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 Irreduc-
ibility,” 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 con-
tents 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 Univer-
sity, 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 Philo-
sophical 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 explana-
tion, and the greatest obstacle to the success of such an argument derives from
phenomenal consciousness. The knowledge and conceivability arguments, which art-
iculate this impediment, have received increasingly sharper formulations, and con-
vincing responses have proven difficult to produce. These two arguments crucially
presuppose that we introspectively represent phenomenal properties as having spe-
cific qualitative natures that are distinct from any features that physical theories repre-
sent 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 rep-
resents 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 prop-
erties, 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 phe-
nomenal 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 knowl-
edge 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 con-
sciousness. While there are nonphysicalist versions of this view, some are amenable to
physicalism. In Chalmers’s terminology, the variants that are potentially physicalism-
friendly 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 repre-
sent 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 metaphys-
ically 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); “Conscious-
ness 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 con-
ceive 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 pos-
itivists 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 metaphys-
ical. 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 psy-
chological and microphysical entities of any sort—including causal powers. The deep-
est 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 the-
ories, 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 agnosti-
cism 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 constitu-
tion. 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 Philos-
ophy 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 con-

cept, 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 some-
what 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 men-
tal entities, even if they come to be accepted by future physics. As she argues, “Physical-
ists 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 phys-
ical.” 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 men-
tal.”5 Wilson’s type of approach is fairly liberal. Physicalism can, for instance, allow pos-
iting consciousness to explain the collapse of the wave function if consciousness itself
has a physical account or, supposing physicalism to be making a methodological rec-
ommendation, 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 counterin-
tuitive. 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 panpsy-
chist 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 men-
tal 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 hypo-
thesis 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 physi-
calism, 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 funda-
mental 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 Univer-
sity 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 uni-
verse—the totality of the actual concrete contingent entities—are physical. My con-
tention 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. Intro-
spective representations thus mediate the subject’s awareness of those mental states,
rendering this awareness in a sense indirect. As a consequence, the subject may repre-
sent 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 mul-
titude 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 con-
sciously 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 intro-
spectively represented by us as being that way, while the stronger, distinctively Kantian claim I am sin-
gling 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 knowl-
edge of every entity that is either microphysical or else wholly microphysically consti-
tuted, 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 phenom-
enal 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,” Phil-
osophical 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 Con-
sciousness: 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 para-
digms, 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 phys-
ical, 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 Con-
cepts,” 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 particu-
lar 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 fun-
damental 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, “Reduc-
tion, 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 knowl-
edge 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 intend-
ing 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 Gen-
dler 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, Possi-
bility, and Consciousness (Cambridge: MIT Press, 2001); see also John Hawthorne, “Advice for Phys-
icalists,” 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-like-
to-sense-red way, and it is intuitive that a qualitative nature that resembles this what-it-
is-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 charac-
terizations, one might say simply that the qualitative nature of the phenomenal prop-
erty 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 discred-
ited 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-is-

like-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 repre-
sent 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 physi-
calist 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 prop-
erties 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 intro-
spective 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 presen-
tation 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 demon-
stratives. 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 presen-
tation 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 phe-
nomenal properties and the qualitative natures we introspectively represent them as
having is consistent with certain claims about the correctness of introspective repre-
sentation. 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 phenom-
enal 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 phenom-
enal 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 ar-
gument 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 phe-
nomenal state one is in is consistent with the position that, in general, there is a dis-
crepancy 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 neverthe-
less 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 pro-
poses 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 them-
selves, 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 tempera-
ture 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 posi-
tion 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 at-
tribute 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, prop-
erties 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 sen-
sory 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 qual-
itative essence or, alternatively, we are disposed by introspection to believe that its
complete qualitative essence is introspectively represented, and also that we repre-
sent 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,” Philosoph-
ical 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 Univer-
sity 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 chal-
lenge 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 out-
lined, is controversial. Some would deny that there is any sense in which our ordi-
nary 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 argu-
ment. 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 repre-
sented 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 cer-
tain qualitative natures they are represented as having. In these respects, the inac-
curacy 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 pro-
vide 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 sec-
ondary 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 representa-
tion 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 rep-
resentations, they are typically caused by external objects (perhaps by their property-
instances) 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 sensa-
tion 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 self-
blind 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 self-
blindness 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, consti-
tutive 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-presen-

tation 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 con-
ception, “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 expe-
rience: —,’ 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 experi-
ence that appropriately resembles this experience or its faint copy.29 Here again, a phe-
nomenal 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 recogni-
tional 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 reason-
able 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 repre-
sentations of phenomenal properties would be an open possibility even if phenom-
enal 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 repre-
sented 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 accu-
rately represented by our ordinary sensory representations, but how could the qualita-
tive 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 real-
ity/appearance distinction at the heart of human subjectivity, but as well as
being scientifically unmotivated, this is metaphysically dubious, because it cre-
ates 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! (Smul-
lyan (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 qualita-

tive 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 rep-
resented 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 phenom-
enon. 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 fea-
tures 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 Univer-

sity Press, 1991), pp. 128–29.
33 Hill takes this example to provide evidence that we make errors of judgment in our introspec-
tion-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 Novo-

cain 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 introspec-
tively 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 phe-
nomenal 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 phe-
nomenal 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 character-
istic 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 knowl-
edge. 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 com-
plete 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 pos-
sibility 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, phenom-
enal 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 introspec-
tive 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 con-
tent, 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 phenom-
enal 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 phe-
nomenal mode of presentation—call this MPR. So Mary does not learn that the sen-
sation 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 sensa-
tions in the what-it-is-like-to-sense-red phenomenal way. Could Mary derive the cor-
responding 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 hypo-
thesis 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 representa-
tion. 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 phys-
ical 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 introspec-
tive 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 unwel-
come 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 pre-
sentation 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 intro-
spective 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 necessi-
tated. 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 representa-
tions 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 con-
tender, 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 phe-
nomenal state or property to its introspective phenomenal mode of presentation. But
the qualitative inaccuracy move can be reiterated for such introspective modes of pre-
sentation. This response does not require, incredibly, that when I introspect phenom-
enal 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 every-
thing 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 sur-
prising 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 presen-
tation. 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 physi-
calist with an effective reply to those, like Joseph Levine and Robert Adams, who sug-
gest 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 repre-
sented 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 con-
tending 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 con-
cepts 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 metaphysi-
cally 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 con-
sciousness, 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 con-
cepts is dependent on acquaintance with what they stand for. He then suggests:

The general concept of consciousness is acquaintance-based: we are acquaint-

ed with consciousness and our concept of consciousness depends upon this
acquaintance. To have the concept of consciousness is to know what con-
sciousness is (in one sense), and this knowledge is produced by our acquain-
tance 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 conscious-
ness 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 con-
cepts 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 objec-
tion can be contested. Although it is attractive to suppose with McGinn that an
acquaintance-based general concept of consciousness gives us knowledge of the es-
sence 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 phe-
nomenal 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 ante-
cedents of the conditionals specify possible worlds considered as actual—that is, pos-
sible 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 con-
siders 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 Chalm-
ers 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 de-
veloped 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 well-
behaved 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 composi-
tional stuff;
and
if a world is actual in which the watery stuff has a small number of sorts of compo-
sition, 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,” Jour-
nal 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 Uni-
versity 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 actu-
ally 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 con-
junct 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 rudimen-
tary change from a situation in which we rely only on the manifest image for its condi-
tions 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 resem-
bles 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 Anti-
Metaphysical 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 pro-
posed 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 resem-
bles 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 result-
ing 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 prop-
erties. 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 ‘phe-
nomenal 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 characteriza-
tion, 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 under-
standing 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 phys-
ical 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 expe-
rience 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 dispo-
sition, 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 dispo-
sitional 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 percep-
tion. Such perception represents physical objects as primitively colored, but they are
not primitively colored, while at the same time they are colored. Visual color percep-
tion sorts colors quite correctly but represents something else about them inaccu-
rately. 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 phe-
nomenal 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 ordi-
nary 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 rep-
resentations of phenomenal redness.
Consider two possible proposals for the ordinary content of representations of phe-
nomenal 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 prop-
erties generally, not to be functional in the sense of properties whose essences are ex-
clusively 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 con-
sequence of OC-P2, recall the example of the fraternity pledge discussed in the pre-
vious 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 possi-
bility 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 intui-
tion that Mary learns something about phenomenal redness upon having an experi-
ence 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 pre-
dicts 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 intro-
spective 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 pos-
sibility 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 ev-
idence 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 Des-
cartes, 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 non-
things,” while in the Principles of Philosophy of 1644 he confidently affirms that sec-
ondary 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 Phi-
losophy,” Philosophical Studies 140 (2008), pp. 19–46.
21 Janice Dowell, “Empirical Metaphysics: The Role of Intuitions about Possible Cases in Phi-
losophy.” 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 red-
ness 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 phe-
nomenal 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 prop-
erties 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 com-
mitted 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 pro-
jectivism 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 projec-
tivist threat to his own position:

