 (2005), pp. 79–104. Byrne, Alex, and David Hilbert. “Color Primitivism,” Erkenntnis 66 (2007), pp. 73–105. Carnap, Rudolf. Der logische Aufbau der Welt, Berlin: Weltkreis, 1928; translated as The Logical Structure of the World by Rolf George, Berkeley and Los Angeles: University of California Press, 1967. Carnap, Rudolf. “Psychology in Physical Language,” Erkenntnis 3 (1932–33), reprinted in Logical Positivism, ed. A. J. Ayer, New York: Macmillan, 1959, pp 165–98. Carnap, Rudolf. “Meaning and Synonymy in Natural Languages,” Philosophical Studies 6 (1955), pp. 33–47. Chalmers, David. The Conscious Mind, Oxford: Oxford University Press, 1996. Chalmers, David. “Materialism and the Metaphysics of Modality,” Philosophy and Phenomenological Research 59 (1999), pp. 473–96. Chalmers, David. “Consciousness and Its Place in Nature,” in Blackwell Guide to the Philosophy of Mind, Oxford: Blackwell, 2002; reprinted in Philosophy of Mind: Classical and Contemporary Readings, ed. David Chalmers, New York: Oxford University Press, 2002, pp. 247–72. Chalmers, David. “Does Conceivability Entail Possibility?” in Conceivability and Possibility, ed. Tamar Gendler and John Hawthorne, Oxford: Oxford University Press, 2002, pp. 145–200. Chalmers, David. “The Content and Epistemology of Phenomenal Belief,” in Consciousness: New Philosophical Perspectives, ed. Q. Smith and A. Jokic, Oxford: Oxford University Press, 2003. Chalmers, David. “Imagination, Indexicality and Intensions,” Philosophy and Phenomenological Research 68 (2004), pp. 182–90. Chalmers, David. “Foundations of Two-Dimensional Semantics,” in Two-Dimensional Semantics, ed. M. Garcia-Carpintero and J. Macia, Oxford: Oxford University Press, 2006, pp. 55–140. Chalmers, David. “Perception and the Fall from Eden,” in Perceptual Experience, ed. Tamar Szabó and John Hawthorne, Oxford: Oxford University Press, 2006, pp. 49–125. Chalmers, David. “Phenomenal Concepts and the Explanatory Gap,” in Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism, ed. Torin Alter and Sven Walter, Oxford: Oxford University Press, 2006, pp. 167–94.

176 Bibliography

Chalmers, David. “The Two-Dimensional Argument against Materialism,” in The Character of Consciousness, New York: Oxford University Press, forthcoming. Chalmers, David, and Frank Jackson. “Conceptual Analysis and Reductive Explanation,” Philosophical Review 110 (2001), pp. 315–61. Chatel, J. C., and R. Peele. “The Concept of Neuraesthenia,” International Journal of Psychiatry 9 (1970), pp. 36–49. Chignell, Andrew. “Belief in Kant,” Philosophical Review 116 (2007), pp. 323–61. Churchland, Patricia S. Neurophilosophy: Toward a Unified Science of the Mind-Brain, Cambridge: Bradford Books/MIT Press, 1986. Churchland, Paul M. “Eliminative Materialism and the Propositional Attitudes,” Journal of Philosophy 78 (1981), pp. 67–90. Churchland, Paul M. “Reduction, Qualia and the Direct Introspection of Brain States,” Journal of Philosophy 82 (1985), pp. 8–28. Churchland, Paul M. “Functionalism at Forty: A Critical Retrospective,” Journal of Philosophy 102 (2005), pp. 33–50. Churchland, Paul M., and Patricia S. Churchland. “Intertheoretic Reduction,” in Paul M. Churchland and Patricia S. Churchland, On the Contrary, Cambridge, MA: MIT Press, 1998. Clapp, Lenny. “Disjunctive Properties: Multiple Realizations,” Journal of Philosophy 98 (2001), pp. 111–36. Clarke, Randolph. “Nonreductive Physicalism and the Causal Powers of the Mental,” Erkenntnis 51 (1999), pp. 295–322. Collingwood, Robin G. The Idea of History, Oxford: Oxford University Press, 1946. Conee, Earl. “The Comforts of Home,” Philosophy and Phenomenological Research 70 (2005), pp. 444–51. Crane, Tim, and Hugh Mellor. “There Is No Question of Physicalism,” Mind 99 (1990), pp. 185–206. Crisp, Thomas, and Ted Warfield. “Kim’s Master Argument,” Noûs 35 (2001), pp. 304–16. Cross, Troy. The Nature of Fundamental Properties, Dissertation, Rutgers University, 2004. Cross, Troy. “What Is a Disposition?” Synthèse 144 (2005), pp. 321–41. Culler, Jonathan. On Deconstruction: Theory and Criticism after Structuralism, Ithaca: Cornell University Press, 1982, 2007. Davidson, Donald. “Mental Events,” in his Essays on Actions and Events, Oxford: Clarendon Press, 1980, pp. 207–25. Dennett, Daniel. Consciousness Explained, Boston: Little, Brown, and Company, 1991. Derrida, Jacques. “Structure, Sign, and Play in the Discourse of the Human Sciences,” in his Writing and Difference, tr. Alan Bass, Chicago: University of Chicago Press, 1978, pp. 278–93; originally L’écriture et la différence, Paris: Editions du Seuil, 1967. Derrida, Jacques. De la grammatologie, Paris: Minuit, 1967; translated as Of Grammatology by Gayatri C. Spivak, Baltimore: Johns Hopkins University Press, 1976, 1998. Descartes, René. Oeuvres de Descartes, 11 vols., ed. Charles Adam and Paul Tannery, Paris: J. Vrin, 1964–76. (Abbreviated as ‘AT’) Descartes, René. The Philosophical Writings of Descartes, 3 vols., tr. and ed. John Cottingham, Robert Stoothoff, and Dugald Murdoch, Cambridge: Cambridge University Press, 1984. Diamond, L. “Neuraesthenia,” in International Encyclopaedia of Psychiatry, Psychology, Psychoanalysis, and Neurology, ed. Benjamin Wolman, New York: Aesculapius, 1977, vol. 8, pp. 27–28.

