The Nature and Metaphysical Significance of the Explanatory Gap

Material Information

The Nature and Metaphysical Significance of the Explanatory Gap
Andrei, Ana
University of Florida
Publication Date:

Thesis/Dissertation Information

Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Committee Chair:
Ludwig, Kirk A.
Committee Members:
Witmer, D Eugene
D'amico, Robert
Ray, Greg B.
Smocovitis, Vassiliki
Graduation Date:


Subjects / Keywords:
Ascriptivism ( jstor )
Consciousness ( jstor )
Intuition ( jstor )
Metaphysics ( jstor )
Phenomena ( jstor )
Physicalism ( jstor )
Physics ( jstor )
Property law ( jstor )
Psychophysics ( jstor )
Sturgeon ( jstor )
chalmers, consciousness, explanatory, gap, levine, physicalism
Unknown ( sobekcm )


General Note:
THE NATURE AND METAPHYSICAL SIGNIFICANCE OF THE EXPLANATORY GAP The main goal of the present study is to answer the following two questions: First, what is the exact nature of the explanatory gap? Second, what are the metaphysical implications of the gap phenomenon? In Chapters 1-4, I aim to reach a deeper understanding of the nature of the explanatory gap phenomenon. To this end, I provide a detailed critical examination of a number of approaches to the gap that are either suggested by the literature on the topic or inspired by some philosophers? discussion of related themes, and argue that in fact they mischaracterize in various ways the difficulty they purport to represent. I also sketch my own construal of the explanatory gap. In Chapter 5, I investigate the metaphysical consequences of the gap, and argue that the gap phenomenon casts doubt upon both a posteriori physicalism and neutral monism, lending a significant degree of plausibility to the view that the phenomenal is ontologically fundamental.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Andrei, Ana. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
Resource Identifier:
814306384 ( OCLC )


This item has the following downloads:

Full Text




2010 Ana Maria Andrei 2


To m y parents 3


ACKNOWLEDGMENTS I wish to thank the members of my comm ittee for invaluable feedback on previous versions of the dissertation. I am greatly inde bted to Kirk Ludwig for abundant comments and illuminating discussion. I owe a special debt of gratitude to Gene Witmer, who has offered tremendous help in structuring th e project, as well as helpful co mments on various drafts. I wish to thank Robert DAmico, Franz Futterknecht and Eric Kliegerman, with whom I worked mostly on topics unrelated to my dissert ation, but who have inspired me to be bolder in my intellectual aspirations. Thanks are also due to Bill Butcha rd, Christopher Lubbers an d Elka Shortsleeve for useful comments and unconditional encouragement. Last but not least, I wish to thank Emil Badici, my husband, for insightful sugges tions and for his boundless generosity. 4


TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........7 LIST OF FIGURES.........................................................................................................................8 ABSTRACT.....................................................................................................................................9 CHAPTER 1 PRELIMINARIES................................................................................................................ ..10 The Explanatory Gap vs. the Gap Intuition in its Various Guises..........................................10 Some Preliminary Desiderata on an Account of the Explanatory Gap..................................14 Ascriptivism vs. Non-Ascriptivism........................................................................................16 Overview of the Project..........................................................................................................17 2 A DERIVABILITY CONSTRUAL OF THE EXPLANATORY GAP (D. CHALMERS).........................................................................................................................21 A Simple Picture of the Explanatory Gap..............................................................................21 A Complication: Making Sense of Reductive Explanation Modulo Phenomenology........22 3 A GENERALIZED DERIVABILITY CONCEPTION OF THE GAP.................................33 Physics-Based Accounts of the P hysical and Hempels dilemma..........................................33 The Lack of Generality Objection against Chalmers.............................................................45 Delimiting the Physical........................................................................................................ ...59 Defense of the Proposed Account of the Physical from Objections.......................................67 A Generalized Derivability Construal of th e Gap and an Account of the Gap Intuition........75 4 EXPLANATORY POWER CONS TRUALS OF THE GAP.................................................82 Two Accounts of the Gap Inspired by J. Levine: NCC and TC.............................................82 A Closer Look at the Two Accounts and their Shortcomings................................................86 The Epistemic Credentials of Conceptual Non-Ascriptivism................................................93 5 THE METAPHYSICAL IMPLICATI ONS OF THE EXPLANATORY GAP...................117 From an Explanatory Gap to an Ontological Gap................................................................117 Denying the Epistemic Asymmetry......................................................................................122 The Epistemic View of Subjectivity (Sturgeon)...................................................................127 Different Kinds of Imagining/C onceiving (Hill and McLaughlin)......................................140 5


Phenom enal Concepts as Recognitional (Loar)....................................................................148 The Prospects for A Posteriori Physicalism and Neutral Monism.......................................153 LIST OF REFERENCES.............................................................................................................161 BIOGRAPHICAL SKETCH.......................................................................................................166 6


LIST OF TABLES Table page 5-1 Types of belief justification.............................................................................................129 7


8 LIST OF FIGURES Figure page 5-1 Generative property explanation......................................................................................130


Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE NATURE AND METAPHYSICAL SIGN IFICANCE OF THE EXPLANATORY GAP By Ana Maria Andrei August 2010 Chair: Kirk Ludwig Major: Philosophy The main goal of the present study is to answer the following two questions: First, what is the exact nature of the explanatory gap? Second, what are the metaphysical implications of the gap phenomenon? In Chapters 1-4, I aim to reach a deeper unde rstanding of the nature of the explanatory gap phenomenon. To this end, I provide a detail ed critical examination of a number of approaches to the gap that are either suggested by the literature on the to pic or inspired by some philosophers discussion of related themes, and argue that in f act they mischaracterize in various ways the difficulty they purport to represent. I also sketch my own construal of the explanatory gap. In Chapter 5, I investigate th e metaphysical consequences of the gap, and argue that the gap phenomenon casts doubt upon both a posteriori physicalism and neutral monism, lending a significant degree of plausibility to the view that the phenomen al is ontologically fundamental. 9