I mentioned two features of my view that might seem to commit me to projec-

tivism. 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 represen-
tation 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 pro-
jected 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 misrepre-
sent 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 ordi-
nary content, even assuming that our visual color sensations misrepresent its qualita-
tive 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 phenom-
enal modes of presentation accurately represent the qualitative natures of the prop-
erties 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 presenta-
tion corefer, for the reason that the phenomenal concept correctly applies to a primi-
tive phenomenal property or a property that has a qualitative nature accurately
represented by introspection, while this is not the property the physical concept ex-
presses. However, if introspective representations of phenomenal properties are qual-
itatively 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 Phenomeno-
logical 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 repre-
sent this property as having. It is then an open possibility that how R seems is a straight-
forwardly 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 possi-
bility 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 prop-
erty 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 dis-
tinction 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 acces-
sible from the third-person perspective. But this is not enough to make the materi-
alism 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 phe-
nomenal 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 re-
semble our sensory representation of it.34 An Edenic content of our temperature con-
cept 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 plau-
sible 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 instanti-
ated. 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 contem-
porary 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 phe-
nomenal 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 anti-
physicalist 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 characteriza-
tion, 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 phenom-
enal-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 inac-
curacy.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 pro-

posal 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 meta-
physically possible, and conclude that physicalism is false.1 Sometimes the conceiv-
ability 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 Epistemol-
ogy and the Rationalist Renaissance.” For expositions of Descartes’s argument, see Margaret Wil-
son, 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 Two-
Dimensional 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 phe-
nomenal 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 conceiv-
able 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 con-
sciousness, and that it therefore sustains valid reasoning to the conclusion that a zom-
bie 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 dual-
ism; “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 sup-
ported 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 con-
cepts 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 ordi-
nary 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 ac-
tual world. If the qualitative natures these phenomenal properties are introspectively
represented as having are incorrectly attributed to them and are features these prop-
erties actually lack, then it might well be that all facts about the real natures of phe-
nomenal 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 ev-
ident 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 con-
ceivability 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-conceiv-
ability, 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 per-
forming 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 fundamen-
tal 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 Re-
duction, 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 argu-
ment.5 To begin, he distinguishes several dimensions of conceivability: prima facie ver-
sus 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 ‘con-
ceivable 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 conceiv-
ability 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 deter-
mine whether a statement S is primarily possible, one needs to evaluate indicative con-
ditionals 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 formaliza-
tion is controversial and has given rise to criticism; see, for example, George Bealer, “Modal Epis-
temology 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 inten-
sion’; 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 sugges-
tion; 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 explic-
itly 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 state-
ment 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-neu-
trality, which he defines as follows: “it is possible that p in the sense of qualitative-evidential-neutral-
ity 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 iden-
tical 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 circum-
stances 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 any-
thing 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 condi-
tionals 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 con-
struing 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 no-
tions 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 un-
derstandable 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 plati-
tudes 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 conse-
quence, 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 propo-
sition 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 rel-
evant 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 categor-
ical and intrinsic protophenomenal properties, so-called because while not phenom-
enal 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 possi-
ble 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 justifica-
tion for believing this can be defeated by better reasoning, reasoning that in this con-
text 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 en-
tities.21 On a microphysical alternative, ‘P’ features only microphysical information,
including information about microphysical laws. Since Chalmers employs this micro-
physical 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 terminol-
ogy 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 con-

cepts 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 microphys-
ical conception of the actual world, together with the minimum phenomenal concep-
tual 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 consid-
ered 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 neces-
sity: “a statement that is falsified by some positively conceivable situation (considered
as actual), but which is nevertheless true in all possible worlds (considered as ac-
tual).”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 expla-
nation as to why primary conceivability does not entail primary possibility.23 If a state-
ment 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 Possibil-
ity?” p. 156.
24 David Chalmers, “Materialism and the Metaphysics of Modality,” pp. 480–91; “Does Con-
ceivability 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 conceiv-
able 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 pos-
sible 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 con-
tingent—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, Chalm-
ers 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 in-
tension 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 contin-
gency; 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 de-
velop the zombie argument. A pure phenomenal concept “characterizes the phenom-
enal 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 (consid-
ered 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 consid-
ered 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 de-
monstrative concept ‘this property of my experience’ in appropriate circumstances, but a pure phe-
nomenal 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 pri-
marily 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 em-
pirical 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 pre-
mise 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 con-
sidered 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 ac-
tual 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 Rus-
sellian monism, rejecting Premise (3) requires denying that pure phenomenal con-
cepts 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 pre-
mise 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 pri-
marily 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 Con-
ceivability 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, phenom-
enal redness, contains the what-it-is-like-to-sense-red qualitative nature as it is intro-
spectively 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 intro-
spectively 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 estab-
lished, 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 in-
accurate. 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 phenom-
enal 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 intro-
spect a secondary quality or a phenomenal property, we thereby represent it as having a specific qual-
itative 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 rep-
resentation, 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 mi-
crophysically, but without phenomenal properties themselves, is not ideally, positively,
primarily conceivable. On analogy with the color case, the correct analysis of our phenom-
enal 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 quasi-
primitive 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 quasi-
primitive 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-prim-
itively red physical objects, one of these conditionals (or a similar alternative) plausi-
bly 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 repre-
sentations 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-primi-
tive phenomenal redness, but there are many different sorts of causes of their
introspectively appearing phenomenally red, and there are no salient similar-
ities 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 red-
ness, 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 phe-
nomenal redness, a conditional of this sort plausibly applies. Phenomenal color prop-
erties would then be instantiated by experiences in that scenario, and they would be
straightforwardly physical. Given the a priori derivability of instantiations of straight-
forwardly 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 inac-
curacy hypothesis is an open possibility, the analysis of phenomenal concepts dis-
closes 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 opposi-
tion 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 rel-
evant sentence [P and ~Q]. One explanation is that according to our commonsense under-
standing, 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 common-
sense 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 pri-
ori 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 con-
tends, the openness of this possibility can be grounded in our envisioning future em-
pirical 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 pos-
sibility 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 struc-
ture 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 con-
tent 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 con-
cepts 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-like-
to-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 phenom-
enal-redness, and ‘Q*’ a truth about R*. The suggestion is that a successful conceiv-
ability 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 Phi-
losophy.”
 63 Conceivability Arguments and Qualitative Inaccuracy

The response is the same: it is also an open possibility that our introspective repre-
sentation 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 micro-
physically identical to ours but without instantiated quasi-primitive introspective
modes of presentation of phenomenal properties, R* would nevertheless be instanti-
ated. 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 disanal-
ogy 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 deter-
mine 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 con-
cept 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). As-
sume again that ‘Q’ is an actual truth about phenomenal property R and that phenom-
enal 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 em-
pirical 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 pri-
mary 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 condi-
tional ‘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 mate-
rialism.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 pos-
sible 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 conceiv-
ability 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 phe-
nomenal 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 qual-
itative inaccuracy hypothesis. This challenge crucially involves the claim that we intro-
spectively 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 conceivabil-
ities 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 Physical-
ists”; 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 prop-
erties, 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 mate-
rialists 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 pos-
sible 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 con-
ceivable 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 con-
ceivability to primary possibility. So his discussion leaves this argument untouched.”6
Do Stalnaker’s and Hawthorne’s reflections serve to explain away the zombie intui-
tion, the sense that

(a) ‘P and ~ Q’ is ideally primarily, positively conceivable?