177 Bibliography

Dilthey, Wilhelm. Der Aufbau der geschichtlichen Welt in den Geisteswissenschaften (1910), Stuttgart, 1956; translated as The Formation of the Historical World in the Human Sciences by Rudolf Makkreel, Princeton, NJ: Princeton University Press, 2002. Dooyeweerd, Herman. De Wijsbegeerte der Wetsidee (The Philosophy of the Law-Idea), Amsterdam: H. J. Paris, 1935–36; translated as A New Critique of Theoretical Thought by David Freeman, Philadelphia: Presbyterian and Reformed Publishers, 1953–58. D’Oro, Guiseppina. “Collingwood, Metaphysics, and Historicism,” Dialogue 41 (2002), pp. 1–20. Dowell, Janice L. “The Physical: Empirical, Not Metaphysical,” Philosophical Studies 131 (2006), pp. 25–60. Dowell, Janice L. “Empirical Metaphysics: The Role of Intuitions about Possible Cases in Philosophy,” Philosophical Studies 140 (2008), pp. 19–46. Dowell, Janice L. “A Priori Entailment and Conceptual Analysis: Making Room Type-C Physicalism,” Australasian Journal of Philosophy 86 (2008), pp. 93–111. Dunn, J. Michael. “Relevant Predication 2: Intrinsic Properties and Internal Relations,” Philosophical Studies 60 (1990), pp. 177–206. Dupré, John. “The Solution to the Problem of the Freedom of the Will,” Philosophical Perspectives 10, Oxford: Blackwell, 1996, pp. 385–402. Elder, Crawford. “Mental Causation versus Physical Causation: No Contest,” Philosophy and Phenomenological Research 62 (2001), pp. 111–27. Endicott, Ronald. “On Physical Multiple Realization,” Pacific Philosophical Quarterly 70 (1989), pp. 212–24. Fara, Michael. “Dispositions,” The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, at http://plato.stanford.edu/entries/dispositions/. Feigl, Herbert. The ‘Mental’ and the ‘Physical,’ Minneapolis: University of Minnesota Press, 1967. Fine, Kit. “The Question of Realism,” Philosophical Imprint (1) 2001, pp. 1–30. Fine, Kit. “The Non-identity of a Material Thing and Its Matter,” Mind 112 (2003), pp. 195–234. Fodor, Jerry. “Special Sciences,” Synthèse 28 (1974), pp. 97–115. Fodor, Jerry. “Special Sciences: Still Autonomous after All These Years,” Philosophical Perspectives 11, Oxford: Blackwell, 1997, pp. 149–63. French, Steven, and James Ladyman. “Remodelling Structural Realism: Quantum Physics and the Metaphysics of Structure,” Synthèse 136 (2003), pp. 31–56. Friedman, Michael. “Carnap’s Aufbau Reconsidered,” Noûs 21 (1987), pp. 521–45. Funkhouser, Eric. “The Determinable-Determinate Relation,” Noûs 40 (2006), pp. 548–69. Galileo Galilei. The Assayer, in Discoveries and Opinions of Galileo, tr. Stillman Drake, New York: Doubleday Anchor, 1957, pp. 217–80. Gertler, Brie. “Explanatory Reduction, Conceptual Analysis, and Conceivability Arguments about the Mind,” Noûs 36 (2002), pp. 22–49. Gertler, Brie. “Daniel Stoljar’s Ignorance and Imagination,” Noûs 43 (2009), pp. 378–93. Gillett, Carl. “The Dimensions of Multiple Realization: A Critique of the Standard View,” Analysis 62 (2002), pp. 316–23. Gillett, Carl. “The Metaphysics of Realization, Multiple Realizability, and the Special Sciences,” Journal of Philosophy 100 (2003), pp. 591–603. Gillett, Carl. “‘A Whole Lot More from Nothing But’: Scientific Composition and the Possibility of Strong Emergence,” delivered at the American Philosophical Association Meetings, February 2009.

178 Bibliography

Harré, Rom, and E. H. Madden. Causal Powers: A Theory of Natural Necessity, Oxford: Blackwell, 1975. Hawthorne, John. “Causal Structuralism,” Philosophical Perspectives 15 (2001), pp. 361–78. Hawthorne, John. “Advice for Physicalists,” Philosophical Studies 109 (2002), pp. 53–74. Hawthorne, John. “Knowledge and Evidence,” Philosophy and Phenomenological Research 70 (2005), pp. 452–58. Heil, John. “Multiple Realizability,” American Philosophical Quarterly 36 (1999), pp. 189–208. Heil, John. “Dispositions,” Synthèse 144 (2005), pp. 343–56. Hill, Christopher. Sensations: A Defense of Type Materialism, Cambridge: Cambridge University Press, 1991. Hill, Christopher. “Imaginability, Conceivability, Possibility, and the Mind-Body Problem,” Philosophical Studies 87 (1997), pp. 61–85. Hill, Christopher, and Brian McLaughlin. “There Are Fewer Things in Reality Than Are Dreamt of in Chalmers’s Philosophy,” Philosophy and Phenomenological Research 59 (1999), pp. 445–54. Holden, Thomas. The Architecture of Matter, Oxford: Oxford University Press, 2004. Horgan, Terence. “Jackson on Physical Information and Qualia,” Philosophical Quarterly 32 (1984), pp. 147–52. Horgan, Terence. “From Supervenience to Superdupervenience: Meeting the Demands of a Material World,” Mind 102 (1993), pp. 555–85. Horgan, Terence. “Kim on Mental Causation and Causal Exclusion,” Philosophical Perspectives 11, Oxford: Blackwell, 1997, pp. 165–84. Howell, Robert. “The Knowledge Argument and Objectivity” Philosophical Studies 135 (2007), pp. 135–77. Howell, Robert. “The Ontology of Subjective Physicalism,” Noûs 43 (2009), pp. 315–45. Humberstone, Lloyd. “Intrinsic/Extrinsic,” Synthèse 108 (1996), pp. 205–67. Hume, David. An Enquiry Concerning the Principles of Morals, Oxford: Oxford University Press, 1975. Hume, David. A Treatise of Human Nature, Oxford: Oxford University Press, 1978. Hume, David. An Enquiry Concerning Human Understanding, Oxford: Oxford University Press, 1999. Jackson, Frank. “Epiphenomenal Qualia,” Philosophical Quarterly 32 (1980), pp. 127–36. Jackson, Frank. “What Mary Didn’t Know,” Journal of Philosophy 83 (1986), pp. 291–95. Jackson, Frank. From Metaphysics to Ethics, Oxford: Oxford University Press, 1998. Jackson, Frank, and Philip Pettit. “Program Explanation: A General Perspective,” Analysis 50 (1990), pp. 107–17. Jacovides, Michael. “Locke’s Resemblance Theses,” Philosophical Review 108 (1999), pp. 461–96. Jaworski, William. “Multiple Realizability, Explanation, and the Disjunctive Move,” Philosophical Studies 108 (2002), pp. 298–308. Johnston, Mark. “Constitution Is Not Identity,” Mind 101 (1992), pp. 89–105. Johnston, Mark. “How to Speak of the Colors,” Philosophical Studies 68 (1992), pp. 221–63. Kant, Immanuel. Kants gesammelte Schriften, 29 vols., ed. Koniglichen Preussischen Akademie der Wissenschaften, Berlin: Walter de Gruyter et al., 1902–1980. (Abbreviated as ‘Ak’) Kant, Immanuel. Critique of Pure Reason, tr. Norman Kemp Smith, London: Macmillan, 1929. Kant, Immanuel. Critique of Pure Reason, tr. Paul Guyer and Allen Wood, Cambridge: Cambridge University Press, 1987.

179 Bibliography

Kant, Immanuel. Critique of the Power of Judgment, tr. Paul Guyer and Eric Matthews, Cambridge: Cambridge University Press, 2000. Kant, Immanuel. Metaphysical Foundations of Natural Science, tr. Michael Friedman, Cambridge: Cambridge University Press, 2004. Kim, Jaegwon. “Phenomenal Properties, Psychophysical Laws, and the Identity Theory,” Monist 56 (1972), pp. 190–91. Kim, Jaegwon. “Psychophysical Supervenience,” Philosophical Studies 41 (1982), pp. 51–70. Kim, Jaegwon. “The Myth of Nonreductive Materialism,” in his Supervenience and Mind, Cambridge: Cambridge University Press, 1993, pp. 265–84; first published in Proceedings and Addresses of the American Philosophical Association 63 (1989), pp. 31–47. Kim, Jaegwon. “Multiple Realizability and the Metaphysics of Reduction,” in his Supervenience and Mind, Cambridge: Cambridge University Press, 1993, pp. 309–35; first published in Philosophy and Phenomenological Research 52 (1992), pp. 1–26. Kim, Jaegwon. Mind in a Physical World, Cambridge, MA: MIT Press, 1998. Kim, Jaegwon. “Making Sense of Emergence,” Philosophical Studies 95 (1999), pp. 3–36. Kitcher, Philip S. “1953 and All That: A Tale of Two Sciences,” Philosophical Review 93 (1984), pp. 335–73. Kobes, Bernard W. “Burge’s Dualism,” in The Waning of Materialism, ed. George Bealer and Robert Koons, Oxford: Oxford University Press, 2009. Kornblith, Hilary. Knowledge without Foundations: A Causal Theory, Dissertation, Cornell University, 1979. Kriegel, Uriah. Subjective Consciousness: A Self-Representational Theory, Oxford: Oxford University Press, 2009. Kripke, Saul. Naming and Necessity, Cambridge, MA: Harvard University Press, 1980. Ladyman, James. “Structural Realism,” The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, at http://plato.stanford.edu/entries/structural-realism. Ladyman, James, and Don Ross, with David Spurrett and John Collier. Every Thing Must Go, Oxford: Oxford University Press, 2007. Langton, Rae. Kantian Humility: Our Ignorance of Things in Themselves, Oxford, Oxford University Press, 1998. Langton, Rae. “Elusive Knowledge of Things in Themselves,” Australasian Journal of Philosophy 82 (2004), pp. 129–36. Langton, Rae, and David Lewis. “Defining ‘Intrinsic,’” Philosophy and Phenomenological Research 58 (1998), pp. 333–45. Leibniz, Gottfried Wilhelm. G. W. Leibniz, Mathemathische Schriften, 7 vols., ed. C. I. Gerhardt, Berlin and Halle, 1849–56. Leibniz, Gottfried Wilhelm. G. W. Leibniz, Die philosophischen Schriften, 7 vols., ed. C. I. Gerhard, Hildesheim, Germany: Olms, 1965. Leibniz, Gottfried Wilhelm. Philosophical Papers and Letters, tr. and ed. L. E. Loemker, Dordrecht, The Netherlands: D. Reidel, 1969. Levine, Joseph. “Materialism and Qualia: The Explanatory Gap,” Pacific Philosophical Quarterly 64 (1983), pp. 354–61. Levine, Joseph. Purple Haze, New York: Oxford University Press, 2001. Lévi-Strauss, Claude. Les structures élementaires de la parenté, Paris: Presses Universitaires, 1949; translated as The Elementary Structures of Kinship by James Harle Bell, John Richard Von Sturmer, and Rodney Needham, Boston: Beacon, 1969.