CHAP TER 1 PRELIMINARIES In Chapter 1, I draw a distinction between an explanatory gap br ought on by consciousness and a gap intuition that consciousness elicits, and whose various m anifestations I lay out. I subsequently formulate a number of preliminary adequacy criteria that a construal of the explanatory gap should meet, and provide some background on certain views about the semantics of linguistic expressions and about co nceptual content (labeled ascriptivism and non-ascriptivism) which are highly relevant both to the project of elucidating the nature of the explanatory gap and to that of clarifying th e metaphysical significance of the gap phenomenon. I end by providing an overview of the project. The Explanatory Gap vs. the Gap In tuition in its Various Guises It is useful to distinguish between something we might call the gap intuition and the explanatory gap itself. To get a grip on what I mean by the gap intuition, let us consider the kind of puzzlement that a claim like pain is the same thing as the firing of C-fibers tends to elicit. (I am employing here the phenomenal concept of pain. Arguably, our ordinary concept of pain involves in addition to the purely phenom enal constituent a further functional component which, for the purposes of this discussion, shoul d be set aside.) We do not quite seem to understand how that could be true. We have the sense that th is is not just a matter of lack of factual information, as we would expect a claim like pain is C-fiber firing to remain equally puzzling no matter how much we found out about our physical constitution. We might even be tempted to go so far as to suggests that pain is the firing of C-fibers sounds like a category mistake. More generally, we seem to have troubl e making sense of the suggestion that the phenomenal is nothing over and above the phys ical. To appeal to a useful metaphor, if 10


physicalism about the phenomenal is true, when God brought about all the physical facts and laws, he thereby brought about the phenomenal as well; he did not have to do any extra work to ensure that the phenomenal was in placethe phenomenal just came for free. But this claim sounds puzzling. How could the phenomenal have just come for free, especially since it does not seem to be conceptually necessitated by the physical? Another way of expressing what is at bo ttom the same kind of puzzlement would be: How could there be pain in a wholly physical world? The intuition this question expresses is that if we assumed that everything wa s physical, we would be unable to explain that there were such things as pains, tickles, color sensations, etc. The gap intuition is a label I would like to us e for what appears to be a cluster of closely connected puzzlements. The following four families of questions are manifestations of the gap intuition in its various guises1: 1. Questioning the intelligibility of identifyi ng phenomenal states with physical states. How could pain be the firing of C-fibers? How could pain a nd the firing of C-fibers pick out the same thing? 2. Questioning the intelligibility of the phenomenal being meta physically supervenient on the physical. How could the phenomenal fail to be so mething over and above the physical? How could the phenomenal be necessitated by the physical in a sense stronger than nomic, given that it is not conceptually necessitated by it? 3. Questioning the possibility of accounting for the existence of phenomena l states in purely physical terms. How could we make sense of the exis tence of pain in a wholly physical world? 4. Questioning the intelligibility of physicalism in general. How could physicalism be true? 1 Subsequent discussion (see especially Chapter 3) will in fact reveal that it is preferable to characterize the gap intuition in a slightly more general fashion, as it consists in a family of puzzlements that arise independently of whether the fundamental terms of physics, chemistry, biology really turn out to refer to physical properties. Something similar holds for the explanatory gap. 11


It is not only newcom ers to th e philosophy of mind who find the gap intuition quite powerful. Physicalists themselves, when not blinded by thei r philosophical allegiances, do admit that they incur an obligation to expl ain how this feeling arises compatibly with their ontolog ical position. Interestingly, philosophers typically come to uphold physicalism for rather indirect reasons. If one finds irresistible both the thesis that the physical is causally closed and the view that the mental is causally efficacious, the conc lusion that physicalism is true looks compelling. There is indeed the rather vexing feeling that it is unclear how it could be true, but the assumption is that, since it must be true, one should just summ on all ones ingenuity in an attempt to show that the nagging feeling can be explained away. To briefly anticipate the conclusion of the present inquiry, this inge nuity will prove worthy of a better cause. When philosophers claim that the phenomenal generates an explanatory gap (or that there is an explanatory gap between the phenomenal a nd the physical), what they often mean is that the phenomenal is not susceptible of a sort of explanation in physical terms which one would expect it to be amenable to if it were phys ical, and which seems to be available for uncontroversially physical phenomena. While this use of the phrase explanatory gap may not be unanimous, it is the one I would like to focus on. Accordingly, the existence of an explanatory gap in this sense counts as at least pr ima facie evidence agai nst physicalism. I think an important connec tion holds between the explanatory gap and the gap intuition. It is natural and, I think, legitimate to regard the gap intuition as gr ounded in certain special features of phenomenal concepts that are also re sponsible for the principled unavailability of an explanation of the phenomenal in physical terms of the appropriate sort. Indeed, in different contexts philosophers focus in th eir characterization of the explanatory gap on different things as the explananda that the physicalist appears to ha ve trouble accounting for. Those can be identity 12


statem ents2 (e.g., pain is the firing of C-fibers), nece ssity statements (e.g., It is metaphysically necessary that, if such and such physical facts are in place, su ch and such a phenomenal fact obtains), and statements to the effect that pa rticulars instantiate such and such phenomenal properties (e.g., S is in pain at t ). This list perfectly matches that of claims susceptible to elicit some form or other of the gap intuition. We cannot help but wonder how there could be pain in a wholly physical world. If pain picks out a physical property, it seems as though it cannot be a basic physical property, for to maintain the contrary would sound like trying to secure the truth of physicalism by arbitrarily redefining the predicate physical. Consequently, an explanation of pain in terms of more basic physical properties must be available, namely, we would expect an explanation akin in form to the explanation that properties picked out by terms like water or temperature are amenable to. The difficulty for physicalism is that such an explanation seems unavailable in principle. Similarly, psychophysical identity statements generate a sense of puzzlement that ordinary theoretical identifications like water is H2O do not. It is quite tempting to take this difference in intuitive reactionsas philosophers who write on this to pic often doas signaling something about phenomenal concepts that is bo und up with an important difference in how the two kinds of identificat ion are justified. The view that the explanat ory gap is closely bound up with a type of puzzlement that certain claims about phenomenal states tend to elicit is sugg ested by the rather unsystematic reflections on the explanatory gap due to Levine who also coined the term explanatory gap. What Levine focuses mostly on as a source of puzzlement are psychophysical identity 2 Its quite common in discussions of the mind-body problem for statements like water is H2 O or pain is the firing of C-fibers to be called identity statements. I will go along with this terminology, as correcting the underlying semantic analysis of such statements is unlikely to influence the discussion in any crucial way. 13


statem ents, called by him gappy identities, which he further spells out as i dentity claim[s] that admit of an intelligible request for explanation.3 While his characterization of the sense in which psychophysical identities are gappy seems incorrect (any informative identity statement admits of an intelligible request for explanation), the type of puzzlement he draws attention to is worth investigating in some detail. Some Preliminary Desiderata on an Account of the Explanatory Gap Giving an account of the e xplanatory gap involves answeri ng the following questions: 1. If the phenomenal were physical, what t ype of explanation of the phenomenal in physical terms should be available, exactly? 2. What are the features of phenomenal concepts that make it impossible to give an explanation of the phenom enal of the specified kind? It is natural to expect that we will get an answer to the first question by examining the unproblematic cases of each of the three kinds of statements (i.e., identity statements, necessity statements and statements about particulars) and offering a story about how their truth is explained, since if the phenomenal were physical it should presumably be susceptible of the same kind of explanation as ordinary macrophysic al phenomena like water or temperature. In fact, to anticipate a later result, the discovery that the phenomenal is conceptually constitutive of macrophysical phenomena such as water or te mperature will motivate reconsidering the intuitively appealing thesis that the explanatory gap brought about the phenomenal simply consists in the lack of an explanation of the phenomenal in physical terms of the sort that is available for water. 3 (Levine, 2001, p. 84). 14