2 John Hawthorne, “Advice for Physicalists,” p. 25.

3 Robert Stalnaker, “What Is It Like to Be a Zombie?” p. 399.
4 Robert Stalnaker, “What Is It Like to Be a Zombie?” p. 399.
5 David Chalmers, “Does Conceivability Entail Possibility?” p. 198–99.
6 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 metaphysi-
cally 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 concep-
tion 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 orac-
ular 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 Condi-
tional 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 cau-
tiously, 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, posi-
tive, 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 con-
ceivability 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 explana-
tion, 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 experien-
tial. 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 persua-
sive. 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 prop-
erties that underlie fundamental dispositional properties; Stoljar call this fundamen-
tal-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 simplic-
ity 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 princi-
ples 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 combi-
nation 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 illustra-
tion 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 inter-
esting 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 rele-
vant 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 conclu-
sion 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 igno-
rance, 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 Black-
well 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 signif-
icant 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 con-
sciousness. 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. Articu-
lating 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 propo-
nents 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 phenom-
ena 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 sup-
port 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 situa-
tion, 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 cred-
ibility 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 possi-
bility 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 alter-
native 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 possi-
bility 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. Coun-
ter-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 counterfac-
tual 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 some-
thing 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 ade-
quate 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 counter-
parts 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 seeming-
possibility 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 psycho-
analytic 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” intui-
tion 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 red-
ness 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 sec-
ondary 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 expe-
rience, 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 primi-
tivist 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 me-
chanics, 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 alter-
ations 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 prop-
erty 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 under-
standing 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 prim-
itive 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 repre-
sentation 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 primi-
tiveness 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 possibil-
ities’ 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 partic-
ular case is to explain away seeming metaphysical possibility, then there must be some-
thing 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 sen-
sation 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 misrepresenta-
tion, 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 fundamen-
tal property whose qualitative nature is not revealed in sensory or introspective ex-
perience 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 primi-
tive, 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 exten-
sion 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 spa-
tially 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 phenom-
enal properties are inaccurately taken to be primitive. When these properties are mis-
takenly taken to be primitive, the illusion is generated that it is metaphysically possible
that states or substances characterized by them should exist without more fundamen-
tal 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 pos-
sess 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 igno-
rance 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 suppo-
sition 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 phe-
nomenal 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 psychoana-
lytical 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 some-
thing 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 qual-
itative 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 hypo-
thesis 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 sym-
pathetically: “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 contin-
gent 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 contin-
gent.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 perceptu-
ally 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,” Philo-
sophical 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 strat-
egy 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 funda-
mentally 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 con-
cern starts with the claim that physical properties are extrinsic, while none are intrin-
sic, 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 funda-
mental sense, but at the same time these physical properties require grounding in fun-
damentally 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 Aristote-
lians suggested prime materiality, and Locke and Newton advocated solidity. Chalm-
ers 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 founda-
tional 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 Rus-
sellian 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 possi-
ble, 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; “Con-
sciousness and Its Place in Nature.”
 87 Russellian Monism I

foregrounds the distinction between dispositional and categorical properties.2 Fra-

gility 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 prop-
erty 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 dispo-
sitional 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 dispositional-
categorical 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 ex-
planatory 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,” Phi-
losophy 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 Philos-
ophy 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 ten-
dency 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 ap-
parent 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 sec-
ondary intension. It can reasonably be argued that physical concepts have their
reference fixed by some dispositional role, but refer to an underlying categori-
cal 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 prop-
erties. 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 phenom-
enal 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 Chalm-
ers’s conception, the protophenomenal option allows for physicalism, for even though
no concepts in current physics specify protophenomenal properties directly, these prop-
erties might nonetheless be similar enough to those over which current physics quan-
tifies 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. Exam-
ining 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 [in-
trinsic] 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 correspon-
dence).
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 disposi-
tions.12 This claim is sometimes grounded in a Humean metaphysics, according to
which all the fundamental properties of the world are categorical, and physical dispo-
sitional 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 igno-
rant? Stoljar specifies three pertinent classes of truths to which we do not lack cogni-
tive 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 configura-

tions of things—hard, massy, shaped things, resisting penetration and displace-
ment 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 dy-
namical 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 disposi-
tions 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 rela-
tional 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 dispo-
sitional 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 fun-
damental 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 cate-
gorical properties, but perhaps a distinction between properties that are categor-
ical 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 rela-
tions 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 spatio-
temporal 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 Straw-
son’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 rela-
tional 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 posi-
tion 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 de-
nomination 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 in-
telligence. 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 Phi-
losophy. 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, “Psy-
chophysical 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 dis-
cussion 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 Des-
cartes, 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 Cri-
tique 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 Uni-
versity Press, 1987). James van Cleve, “Inner States and Outer Relations: Kant and the Case for Monad-
ism,” 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 con-
stitute 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 com-
mitted than the notion of reduction or analysis in this context, and because these ver-
sions 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 proposi-
tion 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 defini-
tions 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 compara-
tively 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 any-
thing 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 multi-
colored is analyzable into, or reduces to, intrinsic properties of two of its parts, namely, one hemi-
sphere’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 primi-
tively 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 multicol-
ored 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, fur-
ther 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 intui-
tive 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 com-
pleted 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 abso-
lutely 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, Spec-
imen 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 prop-
erty 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 primi-
tive force is found in the representational states of the monad, and it is these nonphys-
ical 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 funda-
mental 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 ulti-
mately 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 men-
tal 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 abso-
lutely intrinsic physical property. Locke and Newton were corpuscularians, that is, at-
omists. 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 prop-
erties, 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 pos-
sesses, 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 cer-
tain effects given certain circumstances, and the concern is that this is how Locke con-
ceives 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 cate-
gorical or underlying explanatory basis of absolute impenetrability. Do we have a con-
ception 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 Disposi-
tions,” 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 coun-
terfactual dependency among nondispositional property instances.35 Hume’s discus-
sion 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 ten-
dencies, 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 so-
lidity 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 con-
stitute 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 (Dor-
drecht, 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 sugges-
tion 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 substan-
tival 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 parti-
cles that are perfectly solid. The Locke-Newton proposal would then be that the abso-
lutely 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 prop-
erty 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-sub-
sistent and permanent, and through these we are given a determinate object.39

In material things we find comparatively intrinsic properties but no absolutely intrin-

sic properties. This is not merely an epistemic claim. Kant contends that all properties
of matter, substantia phaenomenon, even its apparently intrinsic properties, are abso-
lutely 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 rela-
tions 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 dis-
positional, and relational for that reason.)
Kant admits that there is something unintuitive about his view that all of the prop-
erties 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 phys-
ical 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 en-
tities—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 substan-

tial 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 prop-
erties, with (2) consciousness- or experience-relevance, the proposal that these categor-
ical 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 substan-
tival 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 interpreta-
tion 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 recep-

tive: 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 deduc-
tive; 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 argu-
ment for ignorance about the properties in question cannot assume such a deducibil-
ity requirement on knowledge.
David Lewis advances a related argument for this sort of ignorance.4 He has us ima-
gine a final theory of the world that incorporates not only all of the actually instanti-
ated 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 Lang-
ton 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 for-
malize 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 The-
oretical 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 espe-
cially 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, fol-
lowing Shoemaker, that we denied quiddities and instead endorsed a causal structur-
alist 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 sug-
gest, 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 vir-
tue 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 Prop-
erties,” 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 Mind-
Body 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 refer-
ence 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 possibil-
ities. What are the other options?