180 Bibliography

Lewis, David. “Attitudes De Dicto and De Se,” Philosophical Review, 88 (1979), pp. 513–43. Lewis, David. “Reduction of Mind,” in A Companion to Philosophy of Mind, ed. Samuel Guttenplan, Oxford: Blackwell, 1994; reprinted in his Papers in Metaphysics and Epistemology, Cambridge: Cambridge University Press, 1999, pp. 291–324. Lewis, David. “Should a Materialist Believe in Qualia?” Australasian Journal of Philosophy 73 (1995), pp. 140–44; reprinted in his Papers in Metaphysics and Epistemology, Cambridge: Cambridge University Press, 1999, pp. 325–31. Lewis, David. “Elusive Knowledge,” Australasian Journal of Philosophy 74 (1996), pp. 549–67. Lewis, David. “Ramseyan Humility,” in Conceptual Analysis and Philosophical Naturalism, ed. David Braddon-Mitchell and Robert Nola, Cambridge, MA: MIT Press, 2009, pp. 203–22. Loar, Brian. “Phenomenal States,” in Philosophical Perspectives 4: Action Theory and Philosophy of Mind, ed. James Tomberlin, Atascadero, CA: Ridgeview, 1990, pp. 81–108. Loar, Brian. “Phenomenal States,” in The Nature of Consciousness: Philosophical Debates, ed. Ned Block, Owen Flanagan, and Guven Güzeldere, Cambridge, MA: MIT Press, 1997. Loar, Brian. “David Chalmers’s The Conscious Mind,” Philosophy and Phenomenological Research 59 (1999), pp. 465–72. Locke, Dustin. “A Partial Defense of Ramseyan Humility,” Conceptual Analysis and Philosophical Naturalism, ed. David Braddon-Mitchell and Robert Nola, Cambridge, MA: MIT Press, 2009, pp. 223–41. Locke, John. An Essay Concerning Human Understanding, ed. P. H. Nidditch, Oxford: Oxford University Press, 1975. Loewer, Barry. “Review of Mind in a Physical World by Jaegwon Kim,” Journal of Philosophy 98 (2001), pp. 315–24. Lycan, William G. “What Is the ‘Subjectivity’ of the Mental?” Philosophical Perspectives 4 (1990), pp. 109–30. Makkreel, Rudolf. Dilthey: Philosopher of the Human Studies, Princeton, NJ: Princeton University Press, 1975. Martin, Charlie B. “On the Need for Properties: The Road to Pythagoreanism and Back,” Synthèse 112 (1997), pp. 193–221. Martin, Charlie B., and John Heil. “The Ontological Turn,” Midwest Studies in Philosophy 23 (1999), pp. 34–60. Matthews, Gareth. “Aristotle’s Theory of Kooky Objects,” ms. Maund, Barry. “The Illusion Theory of Colour: An Anti-Realist Theory,” Dialectica 60 (2006), pp. 245–68. Maxwell, Grover. “The Ontological Status of Theoretical Entities,” Minnesota Studies in the Philosophy of Science 3 (1962), pp. 3–14. Maxwell, Grover. “Structural Realism and the Meaning of Theoretical Terms,” Minnesota Studies in the Philosophy of Science 4 (1970), pp. 181–92. McGinn, Colin. “How Not to Solve the Mind-Body Problem,” in Physicalism and Its Discontents, ed. Carl Gillett and Barry Loewer, Cambridge: Cambridge University Press, 2001. McGinn, Colin. “What Constitutes the Mind-Body Problem,” in his Consciousness and Its Objects, Oxford: Oxford University Press, 2004. McKitrick, Jennifer. “The Bare Metaphysical Possibility of Bare Dispositions,” Philosophy and Phenomenological Research 66 (2003), pp. 349–69.

181 Bibliography

McLaughlin, Brian. “The Rise and Fall of British Emergentism,” in Emergence or Reduction? Essays on the Prospects of Nonreductive Physicalism, ed. A. Beckermann, H. Flohr, and J. Kim, New York: Walter de Gruyter, 1992, pp. 49–93. Melnyk, Andrew. A Physicalist Manifesto, Cambridge: Cambridge University Press, 2003. Melnyk, Andrew. “Can Physicalism Be Non-Reductive?” Philosophy Compass 3, no. 6 (2008), pp. 1281–96. Millikan, Ruth G. “Historical Kinds and the ‘Special Sciences,’” Philosophical Studies 95 (1999), pp. 45–65. Moyer, Mark. “A Semantic Approach to Material Constitution,” Dissertation, Rutgers University. Moyer, Mark. “Statues and Lumps: A Strange Coincidence?” Synthèse 148 (2006), pp. 401–23. Mucciolo, Laurence. “The Identity Thesis and Neurophysiology,” Noûs 8 (1974), pp. 327–42. Nagel, Thomas. “What Is It Like to Be a Bat?” Philosophical Review 83 (1974), pp. 435–50. Nagel, Thomas. “Panpsychism,” in his Mortal Questions, Oxford: Oxford University Press, 1979. Nagel, Thomas. The View from Nowhere, New York: Oxford University Press, 1986. Newman, M. H. A. “Mr. Russell’s Causal Theory of Perception,” Mind 37 (1928), pp. 137–48. Ney, Alyssa. “Can an Appeal to Constitution Solve the Exclusion Problem?” Pacific Philosophical Quarterly 88 (2007), pp. 486–506. Ney, Alyssa. “Physicalism and Our Knowledge of Intrinsic Properties,” Australasian Journal of Philosophy 85 (2007), pp. 41–60. Nida-Rümelin, Martine. “Qualia: The Knowledge Argument,” The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, at http://plato.stanford.edu/entries/qualia-knowledge. Noonan, Harold. “Constitution Is Identity,” Mind 102 (1993), pp. 133–46. Noonan, Harold. “Moderate Monism, Sortal Concepts, and Relative Identity,” Monist, forthcoming. O’Connor, Timothy. “Emergent Properties,” American Philosophical Quarterly 31 (1994), pp. 91–104. O’Connor, Timothy. “Agent-Causal Power,” in Dispositions and Causes, ed. Toby Handfield, Oxford: Oxford University Press, 2008, pp. 189–214. Papineau, David. Thinking about Consciousness, Oxford: Oxford University Press, 2002. Parkinson, G. H. R. Leibniz, Logical Papers, Oxford: Clarendon, 1966. Paul. L. A. “Aspect Causation,” Journal of Philosophy 97 (2000), pp. 235–56. Paul, L. A. “Coincidence as Overlap,” Noûs 40 (2006), pp. 623–59. Paul. L. A. “Constitutive Overdetermination,” in Topics in Contemporary Philosophy, Vol. 4: Causation and Explanation, ed. Joseph K. Campbell and Michael O’Rourke, Cambridge, MA: MIT Press, 2007, pp. 265–90. Pereboom, Derk. “Is Kant’s Transcendental Philosophy Inconsistent?” History of Philosophy Quarterly 8 (1991), pp. 357–71. Pereboom, Derk. “Kant’s Amphiboly,” Archiv für Geschichte der Philosophie 73 (1991), pp. 50–70. Pereboom, Derk. “Why a Scientific Realist Cannot Be a Functionalist,” Synthese 88 (1991), pp. 341–58. Pereboom, Derk. “Bats, Brain Scientists, and the Limitations of Introspection,” Philosophy and Phenomenological Research 54 (1994), pp. 315–29. Pereboom, Derk. “Conceptual Structure and the Individuation of Content,” Philosophical Perspectives 9 (1995), pp. 401–26. Pereboom, Derk. Living without Free Will, Cambridge: Cambridge University Press, 2001.