Another important question that has to be answered concerns the exact nature of the relationship between the explanator y gap and what I called the ga p intuition, both of which we would expect to be grounded in features of phenomenal concepts. From a metaphysical perspective, the cruc ial question (which is a ddressed in Chapter 5) is whether the special features of phenomenal c oncepts that are responsible for the explanatory gap and the gap intuition warrant the conclusion th at phenomenal properties are not physical, and perhaps, more generally, that they are ontologically fundamental. There are physicalists w ho do not regard the gap intuition as bound up with an explanatory asymmetry, because they in effect deny that the mental generates any explanatory gap. I do not have in mind here philosophers committed to type A physicalism (i.e., the view that the mental supervenes on the physical logically ), but rather certain espousers of type B physicalism (i.e., the tenet that the mental does not logically supervene on the physical, but counts as material nonetheless because it is metaphys ically necessitated by the physical), such as Papineau, and Block and Stalnaker.4 Such philosophers have a thr eefold obligation to discharge: first, they must provide a convincing account of how familiar theoretical identifications are justified; second, they must show that this accoun t fits psychophysical identities too; third, they must supply a convincing explan ation of the gap intuition. Th is position is examined and assessed in Chapter 5. As for type A physicalism, I will work under the assumption that it is false. The task of proving type A physicalism wr ong lies outside the scope of this dissertation project. 4 I borrow the terms type A physicalism and type B physicalism from (Chalmers, 1996). 15


Ascriptivism vs. Non-Ascriptivism An important distinction that informs bot h debates over the nature of the gap and physicalist responses to the explanat ory gap problem is that between ascriptivism and nonascriptivism. The two labels were introduced in Levine (2001), and the way he characterizes these views suggests that they are views that are supposed to apply to both linguistic expressions and concepts. The contrast between them regards wh at is deemed to fix the extension of a use of a predicate in a language, a nd, correspondingly, the extension of a concept. Linguistic ascriptivism entails that what is responsible for fi xing the extension of a term as used in a certain context is a mode of presenta tion awareness of which is guaranteed by semantic competence. The mode of presentation is thus supposed to be the same across the entire community of competent speakers. The corresponding story abou t concepts is that what determines the extension of a concept is a mode of presentation that is cogniti vely accessible to the subject merely in virtue of the subject grasping the concept in question. Linguistic non-ascriptivism is the view that what determines the extension of expressions in the language is not a mode of presentation awareness of which is guaranteed by semantic competence, but rather a mode of presentation that, as Levine put s it, works behind the scenes, i.e., one that is merely a matter of certain ex ternal (typically causal) relations holding between the use of the term and certain kinds of things in the world. N on-ascriptivism about conceptual content is the correspondi ng view that what fixes the extensio n of a concept is not a mode of presentation knowledge of which is afforded simply by grasping the concept, but is rather just a matter of certain brute causal relations obtaining. One could think of non-ascriptivism as comi ng in two strengths: The radical variety entails that grasp of a concept/semantic competence in the use of a term does not afford any a priori knowledge whatsoever abou t its extension. The moderate va riety does allow that grasping 16


a concept/sem antic competence may furnish a cert ain amount of such a priori knowledge, but it entails that the content that may be built into a concept/the descriptive content that may be part and parcel of the meaning of a certain term is insufficient to fix the extension of the term/concept. Another important dimension of variation has to do with the category of terms/concepts that are claimed to be non-ascriptive in one of the two senses delineated above. Some philosophers endorse non-ascriptivism about macr ophysical terms/concepts, others suggest it applies only to phenomenal terms/concepts, a nd yet others uphold nonascriptivism about both physical and phenomenal terms/concepts. I think the debates over the mind-body problem often suffer from a failure to distinguish between linguistic ascriptivism/non-ascriptivism and ascriptivism/nonascriptivism as theories of conceptual content, which are erroneously pr esented as a package deal. Drawing such a distinction will prove important for a couple of reasons. First, since non-ascriptivism is often invoked by physicalists concerned to show that the explanatory gap poses no insurmountable difficulties for their metaphysical position, it will be useful to clarify which brand of nonascriptivism the physicalist would have to embrace if she is to properly engage the explanatory gap argument advanced against physicalism (namel y, conceptual, rather than linguistic). Second, bearing in mind the distinction between linguistic and conceptual non-ascriptivism will enable one to see that some arguments typically assumed to support the bra nd of non-ascriptivism needed to resist the explanatory gap argument (i.e., conceptual) in fact on ly warrant a version of linguistic non-ascriptivism. Overview of the Project Construals of the explanatory gap can be clas s ified into two categoriesderivability and explanatory power construalsusin g as a criterion the underlying c onception of how the truth of 17


ordinary theoretical identif ications (e.g., water is H2O), in particular, gets explained, under the assumption that the relevant macroscopic concepts (e.g., water ) have no phenomenal constituents. According to derivability construals of the gap, ordinary theoreti cal identifications are justified by virtue of being derivable from a set of basic truths th at do not involve identities (at least if we assume that the phenomenal is not conceptually constitutive of the reduced phenomenon.) According to the form of this appro ach to the gap that I will focus on in Chapter 2, identity statements that the phenomenal is not conceptually constitutive of are justified because they are entailed by a full description of the microphysical, while, by contrast, the phenomenal generates an explanatory gap because psychophysical identity statements are not so entailed. Chalmers can be interpreted as main taining a view of the gap of this sort. The family of explanatory power construals of the explanatory gap have in common the tenet that ordinary theoretical identifications are justified by their explanatory potentialwhere that gets cashed out in a special waywhile a justification of this sort is unavailable for psychophysical identity statements. According to th is type of construal, the justification of statements like water is H2O or temperature is mean kineti c energy does not consist in their derivability from the microphysical, for they are not so derivable ev en if we idealize away from the fact that macrophysical con cepts may have phenomenal components, but rather in their enabling the explanation of the macrophysical pr operties of phenomena like water, temperature, etc. The explanatory potential of such theoretical identifications is manifested in a priori conditionals from a full description of the microphys ical plus the theoreti cal identifications in question to statements about the macrophysical properties of water, temperature, etc. 18