Ignorance from Lack of Acquaintance

The Kantian argument for ignorance that assumes a deducibility requirement presup-
poses 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 [prop-

erties] 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. Obser-
vation 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 intu-
ition 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 any-
thing 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 prop-
erties 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 pas-
sage suggests that these absolutely intrinsic properties are the nonsensible causes of
appearances. In my interpretation, he is thinking of these properties as the fundamen-
tal, 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 prop-
erties 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 fea-
tures 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 exis-
tence 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 under-
stand the complete nature of this property and that water has it. We have this knowl-
edge 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 igno-
rant of which absolutely intrinsic properties are instantiated. Several distinct candi-
dates 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 prop-
erties 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 individ-
uate 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 pro-
posal.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 men-
tioned that Chalmers suggests a protophenomenalist alternative. Armstrong once pro-
posed 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 compre-
hend 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 phe-
nomena, and none of which now convincingly does so. Unlike Kant’s or Lewis’s argu-
ments, 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-indepen-
dently 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 philos-
ophy of physics by structural realists, who claim that all physical properties reduce to
purely extrinsic or structural properties, and yet physical things are mind-indepen-
dently real.19 Kant would say that if there are no absolutely intrinsic physical prop-
erties, 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 Intrinsic-
ness 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 abso-
lutely intrinsic property. We now have in place an argument that we are currently igno-
rant 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 phys-
ical entities are not available by way of a priori reflection on ‘PT,’ and a posteriori phys-
ical 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 ver-
sions 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 sug-
gest (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 pro-
posals 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-and-
dynamics 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, Novem-
ber 2005.
22 This formulation is from Torin Alter, “Does the Ignorance Hypothesis Undermine the Con-
ceivability 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 conscious-
ness.
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 pos-
sible 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-struc-
ture thesis, but rather that we should not identify nonstructural/nondynamical prop-
erties 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-struc-
ture thesis.
The revised version of Chalmers’s proposal would then be that phenomenal prop-
erties of experiences are intrinsic to them, and not merely comparatively but abso-
lutely 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 experi-
ence. The from-structure-only-structure thesis becomes the claim that from facts
about extrinsic and merely comparatively intrinsic properties, no facts about abso-
lutely 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 struc-
tural 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 Argu-
ments?” 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 phenom-
enal 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-and-
dynamics 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 characteriza-
tions 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 informa-
tion 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 phys-
ical side, which, in my view, can be analyzed as purely extrinsic and merely compara-
tively 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 phenom-
enal-micropsychist and protophenomenal versions. On the phenomenal-micropsy-
chist option, which Strawson would endorse, the absolutely intrinsic properties that
account for our phenomenal consciousness are themselves phenomenal and irreduc-
ibly so, while on the protophenomenalist variant, they are not phenomenal but none-
theless 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 phenom-
enal truth about my current experience of blue. Would ‘P*T and ~ Q’ be ideally, posi-
tively, 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 blue-
ness 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 experi-
ence as having properties it actually lacks, but of the Leibnizian sort, by which intro-
spection 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 blue-
ness, 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 phe-
nomenal 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 Uni-
versity 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 phe-
nomenal micropsychism would need to posit fundamental laws linking the micropsy-
chist 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 prop-
erties 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**’ sup-
plements ‘P’ by adding the truths about protophenomenal absolutely intrinsic prop-
erties 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 cer-
tain 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 prop-
erties? 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 protophe-
nomenal 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 intro-
spective acquaintance-concepts on the one hand, and neural and other physical prop-
erties 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 differ-
ence 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 con-
stituents 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 univer-
sal rule, principle, or law. If the universal is given, Kant says, the judgment that sub-
sumes 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 uni-
versal 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 con-
structed 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 devel-

oped 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 antimeta-
physical conception, any terms for absolutely intrinsic properties are to be elimi-
nated 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 phys-
ical, 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 phys-
ical is expressible in a purely relational logical language. This sort of logical express-
ibility 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 rela-
tion, 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 in-
directly 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 philo-
sophical 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 phe-
nomenal 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 funda-
mental 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 regular-
ities 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, state-
ments 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 con-
sists 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 fur-
ther 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 objec-
tivity 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 illu-
minated 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 decon-
structive 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. Spi-
vak (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 acces-

sible 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 objec-
tively real, not merely a claim about what is objective in the sense of rational inter-
subjective 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 Physi-
calism,” 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 un-
derstand 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 accessibil-
ity, 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 abso-
lutely intrinsic properties are nonmental and similar enough to paradigmatic prop-
erties 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 intro-
spective 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 disadvan-
tage accepts the qualitative inaccuracy thesis. On this option, the knowledge and con-
ceivability 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 proto-
phenomenalism, 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 belief-
desire psychology, are importantly distinct from the natural sciences, paradigmatically
physics, and also chemistry and biology, together with its subfields, such as neuro-
physiology. 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 Sci-
ence” (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 Univer-
sity 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 partic-
ular events by means of such laws. Furthermore, on Carnap’s version of the unity doc-
trine, 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 nonre-
ductivists were largely methodological as opposed to metaphysical, in the sense that
they did not argue for the existence of distinct metaphysical levels of reality corre-
sponding 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 antimeta-
physical 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 compat-
ible with physicalism; in fact, on his nonreductive conception, it is not.
The renewed debate about reductionism in the 1960s and 1970s did feature a nonre-
ductivism 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 neurophysi-
ological 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 influen-
tial 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 prin-
ciple required for reducibility will not be available. If kinds in psychology are multi-
ply 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 disjunc-
tion 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 per-
tinent. 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 prag-

matic 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 well-
developed 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 metaphysi-
cally extravagant; the claim that the functionalism typically endorsed by nonreduc-
tive 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 pro-
jectible 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 signif-
icant differences between the controversial sort of emergentism and a plausible sort of
nonreductive physicalism and that a nonreductive physicalist need not be emergen-
tist 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 composi-
tional: 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 predic-
tively 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 con-
cerns 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 iden-
tity theorists have had for preferring their stance. By way of completing the two phys-
icalist accounts of phenomenal properties I’ve developed, the one that builds in
qualitative inaccuracy and the Russellian monist option, for each I endorse interpreta-
tions 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 psycho-
logical 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 concep-
tion. 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 apt-
ness 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 real-
ized—assuming no specific theory of realization—where realization is just, fol-
lowing 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 microphys-
ical 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 explana-
tion 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 dis-
tinct 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 overdeter-
mined, a result that is implausible. Let us call this phenomenon redundant overdeter-
mination (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 alterna-
tive 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 Superve-
nience 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 micro-
physical causal powers does it turn out that there is genuine mental causation, but
mental causal powers will then ultimately be microphysical. Psychological explana-
tions 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 regu-
larities 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 psy-
chological explanation refers are identical with those to which the underlying
microphysical explanation refers, then there will be no exclusion-generating compe-
tition 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 fol-
lowing 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 nonreduc-
tive 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 men-
tal property to be causally efficacious qua mental, its causal efficacy cannot con-
sist 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 micro-
physical 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 micro-
physical 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 iden-
tical 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 intui-
tion 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 constitu-
tion, 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 Perspec-
tive,” 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 Aristo-
tle endorsed such a notion of sameness alongside absolute identity; see Gareth Matthews, “Aristo-
tle’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 argu-
ment 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 real-
izer? 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 alterna-
tive; 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 alterna-
tive 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 micro-
physical causal powers. Sydney Shoemaker and Jessica Wilson have endorsed a
token-identity thesis for mental and lower-level causal powers, although they op-
pose 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 state-
ment 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 realiz-
ability 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 instan-
tiation 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?” Philosoph-
ical 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 micropro-
cessor—let’s call it a silicon prosthesis. Imagine that the illness never actually materi-
alizes, 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 sup-
position 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 pos-
sesses 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 men-
tal causal power cannot be identified with any neural token causal power, in partic-
ular, 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 neu-
ral-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 physi-
calism 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 “swal-
lowed 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 iden-
tity, 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 re-
striction, 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 with-
out 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 constitu-
tion. I will have more to say on this shortly.
Carl Gillett differentiates between a flat conception of realization, in which prop-
erties 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 constitu-
tion 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 prop-

erty 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 micro-
realized 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 phe-
nomena 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 consti-
tutes 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 accor-
dance 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 funda-
mental law of physical-to-mental emergence.38 We can grant that if this necessitation

36 Sydney Shoemaker, “Realization, Micro-Realization, and Coincidence,” Philosophy and Phe-

nomenological 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 commit-
ted 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 coinci-
dence, 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 consti-
tuting or realizing the other at all. If there is a corresponding metaphysical possi-
bility, 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 nonphys-
ical 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 solid-
ity, 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 lat-
tice 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 consti-
tuting 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 condi-
tions 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 irre-
flexive, asymmetric, and such that their input is more fundamental than their output. A third ele-
ment 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 connec-
tion 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. Chalm-
ers, 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 suf-
fice 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 fun-
damental to the less fundamental, and the made up of relation secures these character-
istics, 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 ‘constitu-
tion’ or ‘realization’? The former seems better because of the affinity between the con-
cepts ‘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 materi-
ally 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 materi-
ally coincident with x at t.