182 Bibliography

Pereboom, Derk. “On Baker’s Persons and Bodies,” Philosophy and Phenomenological Research 64 (2002), pp. 616–23. Pereboom, Derk. “Robust Nonreductive Materialism,” Journal of Philosophy 99 (2002), pp. 499–531. Pereboom, Derk. “The Problem of Evil,” in The Blackwell Guide to Philosophy of Religion, ed. William E. Mann, Oxford: Blackwell, 2004, pp. 148–70. Pereboom, Derk. “Consciousness and Introspective Inaccuracy, in Metaphysics and the Good: Themes from the Philosophy of Robert Merrihew Adams, ed. L. M. Jorgensen and Samuel Newlands, Oxford: Oxford University Press, 2009, pp. 156–87. Pereboom, Derk, and Hilary Kornblith. “The Metaphysics of Irreducibility,” Philosophical Studies 63 (1991), pp. 125–45. Perry, John. Knowledge, Possibility, and Consciousness, Cambridge, MA: MIT Press, 2001. Place, U. T. “Is Consciousness a Brain Process?” British Journal of Psychology 47 (1956), pp. 44–50. Polger, Thomas. Natural Minds, Cambridge, MA: MIT Press, 2004. Prior, Elizabeth. Dispositions, Aberdeen, Scotland: Aberdeen University Press, 1985. Prior, Elizabeth, Robert Pargetter, and Frank Jackson. “Three Theses about Dispositions,” American Philosophical Quarterly 19 (1982), pp. 251–57. Putnam, Hilary. “Brains and Behavior,” in his Philosophical Papers, vol. 2, Cambridge: Cambridge University Press, 1975, pp. 325–41; first published in Analytical Philosophy, second series, Oxford: Blackwell, 1965, pp. 1–19. Putnam, Hilary. “The Nature of Mental States,” in his Philosophical Papers, vol. 2, Cambridge: Cambridge University Press, 1975, pp. 429–40; first published as “Psychological Predicates,” in Art, Mind, and Religion, ed. W. H. Capitan and D. D. Merill, Pittsburgh: Pittsburgh University Press, 1967, pp. 37–48. Putnam, Hilary. “Language and Reality,” in his Philosophical Papers, vol. 2, Cambridge: Cambridge University Press, 1975, pp. 272–90; this paper was delivered as a Machette Lecture at Princeton University, May 22, 1974. Putnam, Hilary. “The Meaning of ‘Meaning,’” in his Philosophical Papers, vol. 2, Cambridge: Cambridge University Press, 1975, pp. 215–71; first published in Minnesota Studies in the Philosophy of Science 7 (1975), pp. 131–93. Ramsey, Frank P. “Theories,” in his The Foundations of Mathematics and Other Logical Essays, ed. R. B. Braithwaite (Paterson, NJ: Littlefield and Adams, 2001), pp. 212–36. Richardson, Alan W. Carnap’s Construction of the World, Cambridge: Cambridge University Press, 1998. Rickert, Heinrich. Die Grenzen der naturwissenschaftlichen Begriffsbilding, Tübingen, Germany: Mohr, 1929; translated as The Limits of Concept Formation in the Natural Sciences by G. Oakes, Cambridge: Cambridge University Press, 1986. Robb, David. “Power Essentialism,” ms. Rosenberg, Gregg. A Place for Consciousness, New York: Oxford University Press, 2004. Rozemond, Marleen. Descartes’s Dualism, Cambridge, MA: Harvard University Press, 1998. Rueger, Alexander. “Physical Emergence, Diachronic and Synchronic,” Synthèse 124 (2000), pp. 297–322. Russell, Bertrand. “Knowledge by Acquaintance and Knowledge by Description,” Proceedings of the Aristotelian Society 11 (1910), pp. 108–28. Russell, Bertrand. The Analysis of Matter, London: Kegan Paul, 1927.

183 Bibliography

Saucedo, Raul. “Parthood and Location,” Oxford Studies in Metaphysics 6 (2010), forthcoming. Schaffer, Jonathan. “Is There a Fundamental Level?” Noûs 37 (2003), pp. 498–517. Schaffer, Jonathan. “Quiddistic Knowledge,” in Lewisian Themes, ed. Frank Jackson and Graham Priest, Oxford: Oxford University Press, 2004, pp. 210–30. Schaffer, Jonathan. “On What Grounds What,” in Metametaphysics: New Essays on the Foundations of Ontology, ed. David J. Chalmers, David Manley, and Ryan Wasserman, Oxford: Oxford University Press, 2009, pp. 345–83. Schaffer, Jonathan. “Monism: The Priority of the Whole,” Philosophical Review 119 (2010), pp. 31–76. Schroeter, Laura. “The Rationalist Foundations of Chalmers’s 2-D Semantics,” Philosophical Studies 118 (2004), pp. 227–55. Shagrir, Oron. “Multiple Realization, Computation, and the Taxonomy of Psychological States,” Synthèse 114 (1998), pp. 445–61. Shapiro, Lawrence. “Multiple Realizations,” Journal of Philosophy 97 (2000), pp. 635–54. Shoemaker, Sydney. “Causality and Properties,” in Time and Change, ed. P. van Inwagen, Dordrecht, The Netherlands: D. Reidel, 1980, pp. 109–35; reprinted in Sydney Shoemaker, Identity, Cause, and Mind, first edition, Cambridge: Cambridge University Press, 1984, pp. 206–33; and in the second, expanded, edition, Oxford: Oxford University Press, 2003, pp. 206–33. Shoemaker, Sydney. “On an Argument for Dualism,” in Knowledge and Mind: Philosophical Essays, ed. Carl Ginet and Sydney Shoemaker, New York: Oxford University Press, 1983; reprinted in his Identity, Cause, and Mind, first edition, Cambridge: Cambridge University Press, 1984, pp. 287–308; and in the second, expanded, edition, Oxford: Oxford University Press, 2003, pp. 287–308. Shoemaker, Sydney. “Qualities and Qualia: What’s in the Mind?” Philosophy and Phenomenological Research 50 (1990), pp. 109–131; reprinted in his The First-Person Perspective and Other Essays, Cambridge: Cambridge University Press, 1996, pp. 97–120. Shoemaker, Sydney. “Self-Knowledge and ‘Inner Sense,’” Philosophy and Phenomenological Research 54 (1994), pp. 249–314; reprinted in his The First-Person Perspective and Other Essays, Cambridge: Cambridge University Press, 1996, pp. 201–68. Shoemaker, Sydney. “Causal and Metaphysical Necessity,” Pacific Philosophical Quarterly 79 (1998), pp. 59–77. Shoemaker, Sydney. “Realization, Micro-Realization, and Coincidence,” Philosophy and Phenomenological Research 67 (2003), pp.1–23. Shoemaker, Sydney. Physical Realization, Oxford: Oxford University Press, 2007. Shoemaker, Sydney. “Physical Realization and Mental Causation,” in The New Ontology of the Mental Causation Debate, ed. S. C. Gibb and Jonathan Lowe, Oxford: Oxford University Press, forthcoming. Sider, Theodore. “Review of Lynne Baker, Persons and Bodies,” Journal of Philosophy 106 (2002), pp. 45–48. Sider, Theodore. “What’s So Bad about Overdetermination?” Philosophy and Phenomenological Research 67 (2003), pp. 719–26. Smart, J. J. C. “Sensations and Brain Processes,” Philosophical Review 68 (1959), pp. 141–56. Smith, Michael, and Daniel Stoljar, “Global Response-Dependence and Noumenal Realism,” Monist 81 (1998), pp. 85–111.