Chapter 2 of the present study is devoted to reconstructing Chal merss view of the explanatory gap and to clarifying how he unders tands the metaphysical significance of the gap phenomenon. In Chapter 3, I first investigate the concept of the physical by reference to which the explanatory gap should be articulated, and I offer an account of this notion, which I compare with the major proposals in the literature on demarcating the physical and show to be superior. In the course of elucidating the notion of the physical relevant to the gap prob lem, I also show that Chalmerss construal of the gap relies on a cer tain assumption about the physical which is, if true, only contingently so, and as a result unhappily entangles questions about the nature and metaphysical import of the gap with the separate issue of whether in the actual world there are any emergent physical properties. The second a im of Chapter 3 is to provide a generalized derivability construal of the gap, which disso ciates the phenomenon from such problematic assumptions and characterizes it in more fundament al terms. The third object ive of the chapter is to examine the relationship between the explanat ory gap and the gap intu ition and to offer an explanation of the latter. In Chapter 4, I present and assess two construa ls of the gap that be long to the explanatory power family and which are inspired by two different stages in Levines reflections on the gap problem. I distinguish between an early no-ca usal-role-concept account of the gap (NCC), and a more recent thick-concept account (TC). While both Chalmers derivability account of the gap and the Levine inspired NCC rest on ascriptivism about both phenomenal and nonphenomenal concepts, TC relies on non-ascr iptivism about macrophysical concepts. A central objection I level against NCC is that, even if we grant the (highly objectionable) underlying conception of scientific reduction (more specifically, the view of the justification of theoretical identifications that NCC presuppos es), NCC is a highly unstable 19


position on the nature of the gap. TC, in turn, rests on non-ascriptivism about the content of macrophysical concepts, and my main objection ag ainst TC is that commitment to conceptual non-ascriptivism is, as I argue, deeply problematic. In Chapter 5, I generalize Chalmerss formulat ion of the argument from the explanatory gap to the falsity of physicalism in light of my own derivability construal of the explanatory gap (presented and defended in Chap ter 3). I subsequently offer a critical ex amination of some central physicalist responses to th e gap argument. The strategies adopted by physicalists are of two sorts. Some reject the picture of scien tific reduction presupposed by the explanatory gap argument, and maintain that identity statements which the phenomenal is not conceptually constitutive of, on the one hand, and psychophysical identity statements, on the other, are in fact epistemically on a par. I argue that this strategy misconstrues scientific reduction and fails to explain the gap intuition. The second approach consists in granting the existence of an explanatory asymmetry between the two sorts of theoretical identifications, and claiming that the asymmetry can be accounted for in terms of sp ecial features of phenomenal concepts which physicalism is able to accommodate. Most versions of what is labeled in the literature the phenomenal concepts strategy belong here. I critical ly discuss major versions of this approach defended by S. Sturgeon, C. Hill, B. McLaughlin and B. Loar. I conclude Chapter 5 with a general argument to show that the explanat ory gap problem strongly suggests that both a posteriori physicalism and neutral monism are d eeply inadequate views about the nature of consciousness, and that we should think of the ph enomenal instead as ontologically fundamental. 20


21 CHAPTER 2 A DERIVABILITY CONSTRUAL OF THE EXPLANATORY GAP (D. CHALME RS) The main goal of the present chapter is two-fold: on the one hand, to offer a reconstruction of Chalmerss c onstrual of the explanatory gap,1 and, on the other hand, to spell out the argument he puts forth from the existence of the gap to the falsity of physicalism. In the first section, I present a simple view of the gap that a cursor y look at Chalmerss texts might encourage the reader to attribute him. In the se cond section, I explain how the simple view needs to be refined if it is to reflect Chalmerss actual commitments, and I also clarify how his argument goes which is intended to bring out th e metaphysical significance of the explanatory gap. A Simple Picture of the Explanatory Gap One m ight be inclined to thi nk that Chalmers subscribes to the following view of the explanatory gap and its metaphysical implicatio ns: Reductive explanation requires a priori deducibility. A reductive explanation of the m acrophysical is possible because the macrophysical is a priori deducible from a full description of the microphysical couched in the vocabulary of a completed microphysics (P). In fact, to be precise, the derivation base should be represented as consisting in the conjunction of P and an indexical as well as a totality claimI and T, respectively. The former claim specifies the thinker/speaker and the present time and is introduced to allow the derivation of context sensitive statements. The latter is a cl aim to the effect that the world is a minimal world that satisfies both P and I, and the motiv ation for it is to ensu re that th e account of reductive explanation in terms of entailment by the microphysical is not falsified by negative 1 My reconstruction is based on Chalmerss remarks about the gap, as well as about closely related phenomena, such as the knowledge and the zombie arguments, from (C halmers, 1996), (Chalmers, 2001), (Chalmers, 2002), (Chalmers, 2005), (Chalmers, 2007) and (Chalmers, 2009).


existential s tatements or universally quantifie d statements. For simplicity, I will omit in the following discussion T and I, but it should be ke pt in mind that full accuracy requires taking them into account. The phenomenal is not a priori deducible from P, and hence no reductive explanation of it in terms of the microphysical is possible. What cannot be reductively explained in microphysical terms in this fashion is not physical. Consequently, phenomenal properties are not physical properties. Thus, one might suggest, we have an explanation of why theo retical identifications like water is H2O are not puzzling in the way that pain is the firing of C-fibe rs is: water is H2O is deducible from P, while pain is the firing of C-fi bers clearly is not the so rt of thing that could be entailed by P. A closer look at Chalmers papers reveals that what he is up to is sl ightly different, as we will see in the next section. The problem with th e view of the gap under scrutiny is that it is doubtful that the ordinary, completely unproblem atic, theoretical identifications that are contrasted with psychophysical ones are really deducible from P, because many natural kind concepts involve a phenomenal component. If psychophysical identity statements are not deducible from P, then the same holds, for example, for heat is molecular motion, or even water is H2O, as the concepts of heat and water cl early have phenomenal constituents. Indeed, in Conceptual Analysis and Reductive Explana tion Chalmers is prepared to extend such considerations to many or most macr oscopic concepts (Chalmers, 2001, p. 18). A Complication: Making Sense of Reduc tive Explanation Modulo Phenomenology Chalm ers does uphold the view that reduc tive explanation requ ires entailment. Accordingly, to accommodate the lack of entailmen t from P to macrophysical truths like heat is the motion of molecules (under the assumption that th e phenomenal is not a priori derivable 22