To avoid such a mereological characterization issuing in identity (we’re assuming the pluralist po-
sition 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 constitu-
tion 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 con-
stitutor, 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 environ-
ment, 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 circum-
stances—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 materi-
ally 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 micro-
physical 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 con-
ceived 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 sat-
isfied 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 phys-
ical.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 over-
determination would be objectionable here, see Ted Sider, “What’s So Bad about Overdetermina-
tion?” 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 iden-
tical with a forward-looking causal power of the physical property instance. Conse-
quently, 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 compo-
nents 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 overdeter-
mination 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 consti-
tutes 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 cur-
rent 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 con-
siderations 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 autono-
mous “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 redun-
dant 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 mi-
crophysical 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 counter-
factual 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 plau-
sibly 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 rele-
vant 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 prop-
erties that x and y fail to share. Kit Fine discusses such failures of property sharing in “The Non-
Identity 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 methodo-
logical 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 predict-
able 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 syn-
chronic and not diachronic predictability. Theoretical predictability contrasts with induc-
tive 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. Plausi-
bly, 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: Black-
well, 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 Timo-
thy 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 de-
velops 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 prop-
erty 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 higher-
level 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 real-
izers 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 controver-
sial 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 identi-
fying 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 real-
izer 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 partic-
ular 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 micro-
physical 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 emer-
gent 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 higher-
level 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 consid-
ering several objections that have been raised against nonreductive views generally, ar-
guing 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, nei-
ther 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 com-
mitment 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 ana-
lyzed, metaphysically, solely in terms of causal relations to perceptual inputs, behav-
ioral 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, Mel-
nyk 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 identi-
fied 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 prop-
erties 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 Psychol-
ogy, 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 functional-
ism, 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 pre-
serving nonreductive physicalism about the mental:

How should we counter this line of argument? I think it will be helpful to con-
sider 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 ER-
functionalist 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 iden-
tical 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 expla-
nation. By contrast, certain types of behaviorists identify mental states with for-
ward-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 for-
ward-looking causal relata of mental states much in the way that a viral infection pro-
vides an explanation for the symptoms of polio:

My own view is that psychological predicates correspond to functional prop-

erties 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 Nonre-
ductive 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: Black-
well, 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 expla-
nation 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 for-
ward-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 essen-
tial 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 rela-
tum 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 prop-
erties 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 them-
selves. 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 explan-
atory 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 overwhelm-
ingly 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 prop-
erties 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-functional-
ism, 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 fea-
tures 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 in-
ternally complex types, for instance, a ball piston engine, a recent version of the ro-
tary 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 Philos-
ophy 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 un-
likely 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 investi-
gating 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 deter-
minable 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 tita-
nium realizations of this property are determinates. Similarly, a mental compositional
property can be construed as a determinable relative to neural and silicon-based reali-
zations of that property. Considering mental properties as sufficiently abstract compo-
sitional 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. Yab-
lo’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 prop-
erty 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 real-
izable 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 Realiza-
tion,” pp. 91–96.
21 Stephen Yablo, “Mental Causation.”
 155 Mental Compositional Properties

compositional property is the determinable of which the steel realization is a determi-

nate. 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 dis-
tinct 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 real-
ized 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 compo-
sitional 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 in-
stantiations 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 under-
mined. It may be that these compositional types, although they fail to correlate
neatly with our ordinary mental state categories, are not specifically neural compo-
sitional 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 indi-
viduation 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 be-
ings 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, where-
upon those properties would be so abstract that they are not even essentially phys-
ical. 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 func-
tionally 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 gen-
eral antirealist point of view. Positivist antirealists advocated an operationalist
characterization of natural kinds, defining them in terms of their observable back-
ward- and forward-looking causal relata. Logical behaviorism provides a good ex-
ample 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 caus-
ally 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 redun-
dant 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 in-
stances, 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 nonreduction-

ist 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 hav-
ing 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 dif-
ferent 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 micro-
physical 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 microphys-
ical token.

26 Alyssa Ney, “Can an Appeal to Constitution Solve the Exclusion Problem?” Pacific Philosoph-
ical 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 effica-
cious. 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 microphys-
ical 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 pro-
vides a positive model of a property instance that is causally efficacious qua mental: an
instance of a requisitely abstract compositional property. Multiple realizability argu-
ments 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 consti-
tution 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 effica-
cious 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 replace-
ment by disjunctive properties—properties that are perhaps even wildly disjunctive
in the sense that the disjuncts have at best little in common. Moreover, the disjunc-
tions 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 state-
ments 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 pro-
jectible 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 law-
like 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 counterar-
gument 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 prop-

erty. Nomic equivalence might be defined in this way: Properties F and G are nomi-
cally 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 equiva-
lent 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 pro-
jectible 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 per-
fectly 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: Black-

well, 1997), pp. 107–32, at p. 109.
31 Jerry Fodor argues that Kim’s example merely involves a sampling error; see “Special Sci-
ences: 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 well-
behaved 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 signifi-
cant 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 pred-
icate 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 de-
scription 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 ex-
plains how, on a causal-powers notion of properties, a disjunctive predicate might indicate a nondis-
junctive 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 instruc-
tive 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 lack-
ing, 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 discov-
ered 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 uni-
tary 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 addi-
tion, researchers discovered a disjunction of properties coextensive with being neu-
raesthenia, 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 in-
stances 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 Uni-
versity, 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 physi-
calism 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 contrib-
ute 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 cork-
screws? 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 charac-
terization 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 iden-
tical.”40 But unlike color, being made of steel or of aluminum does make a causal con-
tribution 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 reduc-
tionism, we talk about multiple realizability. But in addition, the fact that the neural
system and the silicon-based system make identical contributions to psychological ca-
pacities seems to force us to say that the features thus contributed are neither essen-
tially 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 realizabil-
ity 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 character-
izable 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 ap-
pears to be a single mental state type, then it might be that these realizations corre-
spond 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, essen-
tially 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 metal-
and-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 neu-
robiology should prove capable of explaining all psychological phenomena in
humans is not threatened by the possibility that some other theory, say, semicon-
ductor 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 charac-
teristics 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, “Phe-
nomenal 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 tempera-
ture 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 composi-
tional 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 con-
ception. A key motivation for the Churchlands’ remarks is their sense that it is neu-
roscientific 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 neurosci-
ence. 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 reduc-
tionism, 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 dif-
ferences (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, “large-
scale electronic realizations . . . will explore entirely new horizons for informa-
tion 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 re-
duction. 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 identifica-
tion 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 argu-
ment 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 pro-
ducing its characteristic effects under these changes. But this is also the case for a neu-
ral 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 mi-
crophysical 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 micro-
physical 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 Sta-
tion, 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 cred-
ible that she could retain this token belief after it comes to be realized by a neural-and-
silicon-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 nonre-
ductivism has consequences for methodology and explanation in psychology. To pro-
vide 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 phys-
icalism 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 psy-
chological 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 under-
gone a lobotomy is best explained in terms of neural damage, and not by a psycholog-
ical 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 ap-
parently 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 phenom-
enon, and perhaps psychological explanations involving tiredness will be illuminated
and deepened when recast as involving this physiological phenomenon. But the re-
ducibility of some apparently psychological kinds to those in a more basic science

50 Patricia Churchland, Neurophilosophy:Toward a Unified Science of the Mind-Brain (Cam-

bridge, 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 psy-
chological 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 phys-
iological 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, prop-
erties 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-func-
tionalist 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 phys-
icalist 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 func-
tionalist characterizations did capture the nature of mental states, and some version of
nonreductive physicalism could accommodate ER-functionalism, then in an impor-
tant respect nonreductive physicalism would not depend on scientific investigation
for its validation. However, Kim is right to claim that ER-functionalism cannot coun-
tenance genuinely mental causal powers, and for this reason, ER-functionalism is in-
compatible 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 compo-
sitional 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 phe-
nomenal 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 pre-
sentation, 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 presenta-
tion 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), con-
joined 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 signif-
icantly 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 dis-
positions but also explain phenomenal properties as they are introspectively repre-
sented, without being phenomenal themselves. They would then be protophenomenal
properties. We currently have incomplete understanding of what such protophenom-
enal 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 proto-
phenomenal properties are instantiated that are similar enough to paradigmatic phys-
ical 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 nonphysi-
calist position can be affirmed by a nonreductive version of physicalism. On the non-
reductive 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 prob-
lem 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 fea-
tures 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 prop-
erties 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 phys-
ical, on which every physical event has in some sense a complete physical causal expla-
nation.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 psycho-
logical explanation does not require physicalist validation.2 About this he might well
be right. The metaphysical considerations with which we’ve been occupied, not pro-
vided 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 rea-
sonable 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 Phenomeno-
logical 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, Psychoanal-
ysis, 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 Possibil-
ity 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,” Philosophi-
cal 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: Cam-
bridge 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. Samu-
el 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 Philosoph-
ical 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 Univer-
sity 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 Univer-
sity.
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 Philosophi-
cal 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, forth-
coming.
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, Cam-
bridge: 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 Founda-
tions 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 Uni-
versity 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 Phenomeno-
logical 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 Reflec-
tions 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?” Philosophi-
cal 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 comparatively or relatively intrinsic properties,