184 Bibliography

Smullyan, Raymond M. “An Epistemological Nightmare,” in The Mind’s I: Fantasies and Reflections on Self and Soul, ed. D. R. Hofstadter and Daniel Dennett, New York: Basic Books, 1981. Sober, Elliott. “The Multiple Realizability Argument against Reductionism,” Philosophy of Science 66 (1990), pp. 542–64. Stalnaker, Robert. “On Considering a Possible World as Actual,” Proceedings of the Aristotelian Society, 2001; reprinted in his Ways a World Might Be: Metaphysical and Anti-Metaphysical Essays, New York: Oxford University Press, 2003, pp. 188–200. Stalnaker, Robert. “What Is It Like to Be a Zombie?” in Conceivability and Possibility, ed. Tamar Gendler and John Hawthorne, Oxford: Oxford University Press, 2002, pp. 385–400. Stalnaker, Robert. Our Knowledge of the Internal World, Oxford: Oxford University Press, 2008. Stoljar, Daniel. “Two Conceptions of the Physical,” Philosophy and Phenomenological Research 62 (2001), pp. 253-82. Stoljar, Daniel. “Physicalism,” The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, at http://plato.stanford.edu/entries/physicalism/. Stoljar, Daniel. “Physicalism and Phenomenal Concepts,” Mind and Language 20 (2005), pp. 469–94. Stoljar, Daniel. Ignorance and Imagination: The Epistemic Origin of the Problem of Consciousness, New York: Oxford University Press, 2006. Stoljar, Daniel. “Précis of Ignorance and Imagination,” Philosophy and Phenomenological Research 79 (2009), pp. 748–55. Strawson, Galen. “Real Materialism,” in his Real Materialism and Other Essays, Oxford: Oxford University Press, 2008, pp. 19–51; an earlier version appeared in Chomsky and His Critics, ed. L. Antony and N. Hornstein, Oxford: Blackwell, 2003. Strawson, Galen. “Realistic Monism,” in his Real Materialism and Other Essays, Oxford: Oxford University Press, 2008, pp. 54–74; first published in Consciousness and Its Place in Nature, ed. A. Freeman, Thorverston, England: Imprint Academic, 2006, pp. 3–31. Sturgeon, Scott. “Physicalism and Overdetermination,” Mind 107 (1998), pp. 411–32. Thompson, Brad. “Senses for Senses,” Australasian Journal of Philosophy 87 (2009), pp. 99–117. Thomson, Judith. “Parthood and Identity across Time,” Journal of Philosophy 80 (1983), pp. 201–20. Tye, Michael. “The Subjective Qualities of Experience,” Mind 95 (1986), pp. 1–17. van Cleve, James. “Inner States and Outer Relations: Kant and the Case for Monadism,” in Doing Philosophy Historically, ed. Peter H. Hare, Buffalo, NY: Prometheus, 1988, pp. 231–47. van Cleve, James. “Receptivity and Our Knowledge of Intrinsic Properties,” Philosophy and Phenomenological Research 65 (2002), pp. 218–37. Van Gulick, Robert. “Physicalism and the Subjectivity of the Mental,” Philosophical Topics 13 (1985), pp. 51–70. Van Gulick, Robert. “Who’s in Charge Here? And Who’s Doing All the Work?” in Mental Causation, ed. John Heil and Alfred Mele, Oxford: Oxford University Press, 1993, pp. 233–56. van Inwagen, Peter. Material Beings, Ithaca, NY: Cornell University Press, 1990. Wasserman, Ryan. “Material Constitution,” The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, at http://plato.stanford.edu/entries/material-constitution/.

185 Bibliography

Brian Weatherson, “Intrinsic and Extrinsic Properties,” The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta, at http://plato.stanford.edu/entries/intrinsic-extrinsic/. Williamson, Timothy. Knowledge and Its Limits, Oxford: Oxford University Press, 2000. Williamson, Timothy. “Replies to Commentators,” Philosophy and Phenomenological Research 70 (2005), pp. 468–91. Wilson, Jessica. “How Superduper Does a Physicalist Supervenience Need to Be?” Philosophical Quarterly 49 (1999), pp. 33–52. Wilson, Jessica. “Supervenience-Based Formulations of Physicalism,” Noûs 39 (2005), pp. 426–59. Wilson, Margaret. Descartes, London: Routledge, 1978. Windelband, Wilhelm. “Geschichte und Naturwissenschaft,” in his Präludien, vol. 2, Tübingen, Germany: Mohr, 1924, pp. 136–60; translated as “History and Natural Science” by G. Oakes, History and Theory 19 (1980), pp. 169–85. Witmer, Gene. “Functionalism and Causal Exclusion,” Pacific Philosophical Quarterly 84 (2003), pp. 198–214. Wykstra, Stephen. “The Humean Obstacle to Evidential Arguments from Suffering: On Avoiding the Evils of ‘Appearance,’” International Journal for Philosophy of Religion 16 (1984), pp. 73–93; reprinted in The Problem of Evil, ed. M. M. Adams and R. M. Adams, Oxford: Oxford University Press, 1990, pp. 138–60. Yablo, Stephen. “The Real Distinction between Mind and Body,” Canadian Journal of Philosophy 16 (1991), pp. 149–201. Yablo, Stephen. “Mental Causation,” Philosophical Review 101 (1992), pp. 245–80. Yablo, Stephen. “Wide Causation,” Philosophical Perspectives 11 (1997), pp. 251–81. Yablo, Stephen. “Shoulda, Woulda, Coulda,” in Conceivability and Possibility, ed. Tamar Gendler and John Hawthorne, Oxford: Oxford University Press, 2002, pp. 441–92. Yablo, Stephen. “No Fool’s Cold,” in Two-Dimensional Semantics, ed. M. Garcia-Carpintero and J. Macia, Oxford: Oxford University Press, 2006, pp. 327–45. Zimmerman, Dean W. “The Constitution of Persons by Bodies: A Critique of Lynne Rudder Baker’s Theory of Material Constitution,” Philosophical Topics 30 (2002), pp. 295–338. Zimmerman, Dean W. “Properties, Minds, and Bodies: An Examination of Sydney Shoemaker’s Metaphysics,” Philosophy and Phenomenological Research 79 (2009), pp. 673–738.