from P), he qualifies the claim that a fullblown reductive explanation of macrophysical phenomena is available: he states that, if the phenomenal is not a pr iori deducible from P, then macroscopic truths are reductively explaina ble modulo phenomenology that is that we can reductively explain those aspects of them for which the phenomenal plays no conceptually constitutive role. This would f it nicely with the view () that the actual reductive explanation of phenomena such as color and heat has proceeded by explaining their objective aspects while leaving their subjective aspe cts untouched. (Chalmers, 2001, p. 3) Admitting that there is a breach of this sort in the justification of st atements like heat is molecular motion raises a number of important questions. First, how should we think of the explanatory gap the phenomenal generates, exactly? If what it ultimately comes down to is a lack of derivability from P, then the gap brought about by consciousness extends to all phenomena that consciousness is conceptually constitutive of. Second, what is the form of the argument from an explanatory gap to an ontologi cal gap? It is plain that the argument cannot simply go like this: Reductive explanation requires a pr iori derivability, and hence a ny truth that is not derivable from P picks out a non-physical fact; in particular the phenomenal is not a priori derivable from P, therefore, the phenomenal is not physical. Afte r all, that heat is molecular motion is not a priori derivable from P either, and yet we do not conclude that h eat is a non-physical phenomenon. Before addressing these questions, it will be help ful to get clear on what it is for ordinary macrophysical phenomena like temperature to be reductively expl ainable only modulo phenomenology. In what follows, I will prov ide a model for how to think of this. If N is a term that picks out a phenomenon we are trying to give a reductive explanation of, then a reduction is available only if there is a definite description, the M, in an ideal language L that contains terms for all the basic concepts we possess, such that N is the M is 23


knowable to be true a priori.2 The trouble is that among th e terms in the description the M, we might encounter terms that express phenomenal concepts. One might propose then that the claim that the phenomenon N picks out can be reductively explained modulo phenomenology comes down to the claim that If P, then the R is the U is knowable a priori, where the R is a definite description obtained from the M by removing all the clauses that contain terms which express phenomenal concepts, while the U is a term from the vocabulary of physics, chemistry, etc. that reveals what th e reducing phenomenon is. A problem this proposal faces is that in some cases it is not clear that, after the removal of the components that the phenomenal is concep tually constitutive of, enough remains to really fix a referent. Perhaps the phenom enal does not play a crucial ro le as far as concepts like water are concerned, but for concepts like heat or, better yet, red (understood as applied to external objects), if one brackets, as sugge sted, the phenomenal component, it is not quite clear what is left. And even if what is left does impose in many cases substantive constraints on what could count as the phenomenon picked out by the term, it seems possible for there to be instances in which more than one physical phenomenon plays the causal role obtained by purging the M of phenomenal elements. In that case, If P, then the R is the U comes out false. In the face of such objections, one could modify the foregoing account of reduction modulo phenomenology as follows: a reductive e xplanation modulo phenomenology is available for a phenomenon N just in case, for every predicate S in L which is such that If there is a unique M, the M is S is a priori, and such that no phenom enal concept is constitutive of the 2 This presupposes that N behaves as a referring term; for the sake of simplicity, I will not attempt here to modify the story so as to avoid this questionable commitment, but I trust such a modification is possible and wouldnt affect the following argument. 24


concept expressed by it, it is derivable from P that S is satisfied precisely by the objects that satisfy it. While the modified account correctly predicts that reduction may still be possible even if more that one thing satisfies R the requirement on reduction it la ys down is trivially satisfied if nothing remains after removing from the M all the clauses that contain phenomenal terms (as might be the case for red understood as a predicate of external things). That is an awkward result. The problem with the modified account is that it provides an incomplete picture of reductive explanation modulo phenomenology. It is co rrect that there is a reductive explanation modulo phenomenology of a microphysic al phenomenon such as water only if P implies, for every feature included in the concept of wate r and which the phenomenal is not conceptually constitutive of, that feature is in stantiated precisely when and wher e it is instantiated. That is not the whole story, however. If th e causal role expressed by th e term which picks out the explanandum property specifies re lationships to phenomenal featur es, then reductive explanation modulo phenomenology succeeds only if, in addition, all the instantiations of the explanandum property (as expressed by the term in question) are derivable from P conjoined with a set of psychophysical correlations that capture the causal relationshi ps (if any) between those phenomenal features and the reducing prope rty. For the purposes of reduction modulo phenomenology, we just take these causal relationships for granted, and set aside the question whether they are derivative or basic, as that hinges on whether consciousness itself is amenable to reduction or not. It is this further deducibility requirement that helps account for the possibility of reductive explanation modulo pheno menology for properties that the phenomenal is conceptually 25


constitutive of to the degree th at removing all ph enomenal concepts leaves either no descriptive material or one that is multiply satisfied. What is then the exact sense in which th e phenomenal generates an explanatory gap on Chalmerss view? Chalmers maintains that PQTI3 entails any macroscopic truth, and that there are two possibilities: either PTI alone implies any macroscopic truth, in which case PTI must also imply Q, or PTI alone does not imply all macroscopic truths, in which case all failures of PTI to imply a relevant [macr oscopic truth] M will be associated with the failure of PTI to imply Q, in the sense that adding Q will close any epistemic gaps. Putting these cases together, the thesis concerning PQTI [to the effect that PQTI implies any macroscopic truth M] entails the crucial claim that if phenomenal truths are not implied by PTI, then there is a special epistemic gap in the phenomenal case. (Chalmers, 2001, p. 3) Here is how that relates to th e notion of reductive explanation: If we combine these alternatives with the thes is that reductive expl anation goes along with a priori entailment, then the first alternative above leads to the view that all the relevant macroscopic truths, including phenomenal trut hs are reductively explainable. The second alternative leads to the view that phenomenal truths arent reductively explainable, and that other macroscopic truths are only reduc tively explainable modulo phenomenologythat is, that we can reductively explain those aspe cts of them for which the phenomenal plays no conceptually constitutive role. (Chalmers, 2001, p. 3) Of course, Chalmers contends that it is the latter alternative that obtains. Consequently, the underlying picture of the e xplanatory gap seems to be the following. Full blown reductive explanation does require de rivability from the microphysical. The reductive explanation modulo phenomenology of macrosc opic phenomena like water or temperature amounts to reductive explanation proper only if th e phenomenal itself is reductively explainable, i.e., deducible from the microphysical. However, the phenomenal is not so deducible. Here is how Chalmers appears to think of the explanatory gap in relation to the th ree kinds of statements 3 Q is a full description of the phenomenal, while, as explained above, T is a totality claim, and I provides indexical information about the speaker and the present time. 26