a priori conceivability, 24n.35, 50–52, 50–65, 93–94, 94–96, 100–101, 103, 107, 112–4
67–75, 86–89 van Cleve’s characterization, 94–95n.23
a priori derivability, 11, 48, 48–49n.4, 94, 102 compositional properties, 126–27, 148–49,
absolutely intrinsic properties, 86, 93–94, 148–69, 171–72
93–122 conceivability, 47–122
ignorance of, 91–110 ideal 50, 50–86
substantival, 96, 96–102, 118 negative, 50
van Cleve’s characterization, 94–95n.23 positive, 24n.35, 50, 50–63, 86
accuracy, qualitative, for phenomenal repre- and possibility, 50–57, 63–65
sentation, 13, 58n.36, 69, 85, 122 prima facie, 50
acquaintance, with phenomenal properties, 13, primary, 24n.35, 50–52, 50–65, 67–75, 86–89
28, 30–31, 36n.16, 42, 98–101, 106–7, 116 secondary, 50–52
actual, conceiving as, 50–52, 50–69 conceivability argument, 3–7, 24n.35, 47–122
A-intension (see also primary intension), 50n.10 Chalmers’s 47–75, 52–57, 85–122
antirealism, 117–21, 152, 156 Descartes’s, 17, 47–48, 69, 82
Aristotelians, 85, 113 ignorance response, 69–84
indexical response, 12–13
behaviorism, 156 Kripke’s, 17, 47–48, 51, 55, 75–84
building relation, 138n.43 qualitative inaccuracy response, 47–84
Russellian monist response, 53–57, 63, 89,
C-intension (see also secondary intension), 109–10, 85–122, 127, 168
50n.10 conceiving as actual, 50–52, 50–69
Cartesian theatre, 21 conceiving as counterfactual, 50–52
categorical properties, 87, 85–122 conceptual analysis, 29–40
causal closure, 172 conceptual possibility, 39, 51
causal exclusion, 126–44, 146 constitution, 5–7, 126, 131–32, 135–41
statement of Kim’s exclusion argument, 129 Baker’s theory, 131–2, 135–41
causal inheritance principle, 130, 141 material, 131–32, 135–41
causal roles, 36, 48, 52, 87n.5, 88, 104–6, 108, 149 solution to causal exclusion problem, 141–44
causal structuralism, 105–6, 108 content, phenomenal, 14, 17–20, 26, 36–40
causal powers, 126–28, 126–69 externalism, 140, 156n.24
as concrete property instances, 127–8, internalism, 14, 156n.24
127n.13 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 God, 32–4, 67–8, 108, 156

determinable/determinate relation, 127, 154–55 and physicalism, 8
determinative universals (or concepts), 117 Goldbach’s conjecture, 69
dimensional realization, 135–6
dispositional properties, 86–122, 87 heat, Kripke on, 55, 75–84
disjunctions, open-ended or wild, 32–34, Humility, 100–102, 103, 103–10
125–26, 158–61
dual-content view, 14, 26, 29, 35, 36–40, 171 ideal conceivability, 24n.35, 50, 50–84
dualism 7, 44, 48n.4, 67–68, 75, 82, 83, 122, 124, idealism, 97, 101, 124
171–72 identity, 5, 127, 130–35, 148, 160–67
dynamical properties, 90–91, 110–14, 118 token, 5, 130–35, 166–67
type, 5, 127, 130–35, 148, 166–67
Edenic and ordinary content, 14, 26, 29, 35, ignorance,
36–40, 171 of categorical properties, 88–91
eliminativism, 29, 43–46 of absolutely intrinsic properties,
emergence, 7, 126, 136–37, 145–47 91–110
emergentism, 7, 126, 136–37, 145–47 of fundamental properties, 7–8, 88–110
epistemic rigidity, 56 as a response to the knowledge and
epistemic view, 69–75 conceivability argument, 69–84
exclusion, causal and explanatory, 124–44, 146 Stoljar on, 69–75
explanation, psychological, 123–30, 158–62, inaccuracy, qualitative, 3–4, 9, 14, 14–46
167–68 indexical response to the knowledge and
explanatory gap, 29, 43–46 conceivability arguments, 12–13
extensionality, principle 138–39n.44 introspection, models of, 19n.24
external-relations- (ER-) functionalism, 126, broad perceptual model, 19n.24
149, 148–58 introspective inaccuracy, 3–4, 9, 14–46, 58–63,
externalism about content, 140, 156n.24 85, 108–9, 122, 127, 168, 170–71
extrinsic properties, 85–122 intrinsic properties, 85–122
absolutely purely extrinsic properties, 94 absolutely intrinsic properties, 86, 93–94,
purely extrinsic properties, 92–93, 92–122 93–122
van Cleve’s characterization, 94–95n.23, ignorance of, 91–110
102 substantival, 96, 96–102, 118
van Cleve’s characterization, 94–95n.23
facsimile principle (Yablo), 75–82, 77 comparatively intrinsic properties, 93–94,
flat realization, 135–6 94–96, 100–101, 103, 107, 112–4
force, 96–100 van Cleve’s characterization 94–95n.23
derivative, 97 purely intrinsic properties, 92–93, 92–122
Leibniz on force, 97 Intrinsicness principle, 92, 96, 101, 102,
Kant on force, 101 109, 117
primitive, 97
as relational, 101 knowledge argument, 3–4, 9–46, 10–11, 48,
functionalism, 119, 126, 148–58 85, 108, 110
external relations (ER), 126, 149, 148–58 conceptual analysis response, 29–40
Shoemaker’s, 149–50 ignorance response, 69–84
 189 Index of Topics

indexical response, 12–13 nonreductive physicalism, 4–5, 123–72

old fact/new guise response, 11–12, 14n.13, and emergence, 7, 126, 136–37, 145–47
26–27 robust, 126–27, 148–9, 123–72
phenomenal concepts response, 11n.5,
83–84 objectivity, 117–21
qualitative inaccuracy response, 9–46 as metaphysical reality, 120
Russellian monist response, 53–57, 63, 89, as rational intersubjective accessibility, 118
109–10, 85–122, 127, 168 old fact/new guise response to the knowledge
argument, 11–12, 14n.13
logical positivism, 5, 117–21, 124, 156 open possibility, 3–4, 14, 9–122, 170–2
law, causal, 5, 53, 80, 91, 97, 117, 123–26, 145, 150, ordinary and perfect content, 14, 26, 29, 35,
158–61, 167–68 36–40, 171
bridge, 125, 145 overdetermination, 126–44
emergence, 115, 136–37, 145–47 Bennett’s necessary condition for, 141–2
interlevel, 115–16, 167–68 redundant, 126–44
law spheres, 124
and Dooyeweerd 124 pain, Kripke on, 55, 75–84
luminosity, 15 panpsychism, 7, 85, 89, 108, 110, 114–15, 122
perfect and ordinary content, 14, 26, 29, 35,
made-up-of relation, 138–40 36–40, 171
material coincidence, 138, 138n.44, 138–40 phenomenal concepts, 29–40, 58–63
material constitution, 131–32, 135–41 phenomenal concept strategy, 11n.5, 83–84
materialism, see physicalism phenomenal misrepresentation, 3–4, 9, 14–46,
matter, 91–101 58–63, 85, 108–9, 122, 127, 168, 170–71
mental causal efficacy, 148–50 dentist example, 6, 23
mental causal powers, 128–69 fraternity example (Albritton/Hill), 22, 38
micropsychism, 110, 114–5, 122 Leibnizian view, 9n.1, 114–15
microrealization, 136 Kantian view, 3–4, 9, 9n.1, 14–46, 58–63, 85,
misrepresentation, 3–4, 9, 14–46, 58–63, 85, 108–9, 122, 127, 168, 170–71
108–9, 122, 127, 168, 170–71 phenomenal representation, theories of, 13,
Descartes’s tower, 18, 23 19–20
Müller-Lyer lines, 18, 23 causal, 19, 19–28
phenomenal, 3–4, 9, 14–46, 58–63, 85, constitution, 20, 20–22
108–9, 122, 127, 168, 170–71 resemblance, 13, 13–16, 33–45, 83
secondary quality 15–40, 58–63 self-presentation, 19–20, 19–22
modal arguments for nonidentity, 131–32, phenomenal properties, 10, 10n.4, 3–122, 127,
142–3, 166–7 148, 152–53, 168, 171–72
mode of presentation, 12, 12n.8, 26–28, 85 acquaintance with, 13, 28, 30–31, 36n.16, 42,
monads, 97 98–101, 106–7, 116
multiple realizability, 125–27, 133–39, 143, misrepresentation of, 3–4, 9, 14–46, 58–63,
144n.59, 148–67 85, 108–9, 122, 127, 168, 170–71
significance for non-identity claims, 161–67 representation of, 3–4, 9–122, 171–2
physical, 5–8, 117–22
natural kinds, 5, 125, 152–53, 156 physicalism, 5–8, 117–22
neuraesthenia, 161 argument from causal closure for, 172
non-Euclidean geometry, 39 best explanation argument for, 3, 172
 190 Index of Topics