This page intentionally left blank

INDEX OF TOPICS

Bold type indicates definitions and core characterizations.

a posteriori conceivability, 50–52, 50–75 a priori conceivability, 24n.35, 50–52, 50–65, 67–75, 86–89 a priori derivability, 11, 48, 48–49n.4, 94, 102 absolutely intrinsic properties, 86, 93–94, 93–122 ignorance of, 91–110 substantival, 96, 96–102, 118 van Cleve’s characterization, 94–95n.23 accuracy, qualitative, for phenomenal representation, 13, 58n.36, 69, 85, 122 acquaintance, with phenomenal properties, 13, 28, 30–31, 36n.16, 42, 98–101, 106–7, 116 actual, conceiving as, 50–52, 50–69 A-intension (see also primary intension), 50n.10 antirealism, 117–21, 152, 156 Aristotelians, 85, 113 behaviorism, 156 building relation, 138n.43 C-intension (see also secondary intension), 50n.10 Cartesian theatre, 21 categorical properties, 87, 85–122 causal closure, 172 causal exclusion, 126–44, 146 statement of Kim’s exclusion argument, 129 causal inheritance principle, 130, 141 causal roles, 36, 48, 52, 87n.5, 88, 104–6, 108, 149 causal structuralism, 105–6, 108 causal powers, 126–28, 126–69 as concrete property instances, 127–8, 127n.13

comparatively or relatively intrinsic properties, 93–94, 94–96, 100–101, 103, 107, 112–4 van Cleve’s characterization, 94–95n.23 compositional properties, 126–27, 148–49, 148–69, 171–72 conceivability, 47–122 ideal 50, 50–86 negative, 50 positive, 24n.35, 50, 50–63, 86 and possibility, 50–57, 63–65 prima facie, 50 primary, 24n.35, 50–52, 50–65, 67–75, 86–89 secondary, 50–52 conceivability argument, 3–7, 24n.35, 47–122 Chalmers’s 47–75, 52–57, 85–122 Descartes’s, 17, 47–48, 69, 82 ignorance response, 69–84 indexical response, 12–13 Kripke’s, 17, 47–48, 51, 55, 75–84 qualitative inaccuracy response, 47–84 Russellian monist response, 53–57, 63, 89, 109–10, 85–122, 127, 168 conceiving as actual, 50–52, 50–69 conceiving as counterfactual, 50–52 conceptual analysis, 29–40 conceptual possibility, 39, 51 constitution, 5–7, 126, 131–32, 135–41 Baker’s theory, 131–2, 135–41 material, 131–32, 135–41 solution to causal exclusion problem, 141–44 content, phenomenal, 14, 17–20, 26, 36–40 externalism, 140, 156n.24 internalism, 14, 156n.24 perfect and ordinary, 26, 36–40

187

188 Index of Topics

counterfactual, conceiving as 50–52

fundamental properties, 6–8, 94–95n.23, 96

decomposition, complete, 138n determinable/determinate relation, 127, 154–55 determinative universals (or concepts), 117 dimensional realization, 135–6 dispositional properties, 86–122, 87 disjunctions, open-ended or wild, 32–34, 125–26, 158–61 dual-content view, 14, 26, 29, 35, 36–40, 171 dualism 7, 44, 48n.4, 67–68, 75, 82, 83, 122, 124, 171–72 dynamical properties, 90–91, 110–14, 118

God, 32–4, 67–8, 108, 156 and physicalism, 8 Goldbach’s conjecture, 69

Edenic and ordinary content, 14, 26, 29, 35, 36–40, 171 eliminativism, 29, 43–46 emergence, 7, 126, 136–37, 145–47 emergentism, 7, 126, 136–37, 145–47 epistemic rigidity, 56 epistemic view, 69–75 exclusion, causal and explanatory, 124–44, 146 explanation, psychological, 123–30, 158–62, 167–68 explanatory gap, 29, 43–46 extensionality, principle 138–39n.44 external-relations- (ER-) functionalism, 126, 149, 148–58 externalism about content, 140, 156n.24 extrinsic properties, 85–122 absolutely purely extrinsic properties, 94 purely extrinsic properties, 92–93, 92–122 van Cleve’s characterization, 94–95n.23, 102 facsimile principle (Yablo), 75–82, 77 flat realization, 135–6 force, 96–100 derivative, 97 Leibniz on force, 97 Kant on force, 101 primitive, 97 as relational, 101 functionalism, 119, 126, 148–58 external relations (ER), 126, 149, 148–58 Shoemaker’s, 149–50

heat, Kripke on, 55, 75–84 Humility, 100–102, 103, 103–10 ideal conceivability, 24n.35, 50, 50–84 idealism, 97, 101, 124 identity, 5, 127, 130–35, 148, 160–67 token, 5, 130–35, 166–67 type, 5, 127, 130–35, 148, 166–67 ignorance, of categorical properties, 88–91 of absolutely intrinsic properties, 91–110 of fundamental properties, 7–8, 88–110 as a response to the knowledge and conceivability argument, 69–84 Stoljar on, 69–75 inaccuracy, qualitative, 3–4, 9, 14, 14–46 indexical response to the knowledge and conceivability arguments, 12–13 introspection, models of, 19n.24 broad perceptual model, 19n.24 introspective inaccuracy, 3–4, 9, 14–46, 58–63, 85, 108–9, 122, 127, 168, 170–71 intrinsic properties, 85–122 absolutely intrinsic properties, 86, 93–94, 93–122 ignorance of, 91–110 substantival, 96, 96–102, 118 van Cleve’s characterization, 94–95n.23 comparatively intrinsic properties, 93–94, 94–96, 100–101, 103, 107, 112–4 van Cleve’s characterization 94–95n.23 purely intrinsic properties, 92–93, 92–122 Intrinsicness principle, 92, 96, 101, 102, 109, 117 knowledge argument, 3–4, 9–46, 10–11, 48, 85, 108, 110 conceptual analysis response, 29–40 ignorance response, 69–84

189 Index of Topics

indexical response, 12–13 old fact/new guise response, 11–12, 14n.13, 26–27 phenomenal concepts response, 11n.5, 83–84 qualitative inaccuracy response, 9–46 Russellian monist response, 53–57, 63, 89, 109–10, 85–122, 127, 168 logical positivism, 5, 117–21, 124, 156 law, causal, 5, 53, 80, 91, 97, 117, 123–26, 145, 150, 158–61, 167–68 bridge, 125, 145 emergence, 115, 136–37, 145–47 interlevel, 115–16, 167–68 law spheres, 124 and Dooyeweerd 124 luminosity, 15 made-up-of relation, 138–40 material coincidence, 138, 138n.44, 138–40 material constitution, 131–32, 135–41 materialism, see physicalism matter, 91–101 mental causal efficacy, 148–50 mental causal powers, 128–69 micropsychism, 110, 114–5, 122 microrealization, 136 misrepresentation, 3–4, 9, 14–46, 58–63, 85, 108–9, 122, 127, 168, 170–71 Descartes’s tower, 18, 23 Müller-Lyer lines, 18, 23 phenomenal, 3–4, 9, 14–46, 58–63, 85, 108–9, 122, 127, 168, 170–71 secondary quality 15–40, 58–63 modal arguments for nonidentity, 131–32, 142–3, 166–7 mode of presentation, 12, 12n.8, 26–28, 85 monads, 97 multiple realizability, 125–27, 133–39, 143, 144n.59, 148–67 significance for non-identity claims, 161–67 natural kinds, 5, 125, 152–53, 156 neuraesthenia, 161 non-Euclidean geometry, 39

nonreductive physicalism, 4–5, 123–72 and emergence, 7, 126, 136–37, 145–47 robust, 126–27, 148–9, 123–72 objectivity, 117–21 as metaphysical reality, 120 as rational intersubjective accessibility, 118 old fact/new guise response to the knowledge argument, 11–12, 14n.13 open possibility, 3–4, 14, 9–122, 170–2 ordinary and perfect content, 14, 26, 29, 35, 36–40, 171 overdetermination, 126–44 Bennett’s necessary condition for, 141–2 redundant, 126–44 pain, Kripke on, 55, 75–84 panpsychism, 7, 85, 89, 108, 110, 114–15, 122 perfect and ordinary content, 14, 26, 29, 35, 36–40, 171 phenomenal concepts, 29–40, 58–63 phenomenal concept strategy, 11n.5, 83–84 phenomenal misrepresentation, 3–4, 9, 14–46, 58–63, 85, 108–9, 122, 127, 168, 170–71 dentist example, 6, 23 fraternity example (Albritton/Hill), 22, 38 Leibnizian view, 9n.1, 114–15 Kantian view, 3–4, 9, 9n.1, 14–46, 58–63, 85, 108–9, 122, 127, 168, 170–71 phenomenal representation, theories of, 13, 19–20 causal, 19, 19–28 constitution, 20, 20–22 resemblance, 13, 13–16, 33–45, 83 self-presentation, 19–20, 19–22 phenomenal properties, 10, 10n.4, 3–122, 127, 148, 152–53, 168, 171–72 acquaintance with, 13, 28, 30–31, 36n.16, 42, 98–101, 106–7, 116 misrepresentation of, 3–4, 9, 14–46, 58–63, 85, 108–9, 122, 127, 168, 170–71 representation of, 3–4, 9–122, 171–2 physical, 5–8, 117–22 physicalism, 5–8, 117–22 argument from causal closure for, 172 best explanation argument for, 3, 172