I m entioned in Chapter 1 (i.e., statements about particulars, identity statements and necessity statements): Regarding true statements about when and where macrophysical properties are instantiated, if the concepts employed to pick out those properties have no phenomenal constituents, then such statements are simp ly derivable from P, and the macrophysical phenomena at issue are reductively explainable simp liciter. One manifestation of the explanatory gap generated by the phenomenal is that a true statem ent like S is in pain at t is not derivable from P. As a result, macrophysical phenomena pi cked out via concepts that have phenomenal components are only reductively explainable modulo phenomenology in the sense delineated above, and so the explanatory gap that the phenomenal gives rise to extends in fact to an impressive number of macrophysical phenomena. Theoretical macro-micro identifications expr essed using terms such that the concepts associated with them have no phenomenal component s, if true, are derivable from P, if true. Another manifestation of the explanatory gap, th is time with regard to psychophysical identity statements, is that, by contrast, no such statem ents are entailed by P. This has important consequences for theoretical identifications involving macr ophysical phenomena that are reductively explainable only modulo phenomenol ogy, as those are derivable only from a full description of the microphysical augmented with the assumption that a number of relevant psychophysical causal connectio ns are in place. Something similar holds for metaphysical neces sity statements. If the phenomenal is not constitutive of the thoughts such statements express, then they are derivable from P.4 However, since the phenomenal is not entailed by the micr ophysical, in the case of metaphysical necessity 4 Its important that P includes all the microphysical laws. 27


statem ents about macrophysical phenomena that the phenomenal is conceptually constitutive of (e.g., Necessarily, water contains hydrogen), to get entailment, the full description of the microphysical would have to be supplemented with certain statements expressing psychophysical causal relations. Consequently, according to Chalmers, the e xplanatory gap is tantamount to lack of derivability from the microphysical. Since phenome nal concepts turn out to be constitutive of many ordinary macroscopic concepts (e.g., water, temperature, etc. ), the explanatory gap extends to such phenomena as well. This has a couple of interesting implications: First, thinking of the explanatory gap as the unavailability for th e phenomenal of a type of explanation that macrophysical phenomena like water are amenable to is strictly speaking mistaken, for it rests on a certain amount of idealizati on, to the extent that it pres upposes that concepts like water lack phenomenal constituents. Second, the connectio n between the explanatory gap and the gap intuition turns out to be less straightforwar d than one might have expected. While the phenomenal generates a gap intuition, nothing like that arises for water, temperature, etc., despite the fact that these cases are likewise charac terized by a failure of derivability from the microphysical. Chalmers argues that from the existence of the explanatory gap the falsity of physicalism follows. There are two versions of his argumen t from an explanatory gap to a metaphysical gapone that does not rely on his two-dime nsional framework and one that does. In light of the discovery that tr uths about water are not deducible from P either it is important to emphasize how the version of the argument from l ack of derivability from P to the falsity of physicalism which resorts to two-dimensional semantics goes. While the two-dimensional framework is explained in more detail in Chapter 5, a brief characterization of it will suffice for 28


our purposes here. In Chalm erss two-dimensional framework, every expression-token/concept is assigned two intensions. Its primary intension is a function from cente red logically possible worlds considered as actual to extensions, wh ile its secondary intens ion is a function from logically possible worlds consider ed as counterfactual to extens ions. A centered world is an ordered pair of a world and a center in that world, whereas a center is an ordered pair of a subject and a time. Let us turn now to the version of the argumen t from lack of derivability from P to the denial of physicalism which makes use of the twodimensional framework: If S is a truth that is not deducible from P, then it follows that physica lism is false, though not necessarily in virtue of the secondary intension of S, but rather because ther e must be some term in S such that either the property determined by its secondary intension, or that determined by its primary intension, or perhaps some property that actually realizes one of those is not a physical property, while it is by hypothesis instantiated in the actual world. Chalmers also advances a version of the argument from the gap to the falsity of physicalism which does not rely on the two-dimens ional framework. I will seek to clarify how that version of the argument unfolds. Chalmers argues that physicalism that postula tes psychophysical identi ties is implausible because psychophysical identities would be epistemically primitive, in the sense that they would not be derivable from underlying truths that dont involve identities (Chalmers, 2001, p. 19). That would make them have a quite singular status, for outside th e context of the mind-body problem, there simply are not any epistemically pr imitive identities. If the identity theory were right, then the only way for psychophysical identiti es not to be epistemically primitive would be to be entailed by P in conjunction with a comple te description of the phenomenal couched in a 29


phenom enal vocabulary (assuming, as Chalmers does, that P entails all the physical truths), but they are not so entailed. No te that explanatory reducti on modulo phenomenology furnishes identity statements that clearly are not epistemica lly primitive, for they are deducible from P plus statements of psychophysical causal connections. (It is important that appeal to causal connections here is suitably neutral between physicalism and dualismit might turn out that such causal relations obtain because phenomenal states are themselves identical with physical states, or because, even though phenomenal states are non-phys ical, there are irreducible psychophysical laws.) Chalmers seems to raise a similar objection against a supervenience type of physicalism: postulating necessary psychophysical connections that are weaker th an logical but stronger than nomic amounts to countenancing epistemically primitive necessities of a sort we have no good reason to believe exist. However, he does not spell out what it is for a necessity to be epistemically primitive. By analogy with the ca se of identities, one might propose that an epistemically primitive necessity is one that is not derivable from underlying truths not involving necessities, and the suggestion w ould be that the only kind of ep istemically primitive necessities we are familiar with are the nomic onesmoreover, as Chalmers says, one can argue that what it is to be a fundamental law of nature is precisely to be an objective [non-indexical], epistemically primitive counterfactual-suppor ting regularity (Chalmers, 2001, p. 20). There is an obvious problem with this attemp t to think of the epistemic primitiveness of necessities as analogous to that of identities. If a Humean account of laws is incorrect, then the laws of nature are not derivable from non-modal truths. On the account under scrutiny, that does render laws of nature epistemically primitive, as Chalmers claims they are, but leads to the unhappy consequence that statements like It is (metaphysically) necessary that water is 30