physicalism (continued) qualitative inaccuracy hypothesis for

inductive argument for, 172 phenomenal representation, 14, 3–4,
nonreductive, 4–5, 123–72 9, 14–46, 58–63, 85, 108–9, 122, 127, 168,
reductive, 4–5, 123–69 170–71
robust nonreductive, 126–27, 148–9, 123–72 in a response to the knowledge argument,
subjective, 121n.48 9–46
twice grounded, in identity and in a response to conceivability arguments,
constitution, 127, 171–72 47–84
positive conceivability, 24n.35, 50, 50–63, 86 qualitative inaccuracy for secondary quality
positivism, 5, 117–21, 124, 156 representation, 15–40, 58–63
possibility, 47–84 qualitative natures, 3–4, 13–46, 47, 54, 57–59,
conceivability and, 50–52, 50–75 62, 79–84, 85, 108–110, 115, 170–71
conceptual, 39, 51 quasi-primitive properties, 59, 59–63
epistemic, 14, 21, 50–51, 81 quiddities, 105, 108
‘could’ of evidential qualitative identity, 51
intuition of, 47, 54n.23 Ramsey sentences, 104n.5, 104–105, 119
metaphysical, 50–52, 50–75 realization, 5, 126, 128, 128–69.
open, 3–4, 14, 9–122, 170–2 dimensional, 135–36
primary, 50–52, 50–65, 67–75 flat, 135–36
secondary, 50–52, 50–75 microrealization, 136
seeming , 47, 55, 69, 75–84 Shoemaker’s (subset) theory, 132–33,
primary conceivability, 24n.35, 50–52, 50–65, 130–35, 143–43, 157–58
67–75, 86–89 reductionism, 4–5, 123–69
primary intension, 50n.10, 55n.26, 63, 63–65, 88 methodological versus metaphysical,
primary possibility, 50–52, 50–65, 67–75 124–26
primary rigidity, 56, 56–57, 63–65 redundant overdetermination, 126–44
prime materiality, 85, 113 Bennett’s necessary condition for, 141–2
primitive properties, 3–4, 16–18, 30, 36–40, reflective universals (or concepts), 117
59–63, 75–83 representation, theories of phenomenal
quasi-primitive properties, 59, 59–63 causal, 15, 19–21, 23, 24n, 28, 36n
representing a property as primitive, 16–18, constitution, 20
30, 36–40, 59–63, 75–83 resemblance, 13, 13n.12, 16, 20
in a response to Kripke’s conceivability self-representation, 19–20
argument, 75–83 revelation, 17, 17n.20
projectibility, 126, 152, 158–61 robust nonreductive physicalism, 126–27,
projectivism, 40, 40–43 148–9, 123–72
literal, 41, 41–43 Russellian monism, 53–57, 63, 89, 109–10,
figurative 41, 41–43 85–122, 127, 168
and Hume, 40n.24
and Shoemaker, 41–43 secondary conceivability, 50–52, 50–75
protophenomenal properties, 4, 7, 53, 86, 89, secondary intension, 50n, 55n, 63–64, 65, 88
108–11, 114–17, 122 secondary possibility, 50–52, 50–75
purely intrinsic properties, 92–93, 92–122 secondary qualities, 15–40, 58–63
analogy to phenomenal properties 15–40,
qualitative accuracy, for phenomenal 58–63
representation, 13, 58n.36, 69, 85, 122, 171 concepts of, 18, 29–40, 58–63
 191 Index of Topics

secondary rigidity, 56 microphysical /mental, 5, 130–67

ship of Theseus, 131–32 arguments for, 130–35, 166–67
solidity (perfect), 85, 97–100, 107–9, 113, 137n neural/mental, 5, 130–35, 166–67
spatial coincidence, 138–40 arguments for, 130–35, 166–67
structural properties, 91, 101,110–14, 118–19, type identity, mental with abstract composi-
127n.10 tional properties, 127, 148–69, 171–72
ambiguity in the term ‘structural’, 127n.10 type nonidentity, 124–66
structuralism, 104n.5, 105–6, 108, 110n.21, microphysical/mental, 5, 124–66
118n.41, 120n.45 arguments for, 124–27, 161–66
causal, 105–6, 108 neural/mental, 5, 124–66
as the denial of absolutely intrinsic prop- arguments for, 124–27, 161–66
erties, 104n.5, 110n.21, 118n.41, 120n.45 two-dimensionalism, 29–36, 47–69, 86–89
subjective physicalism, 121n.48
substantival absolutely intrinsic properties, unity of science, 5, 120n.45, 124
96, 96–102, 118 universals (Kant, Critique of Judgment) 117
subset view, 132–33, 130–35, 143–43, 157–58 determinative, 117
given, 117
T (that’s all), 53n.19 reflective, 117
temporal arguments for nonidentity, 134
token identity, mental with abstract well-founded phenomena, 97
compositional properties, 127, 148–69,
171–72 zombie argument, 4, 7, 24n.35, 43, 47–75,
token nonidentity 52–57, 83, 85–122, 171–2
of causal powers, 130–35 zombies, denying the conceivability of, 57–63
This page intentionally left blank
 INDEX OF NAMES

Adams, Frederick, 155n.22 Chignell, Andrew, 120n.46

Adams, Robert, 9n.1, 11n.5, 29, 43–45, 108 Churchland, Patricia S., 164–68
Albritton, Rogers, 22 Churchland, Paul M., 11n.7, 69, 151n.9,
Almog, Joseph, 47n.1 164–68
Alston, William, 16 Clapp, Lenny, 159n.32, 160n.36
Alter, Torin, 26–28, 68, 110–14 Clarke, Randolph, 145n.63, 147
Antony, Louise, 160n.35 Collingwood, Robin G., 123–24
Armstrong, David, 19n.24, 87, 100n.39, 108, 110, Conee, Earl, 15n.15
127n.10 Crane, Tim, 6n.3
Aristotle 85, 113 Crisp, Thomas, 142n.54
Arnauld, Antoine, 48, 69 Cross, Troy, 87, 105
Ayers, Michael R., 98n.30 Culler, Jonathan, 120n.45