190 Index of Topics

physicalism (continued) inductive argument for, 172 nonreductive, 4–5, 123–72 reductive, 4–5, 123–69 robust nonreductive, 126–27, 148–9, 123–72 subjective, 121n.48 twice grounded, in identity and constitution, 127, 171–72 positive conceivability, 24n.35, 50, 50–63, 86 positivism, 5, 117–21, 124, 156 possibility, 47–84 conceivability and, 50–52, 50–75 conceptual, 39, 51 epistemic, 14, 21, 50–51, 81 ‘could’ of evidential qualitative identity, 51 intuition of, 47, 54n.23 metaphysical, 50–52, 50–75 open, 3–4, 14, 9–122, 170–2 primary, 50–52, 50–65, 67–75 secondary, 50–52, 50–75 seeming , 47, 55, 69, 75–84 primary conceivability, 24n.35, 50–52, 50–65, 67–75, 86–89 primary intension, 50n.10, 55n.26, 63, 63–65, 88 primary possibility, 50–52, 50–65, 67–75 primary rigidity, 56, 56–57, 63–65 prime materiality, 85, 113 primitive properties, 3–4, 16–18, 30, 36–40, 59–63, 75–83 quasi-primitive properties, 59, 59–63 representing a property as primitive, 16–18, 30, 36–40, 59–63, 75–83 in a response to Kripke’s conceivability argument, 75–83 projectibility, 126, 152, 158–61 projectivism, 40, 40–43 literal, 41, 41–43 figurative 41, 41–43 and Hume, 40n.24 and Shoemaker, 41–43 protophenomenal properties, 4, 7, 53, 86, 89, 108–11, 114–17, 122 purely intrinsic properties, 92–93, 92–122 qualitative accuracy, for phenomenal representation, 13, 58n.36, 69, 85, 122, 171

qualitative inaccuracy hypothesis for phenomenal representation, 14, 3–4, 9, 14–46, 58–63, 85, 108–9, 122, 127, 168, 170–71 in a response to the knowledge argument, 9–46 in a response to conceivability arguments, 47–84 qualitative inaccuracy for secondary quality representation, 15–40, 58–63 qualitative natures, 3–4, 13–46, 47, 54, 57–59, 62, 79–84, 85, 108–110, 115, 170–71 quasi-primitive properties, 59, 59–63 quiddities, 105, 108 Ramsey sentences, 104n.5, 104–105, 119 realization, 5, 126, 128, 128–69. dimensional, 135–36 flat, 135–36 microrealization, 136 Shoemaker’s (subset) theory, 132–33, 130–35, 143–43, 157–58 reductionism, 4–5, 123–69 methodological versus metaphysical, 124–26 redundant overdetermination, 126–44 Bennett’s necessary condition for, 141–2 reflective universals (or concepts), 117 representation, theories of phenomenal causal, 15, 19–21, 23, 24n, 28, 36n constitution, 20 resemblance, 13, 13n.12, 16, 20 self-representation, 19–20 revelation, 17, 17n.20 robust nonreductive physicalism, 126–27, 148–9, 123–72 Russellian monism, 53–57, 63, 89, 109–10, 85–122, 127, 168 secondary conceivability, 50–52, 50–75 secondary intension, 50n, 55n, 63–64, 65, 88 secondary possibility, 50–52, 50–75 secondary qualities, 15–40, 58–63 analogy to phenomenal properties 15–40, 58–63 concepts of, 18, 29–40, 58–63

191 Index of Topics

secondary rigidity, 56 ship of Theseus, 131–32 solidity (perfect), 85, 97–100, 107–9, 113, 137n spatial coincidence, 138–40 structural properties, 91, 101,110–14, 118–19, 127n.10 ambiguity in the term ‘structural’, 127n.10 structuralism, 104n.5, 105–6, 108, 110n.21, 118n.41, 120n.45 causal, 105–6, 108 as the denial of absolutely intrinsic properties, 104n.5, 110n.21, 118n.41, 120n.45 subjective physicalism, 121n.48 substantival absolutely intrinsic properties, 96, 96–102, 118 subset view, 132–33, 130–35, 143–43, 157–58 T (that’s all), 53n.19 temporal arguments for nonidentity, 134 token identity, mental with abstract compositional properties, 127, 148–69, 171–72 token nonidentity of causal powers, 130–35

microphysical /mental, 5, 130–67 arguments for, 130–35, 166–67 neural/mental, 5, 130–35, 166–67 arguments for, 130–35, 166–67 type identity, mental with abstract compositional properties, 127, 148–69, 171–72 type nonidentity, 124–66 microphysical/mental, 5, 124–66 arguments for, 124–27, 161–66 neural/mental, 5, 124–66 arguments for, 124–27, 161–66 two-dimensionalism, 29–36, 47–69, 86–89 unity of science, 5, 120n.45, 124 universals (Kant, Critique of Judgment) 117 determinative, 117 given, 117 reflective, 117 well-founded phenomena, 97 zombie argument, 4, 7, 24n.35, 43, 47–75, 52–57, 83, 85–122, 171–2 zombies, denying the conceivability of, 57–63

This page intentionally left blank

INDEX OF NAMES

Adams, Frederick, 155n.22 Adams, Robert, 9n.1, 11n.5, 29, 43–45, 108 Albritton, Rogers, 22 Almog, Joseph, 47n.1 Alston, William, 16 Alter, Torin, 26–28, 68, 110–14 Antony, Louise, 160n.35 Armstrong, David, 19n.24, 87, 100n.39, 108, 110, 127n.10 Aristotle 85, 113 Arnauld, Antoine, 48, 69 Ayers, Michael R., 98n.30

Chignell, Andrew, 120n.46 Churchland, Patricia S., 164–68 Churchland, Paul M., 11n.7, 69, 151n.9, 164–68 Clapp, Lenny, 159n.32, 160n.36 Clarke, Randolph, 145n.63, 147 Collingwood, Robin G., 123–24 Conee, Earl, 15n.15 Crane, Tim, 6n.3 Crisp, Thomas, 142n.54 Cross, Troy, 87, 105 Culler, Jonathan, 120n.45

Baker, Lynne R., 127, 131, 135–41 Balog, Katalin, 38n.19 Bealer, George, 31n.8, 47, 50n.5, 51, 54n.23 Bechtel, William, 164 Bedau, Mark, 145n.63 Bennett, Karen, 73, 115, 135n.34, 138n.43, 142–3 Berkeley, George, 7, 32, 34, 97 Blackburn, Simon, 87n.5, 90–91 Block, Ned, 31n.8, 48–49n.4, 149, 152n.14, 159–60 Boghossian, Paul, 16n.18 Boscovich, Roger, 100n.40 Boyd, Richard, 5, 126n.9, 152–55 Braddon-Mitchell, David, 32n.11, 66–69 Brentano, Franz, 19 Broad, C. D., 72–75 Brown, Robin, 6n.4, n.5 Burge, Tyler, 140, 156n.24, 172 Byrne, Alex , 13n.11, 16, 19n.24 Carnap, Rudolf, 31n.8, 117–21, 124 Chalmers, David, 4, 6–8, 10–12, 16n.16, 20, 24n.35, 26–31, 36–40, 43, 45, 47–65, 85–91, 94, 100n.37, 108–117, 118n.41, 170