H2O/water contains hydrogen are epistemically primitive as well, for to have any chance of deriving those from underlying truths, one must rely on nomic necessities. In addition, by the above account, logical necessities do come out as they should--i.e., non-primitive--but for the wrong reason. Logical necessities are not epistemica lly primitive, but that is not because they are derivable from non-modal truths about the world; rather, logical necessities are not primitive because they are knowable a priori, which does ha ve the consequence that they are trivially derivable from anything. What makes their status as necessities intelligible is their a priori character. The following better captures Chalmerss notion of an epistemically primitive necessity: A necessity statement is epistemically primitive just in case there is no adequate explanation of its presumed truth. Further, there are only two ki nds of explanation of the presumed truth of a necessity statement that count as adequate: first, a necessity statement is adequately explainable if it is a priori; second, a necessi ty statement of a strength weaker than logical counts as adequately explainable if it is implied by non-moda l truths plus truths i nvolving only necessities of lower strength, if any. Thus, no mic necessities count as primitive if a Humean account of laws is incorrect. Note that the Kripk ean type of a posteriori necessities can provide no support for the claim that the phenomenal supervenes on the phy sical metaphysically but not logically, because statements like It is necessary that water is H2O/water contains hydrogen come out not epistemically primitive. Consequently, it seems that psychophysical correlations that are claimed to be metaphysically necessary have a peculiar feature among a posteriori necessities: they are not derivable from non-modal truths plus nomic necessities, and would thus be the only epistemically primitive a posteriori necessities. Given Chalmerss commitments, the problem with merely metaphysically necessary psychophysical correlations can be described in a more 31


specific fash ion as follows: Postulating such things is ad-hoc because they are not derivable from the conjunction of P with a full description of the phenomenal couched in phenomenal terms and with any statements expressing merely nomica lly necessary psychophysical correlations. 32


CHAPTER 3 A GENERALIZED DERIVABILITY CONCEPTION OF THE GAP In this chapter I accom plish a number of things. First, I turn my a ttention to the question as to the exact notion of the physical by refe rence to which the explanatory gap should be articulated, and I provide an account of what th is notion comes to, which I compare with the major contenders in the literature on demarcating the physical and show to be superior. In the process, I argue that Chalmerss construal of the explanatory gap rests on an assumption about the physical that is only contingently true, if true at all. Second, I offer a generalized derivability construal of the gap which dissociates the phe nomenon from such problematic assumptions and characterizes it in more fundamental terms. I end the chapter with an examination of the relationship between the explanatory gap and the ga p intuition and an explanation of the latter. Physics-Based Accounts of the Physical and Hempels dilemma Both those who accept and those who deny th at there is an explanatory gap usually interpret the phrase as purporti ng to refer to a gap between co nsciousness and the physical. Chalmers himself thinks that the physical in th e end boils down to the microphysical, and I will investigate in due course the natu re of the relationship that, acco rding to him, holds between the two, as well as the merits of his view. Typically, however, philosophers understand the gap phenomenon in terms of a general notion of the physical, which is presupposed to be antecedently familiar, and as such receives little attention in the context of discussing the nature of the gap and its metaphysical significance. Clarifying just what notion of the physical is at issue will be instrumental bot h in assessing Chalmerss view of the gap and in providing a perspicuous articulation of the explanatory gap phenomenon. How the physical should be understood is a vexed question. A tempting reaction, and one that is rather popular in cont emporary discussions, is to dema rcate the physica l by appeal to 33


physical theory. The difficulty which notoriously plagues attem pts to spell out the notion of the physical by reference to physics goes in the literature by the name Hempels dilemma.5 If the physical encompasses exactly the properties th at current physics quantifies over (together perhaps with those that are lo gically or metaphysically superv enient on them), since current physics is very likely incorrect, physicalism itself comes out as a false doctrine quite apart from the question about the metaphysical status of the mental. The histor y of science supplies a strong inductive argument in support of the claim that current physics might very well be neither true (as future generations might discover they have to withdraw some of the current posits), nor complete (for new things have been constantly added to the physicist s stock of fundamental entities, and there is no good reason to expect that this trend is a bout to come to an abrupt end). Alternatively, if, in order to bypa ss the errors and limita tions of current physics and not let them vitiate the conceptual framework used to articulate the mind-body problem, the physical is spelled out by reference to an id eal physics, the charge is that our current ideas about what ideal physics might look like are too h azy to give the associated no tion of the physical determinate content. An interesting twist on the second horn of the dilemma is to argue that in fact physics might end up countenancing patently mental proper ties (after all, the things it quantifies over have become increasingly stranger), in which case the definitional project that appeals to ideal physics has the consequence that physicalism is true, even though, intuitively, under the circumstances physicalism should be false. Two prominent reactions to Hempels dilemma are the so-called currentism and futurism (the terms were coined by Crook a nd Gillett), each of which challenges one of the two horns of the dilemma. The currentist maintain s that physicalism can be formulated in terms 5 See (Hempel, 1969) and (Hempel, 1980). 34


of current physics, and grants that, so unders tood, it is probably fals e, but rethinks the relationship that a philosopher m ust bear to physicalism if he is to qualify as a physicalist. The futurist spells out physicalism by reference to an ideal physics and argue s that the prima facie difficulties awaiting any attempt to give an accoun t of ideal physics that is congruent with physicalist aspirations can be ove rcome. The prominent espousers of currentism and futurism in the literature ar e Melnyk and Poland, respectively. Melnyk is far from impressed with the first horn of Hempels dilemma.6 He argues that the predicate physical used in the formulation of the phys icalist thesis s hould nonetheless be interpreted as referring to the entities and pr operties mentioned by current physics. Let us label physicalism so understood Pcp. The philosophical respectability of the physicalists commitments is, in his view, in no way diminished by inductive arguments extrapolating from systematic past failures to come up with a true physical theory to the present, because being a physicalist does not in fact require subscribing to Pcp. Nor does it require believing that Pcp is likely to be true, nor even that it is more likely true than false: To be a physicalist is to take the same attitude toward the hypothesis of physicalism which those who have broadly scientific realist and antirelativist intuitions take toward what they regard as the best of current scientific hypotheses (Melnyk, 1997, p. 625). That attitude, labele d SR is characterized as follows: To take the SR attitude toward a hypot hesis is (1) to regard the hypothe sis as true or false in virtue of the way the mind independent world is, and (2) to assign the hypothesis a higher probability than that of its relevant rivals (Melnyk, 1997, p. 625). It should be kept in mind that relevant rival is a theoretical te rm, which is defined by Mel nyk like this: [h]ypothesis H1 is a relevant rival to H2 if and only if (a) H1 is sensibly intended to ach ieve a significant number of H2s 6 See (Melnyk, 1997). 35