Baker, Lynne R., 127, 131, 135–41 Davidson, Donald, 134n.32

Balog, Katalin, 38n.19 Dennett, Daniel, 21–24
Bealer, George, 31n.8, 47, 50n.5, 51, 54n.23 Derrida, Jacques, 120n.45
Bechtel, William, 164 deRossett, Louis, 18n.23, 20n.30, 115n.32
Bedau, Mark, 145n.63 Descartes, René, 7, 17–18, 23, 39, 46–47, 69,
Bennett, Karen, 73, 115, 135n.34, 138n.43, 142–3 73–74, 82, 91–93
Berkeley, George, 7, 32, 34, 97 Dilthey, Wilhelm, 123–24
Blackburn, Simon, 87n.5, 90–91 Dooyeweerd, Herman, 124
Block, Ned, 31n.8, 48–49n.4, 149, 152n.14, D’Oro, Guiseppina, 123n.1
159–60 Dowell, Janice L., 6, 39–40, 62
Boghossian, Paul, 16n.18 Dunn, J. Michael, 92n.15
Boscovich, Roger, 100n.40 Dupré, John, 145n.63
Boyd, Richard, 5, 126n.9, 152–55
Braddon-Mitchell, David, 32n.11, 66–69 Feigl, Herbert, 118, 121n.48
Brentano, Franz, 19 Fine, Kit, 138n.43, 144n.58
Broad, C. D., 72–75 Fodor, Jerry, 5, 125–26, 134n.32, 158n.28,
Brown, Robin, 6n.4, n.5 159n.31
Burge, Tyler, 140, 156n.24, 172 French, Steven, 117
Byrne, Alex , 13n.11, 16, 19n.24 Funkhouser, Eric, 127n.11
Carnap, Rudolf, 31n.8, 117–21, 124
Chalmers, David, 4, 6–8, 10–12, 16n.16, 20, Galileo Galilei, 34, 45, 58–59
24n.35, 26–31, 36–40, 43, 45, 47–65, 85–91, Gertler, Brie, 48–49n.4, 74–75
94, 100n.37, 108–117, 118n.41, 170 Gillett, Carl, 135–36, 139, 145n.63, 163n.41

193
 194 Index of Names

Harré, Rom, 99n.35, 128n.15 Locke, John, 13, 15–18, 31, 49, 85, 97–100, 113,
Hawthorne, John, 12n.10, 15, 66–69, 83, 90n.12, 137n.41
105 Loewer, Barry, 142n.54
Heil, John, 87, 98n.34, 128n.15, 135n.33 Lycan, William G., 11n.7, 12
Hilbert, David, 16
Hill, Christopher, 20n.26, 22–23, 83–84 Madden, E. H., 99n.35, 128n.15
Holden, Thomas, 90n.13, 98, 99n.36, 100n.40, Makkreel, Rudolf, 123n.1
101n.43 Martin, Charlie B, 87
Horgan, Terence, 11n.7, 130n.23 Matthews, Gareth, 131n.26
Howell, Robert, 121n.48 Maxwell, Grover, 104n.5
Humberstone, Lloyd, 92n.15 Maund, Barry, 16n.18
Hume, David, 24n.34, 40n.24, 90, 98–99, 107, McKitrick, Jennifer, 98n.32, 109n.19, 117
119–20, 128 McGinn, Colin, 7–8, 30–36, 105n.11, 110, 116–17,
136n.38
Jackson, Frank, 3–4, 9–46, 48, 48–49n.4, McGonigal, Andrew, 93n.22
50n.10, 52–60, 87, 131n.24 McLaughlin, Brian, 20n.26, 126n.8, 145n.63
Jacovides, Michael, 16n.17 Mellor, Hugh, 6n.3
Jaworski, William, 158n.28 Melnyk, Andrew, 136–39, 149
Johnston, Mark, 17n.20, 132n.28 Moyer, Mark, 20n.30, 131, 132n.28
Mucciolo, Laurence, 164n.42
Kant, Immanuel, 3–9, 28, 46, 86, 89, 92n.16, Mundale, Jennifer, 164
93–94, 97, 99n.36, 100–104, 106–109, 112, 114,
116–17, 120n.45, 121, 123–24, 172 Nagel, Ernest, 146
Kim, Jaegwon, 92n.15, 94n.23, 126–61, 164n.42, Nagel, Thomas, 7–8, 10n.2, 83, 110, 114n.30, 116,
168 120, 121n.48
Kitcher, Philip S., 144n.59, 154n.18 Newman, M. H. A., 119n.43, 120n.45
Kobes, Bernard W., 172n.2 Newton, Isaac, 85, 97–100, 113, 137n.41,
Kornblith, Hilary, ix , 126n.7, 131n.27, 135n.34, Ney, Alyssa, 93n.19, 99n.36, 105n.6, 107, 117n.37,
158n.28, 161, 164n.44, 167–68 157–58
Kriegel, Uriah, 19n.25 Nida-Rümelin, Martine, 10n.2
Kripke, Saul, 17, 38, 39n.21, 47–48, 51, 55, Noonan, Harold, 132n.28
66–67, 75–84, 131n.27
O’Connor, Timothy, 145n.63, 146–47
Ladyman, James, 6n.4, n.5, 104n.5, 109n.19, 117,
119n.43 Papineau, David, 20–21, 30n.3, 172
Langton, Rae, 92n.15, 99n.36, 103–106, 107n.12 Pargetter, Robert, 87n.5
Leibniz, Gottfried Wilhelm, 7, 9n.1, 85–86, Parkinson, G. H. R., 9n.1
90–97, 101, 102, 108–109, 114, 117–18, Paul, L. A., 127, 133n.30, 157–58
120n.45 Pereboom, Derk, ix , 26n.38, 27n.40, 73n.17,
Levine, Joseph, 13n.11, 24n.35, 29, 43–46, 92n.16, 100n.38, 103n.1, 107n.12, 126n.7,
160n.35 131n.27, 138n.42, 141n.50, 151n.10, 152n.13,
Lévi-Strauss, Claude, 120n.45 156n.24, n.25, 164n.44, 167n.51
Lewis, David, 7, 17n.20, 64, 86, 92n.15, 104–105, Perry, John, 12n.10
108–109, 117 Pettit, Philip, 131n.24
Loar, Brian, 11n.7, 24n.35, 30, 43 Place, U. T., 148
Locke, Dustin, 104n.4, 108 Polger, Thomas, 151n.9, 154n.19, 155n.22
 195 Index of Names

Prior, Elizabeth, 87n.5 Stoljar, Daniel, 6n.5, 11n.5, 24n.35, 44n.32, 66,
Putnam, Hilary, 5, 29–31, 124–25, 129, 151–53, 69–75, 81, 83–84, 87n.2, n.5, 89–91, 110–14
154n.18 Strawson, Galen, 45n.35, 91n.14, 108–10

Ramsey, Frank P., 104–105, 119 Thompson, Brad, 12n.8

Rickert, Heinrich, 123 Thomson, Judith, 138–39n.44
Robb, David, 127n.13, 128n.15 Tye, Michael, 11n.7
Rosenberg, Gregg , 87n.2
Ross, Don, 117 van Cleve, James, 93, 94–95n.23, 103–104
Rozemond, Marleen, 47n.1 Van Gulick, Robert, 11n.7, 153n.17
Russell, Bertrand, 4, 30, 53–55, 57, 63, 85–122, van Inwagen, Peter, 93n.22, 137n.41
127, 168 Velleman, David, 16n.18
Rueger, Alexander, 145n.63
Warfield, Ted, 142n.54
Saucedo, Raul, 139n.45 Wasserman, Ryan, 132n.28, 138n.43, 138–
Shagrir, Oron, 164 39n.44
Schaffer, Jonathan, 96n.26, 104n.4, 105, Weatherson, Brian, 20n.30, 92n.15
138n.43, 145n.63 Williamson, Timothy, 15
Schroeter, Laura, 61n.37 Wilson, Margaret, 47n.1
Shapiro, Lawrence, 162–63 Wilson, Jessica, 6–7, 39n.21, 132–35, 136n.38,
Shoemaker, Sydney, 19n.24, 28n.41, 29, 40–43, 137n.39, 142–43
82n.28, 99, 105, 108, 128, 130–36, 142, 149 Windelband, Wilhelm, 123
Sider, Theodore, 137, 142n.52 Witmer, Gene, 143n.57
Silins, Nico, 9n.1, 25n.36, 28n.41, 68n.7, 114n.31 Wykstra, Stephen, 24n.34
Smart, J. J. C., 148
Smith, Michael, 87n.5 Yablo, Stephen, 32n.10, 47n.1, 50n.11, 51–52, 66,
Smullyan, Raymond, 21 75–84, 127, 129n.18, n.19, 131n.27, 154–55
Sober, Elliott, 129n.18
Stalnaker, Robert, 12n.10, 31n.8, 33n.13, Zimmerman, Dean W., 128n.15, 137n.41,
48–49n.4, 61n.38, 66–69, 83 138–39