Davidson, Donald, 134n.32 Dennett, Daniel, 21–24 Derrida, Jacques, 120n.45 deRossett, Louis, 18n.23, 20n.30, 115n.32 Descartes, René, 7, 17–18, 23, 39, 46–47, 69, 73–74, 82, 91–93 Dilthey, Wilhelm, 123–24 Dooyeweerd, Herman, 124 D’Oro, Guiseppina, 123n.1 Dowell, Janice L., 6, 39–40, 62 Dunn, J. Michael, 92n.15 Dupré, John, 145n.63 Feigl, Herbert, 118, 121n.48 Fine, Kit, 138n.43, 144n.58 Fodor, Jerry, 5, 125–26, 134n.32, 158n.28, 159n.31 French, Steven, 117 Funkhouser, Eric, 127n.11 Galileo Galilei, 34, 45, 58–59 Gertler, Brie, 48–49n.4, 74–75 Gillett, Carl, 135–36, 139, 145n.63, 163n.41 193

194 Index of Names

Harré, Rom, 99n.35, 128n.15 Hawthorne, John, 12n.10, 15, 66–69, 83, 90n.12, 105 Heil, John, 87, 98n.34, 128n.15, 135n.33 Hilbert, David, 16 Hill, Christopher, 20n.26, 22–23, 83–84 Holden, Thomas, 90n.13, 98, 99n.36, 100n.40, 101n.43 Horgan, Terence, 11n.7, 130n.23 Howell, Robert, 121n.48 Humberstone, Lloyd, 92n.15 Hume, David, 24n.34, 40n.24, 90, 98–99, 107, 119–20, 128 Jackson, Frank, 3–4, 9–46, 48, 48–49n.4, 50n.10, 52–60, 87, 131n.24 Jacovides, Michael, 16n.17 Jaworski, William, 158n.28 Johnston, Mark, 17n.20, 132n.28 Kant, Immanuel, 3–9, 28, 46, 86, 89, 92n.16, 93–94, 97, 99n.36, 100–104, 106–109, 112, 114, 116–17, 120n.45, 121, 123–24, 172 Kim, Jaegwon, 92n.15, 94n.23, 126–61, 164n.42, 168 Kitcher, Philip S., 144n.59, 154n.18 Kobes, Bernard W., 172n.2 Kornblith, Hilary, ix , 126n.7, 131n.27, 135n.34, 158n.28, 161, 164n.44, 167–68 Kriegel, Uriah, 19n.25 Kripke, Saul, 17, 38, 39n.21, 47–48, 51, 55, 66–67, 75–84, 131n.27

Locke, John, 13, 15–18, 31, 49, 85, 97–100, 113, 137n.41 Loewer, Barry, 142n.54 Lycan, William G., 11n.7, 12 Madden, E. H., 99n.35, 128n.15 Makkreel, Rudolf, 123n.1 Martin, Charlie B, 87 Matthews, Gareth, 131n.26 Maxwell, Grover, 104n.5 Maund, Barry, 16n.18 McKitrick, Jennifer, 98n.32, 109n.19, 117 McGinn, Colin, 7–8, 30–36, 105n.11, 110, 116–17, 136n.38 McGonigal, Andrew, 93n.22 McLaughlin, Brian, 20n.26, 126n.8, 145n.63 Mellor, Hugh, 6n.3 Melnyk, Andrew, 136–39, 149 Moyer, Mark, 20n.30, 131, 132n.28 Mucciolo, Laurence, 164n.42 Mundale, Jennifer, 164 Nagel, Ernest, 146 Nagel, Thomas, 7–8, 10n.2, 83, 110, 114n.30, 116, 120, 121n.48 Newman, M. H. A., 119n.43, 120n.45 Newton, Isaac, 85, 97–100, 113, 137n.41, Ney, Alyssa, 93n.19, 99n.36, 105n.6, 107, 117n.37, 157–58 Nida-Rümelin, Martine, 10n.2 Noonan, Harold, 132n.28 O’Connor, Timothy, 145n.63, 146–47

Ladyman, James, 6n.4, n.5, 104n.5, 109n.19, 117, 119n.43 Langton, Rae, 92n.15, 99n.36, 103–106, 107n.12 Leibniz, Gottfried Wilhelm, 7, 9n.1, 85–86, 90–97, 101, 102, 108–109, 114, 117–18, 120n.45 Levine, Joseph, 13n.11, 24n.35, 29, 43–46, 160n.35 Lévi-Strauss, Claude, 120n.45 Lewis, David, 7, 17n.20, 64, 86, 92n.15, 104–105, 108–109, 117 Loar, Brian, 11n.7, 24n.35, 30, 43 Locke, Dustin, 104n.4, 108

Papineau, David, 20–21, 30n.3, 172 Pargetter, Robert, 87n.5 Parkinson, G. H. R., 9n.1 Paul, L. A., 127, 133n.30, 157–58 Pereboom, Derk, ix , 26n.38, 27n.40, 73n.17, 92n.16, 100n.38, 103n.1, 107n.12, 126n.7, 131n.27, 138n.42, 141n.50, 151n.10, 152n.13, 156n.24, n.25, 164n.44, 167n.51 Perry, John, 12n.10 Pettit, Philip, 131n.24 Place, U. T., 148 Polger, Thomas, 151n.9, 154n.19, 155n.22

195 Index of Names

Prior, Elizabeth, 87n.5 Putnam, Hilary, 5, 29–31, 124–25, 129, 151–53, 154n.18

Stoljar, Daniel, 6n.5, 11n.5, 24n.35, 44n.32, 66, 69–75, 81, 83–84, 87n.2, n.5, 89–91, 110–14 Strawson, Galen, 45n.35, 91n.14, 108–10

Ramsey, Frank P., 104–105, 119 Rickert, Heinrich, 123 Robb, David, 127n.13, 128n.15 Rosenberg, Gregg , 87n.2 Ross, Don, 117 Rozemond, Marleen, 47n.1 Russell, Bertrand, 4, 30, 53–55, 57, 63, 85–122, 127, 168 Rueger, Alexander, 145n.63

Thompson, Brad, 12n.8 Thomson, Judith, 138–39n.44 Tye, Michael, 11n.7

Saucedo, Raul, 139n.45 Shagrir, Oron, 164 Schaffer, Jonathan, 96n.26, 104n.4, 105, 138n.43, 145n.63 Schroeter, Laura, 61n.37 Shapiro, Lawrence, 162–63 Shoemaker, Sydney, 19n.24, 28n.41, 29, 40–43, 82n.28, 99, 105, 108, 128, 130–36, 142, 149 Sider, Theodore, 137, 142n.52 Silins, Nico, 9n.1, 25n.36, 28n.41, 68n.7, 114n.31 Smart, J. J. C., 148 Smith, Michael, 87n.5 Smullyan, Raymond, 21 Sober, Elliott, 129n.18 Stalnaker, Robert, 12n.10, 31n.8, 33n.13, 48–49n.4, 61n.38, 66–69, 83

van Cleve, James, 93, 94–95n.23, 103–104 Van Gulick, Robert, 11n.7, 153n.17 van Inwagen, Peter, 93n.22, 137n.41 Velleman, David, 16n.18 Warfield, Ted, 142n.54 Wasserman, Ryan, 132n.28, 138n.43, 138– 39n.44 Weatherson, Brian, 20n.30, 92n.15 Williamson, Timothy, 15 Wilson, Margaret, 47n.1 Wilson, Jessica, 6–7, 39n.21, 132–35, 136n.38, 137n.39, 142–43 Windelband, Wilhelm, 123 Witmer, Gene, 143n.57 Wykstra, Stephen, 24n.34 Yablo, Stephen, 32n.10, 47n.1, 50n.11, 51–52, 66, 75–84, 127, 129n.18, n.19, 131n.27, 154–55 Zimmerman, Dean W., 128n.15, 137n.41, 138–39