theoretical goals; (b) the hypotheses H1 and H2 fail to supervene on one another; (c) H1 has actually been formulated (Melnyk, 1997, p. 626). Melnyk further claims that adopting the SR attitude toward a hypothesis is in general compatible with not regarding it as more likely true than not, and a fortiori, it is consistent with not deeming it very likely. This is so because the negation of a hypothesis, taken in isolation, typically does not count as a relevant alternative to that hypo thesis, on account of failing to meet requirement (a) in the characterization of a rele vant alternative: it n eed not indicate even a minimal set of basic principles (Melnyk, 1997, p. 626), and therefore may not give any clue whatsoever as to how the explananda of the original hypothesis might be accounted for. Melnyk argues that Hempels dilemma fails to show that there are insurmountable obstacles in the way of articulating the philosoph ical position adopted by physicalists, which in the final analysis is tantam ount to holding SR toward Pcp. First, the mere negation of Pcp does not qualify as a relevant riva l. Second, the fact that Pcp is likely to be false, as the inductive argument from the history of physics suggests, in no way impugns the physicalists claim that Pcp is more likely than its relevant rivals, because, Melnyk maintains, the history of physics does not place physicalism at a disadvantage ove r its relevant rivals (he discusses in the context of this argument standard alternative views such as dualism). He then spends some time outlining the case for the explanatory superior ity of physicalism over some other major traditional competitors to emphasize that even though Pcp is likely false, the view proposed by physicalists might very well be likely true. Before looking closely at what I think is deeply wrong with Melnyks response to Hempels dilemma, let us briefly examine the me rits of an objection leveled against Melnyks 36


proposal by Crook and Gillett.7 They grant for the sake of th e argument that Melnyks account of being in good scientific standing is adequate, and object that Pcp does not qualify as in good scientific standing by Melnyks own standards. They propose that th ere is a relevant rival to Pcp, (P+), which is more likely than it. (P+) (a) There is some science, S, distinct from the totality of all the sciences, such that Ss laws and theories directly me ntion all the entities of physicalCUR entities and also at least one further entity which is wholly distinct from any physicalCUR entities, but which has features and magnitudes simila r to those of a physicalismCUR entity (i.e., certain types of particles, force, field, etc.); and (b) all ent ities either are identical to entities directly mentioned in the laws or theories of S, or are entities that bear the relation RN to entities directly mentioned in the laws or theo ries of S. (Crook & Gillett, 2001, p. 339) The justification for the claim that (P+) is more likely than Pcp consists in the familiar inductive argument lending support to the conclusion that current physics is incomplete. If compelling, this argument shows that Melnyk manages to inte rpret physicalism so as to avoid the horns of Hempels dilemma only at the cost of compromising the views epistemic credentials. Note that the objection is not that Pcp is not more likely than each of its relevant rivals. Melnyk is after all not committed to the higher likelihood of Pcp over traditional rivals such as dualism. The objection is that, even setting aside physicalism traditional rivals, dualism and idealism, there are alternative theories that are more likely than it. Nevertheless, the argument proposed by Crook and Gillett is far from convincing, as can be seen from the fact that it is apt to prove significantly more than its proponents have bargained for. The authors grant for the sake of the argumen t that it is necessary fo r a scientific hypothesis to be in good standing that it be more likely th an its relevant rivals, as characterized by Melnyk. With this assumption in the background, it can be shown that a similar argumentative strategy as the one directed against Melnyks physicalism can be employed to show that current physics is 7 (Crook, Gillett, 2001). 37


not in good scientific standing. Cu rrent physics is presum ably less likely than an alternative hypothesis, (C+), just as sketchily outlined as (P+), and according to which there is a future, ontologically more inclusive, physical theory w hose stock of entities is exhaustive and which posits at least one additional ent ity that has features and magnit udes similar to those possessed by the entities postulated by current physics. We do not fault the physicalist for failing to trade current physics for the sketchy (C+). By the same token, (P+) should not be considered a threat to the epistemic status of Pcp. (It is noteworthy that Crook and G illett formulate (P+) in terms of a relationship of similarity to the entities postulated by current physics. The very difficulty posed by Hempels dilemma suggests that spelling out that notion cannot be straightforward. Putting enough flesh on the bones of (P+) to tu rn it into a genuine rival of Pcp, as opposed to a hypothesis with hopelessly meager content, might requ ire having already found a way of spelling out physical that avoids Hempels dilemma. Note also that Crook and Gillett cannot drop the similarity requirement altogether, for were th ey to do so, (P+) could not be substantiated anymore by a straightforward appeal to the history of physics.) Perhaps Crook and Gillett could regroup and present their argument as an objection directed instead against Melnyk s understanding of what it is for a hypothesis to be in good scientific standing. In particular it might be tempting to conclude that Melnyks characterization of what counts as a relevant rival of a given hyp othesis is not restrictiv e enough. Note, first, that if they reworked their argument as suggested, they wouldnt ther eby be any closer to attaining their goal, which is to undermine Melnyks unde rstanding of physicalism. They have provided no reasons to suspect that, once the account of being in good scie ntific standing is suitably modified by restricting the notion of relevant rivals so as to rule out hypotheses like (P+) (which intuitively do not count as such), Pcp might not qualify as in g ood scientific standing. 38


Melnyks definition of a relevant rival is undoubtedly vague. I suspect that his reaction to (P+) would be to say that it is disqualified because it is not detailed enough to make clear either what sort of entities have been added to the ontology of current physics, or how adding them helps explain the relevant data Perhaps he would respond that (P+) does not meet his first requirement for being a relevant rival: because of its sketchiness it cannot be sensibly intended to achieve a significant number of [Pcps] theoretical goals (Melnyk, 1997, p. 626). In any case, the temptation to regard (P+) as a legitimate cont ender might be due to the presumed existence of another hypothesis, (P++), which physicists will be able to fo rmulate, though, only when they actually uncover a new kind of entity. The crucial di fference is that (P++) will mention explicitly that kind of entity as well as the properties of it that are s upposed to enhance the explanatory powers of current physics, and pe rhaps formulate new laws descri bing its behavior, while (P+) makes only indirect reference to these things via the expedient device of existential quantification. (P++) is of course no threat to Pcp, because we are not yet in a position to formulate it.8 Even though the criticism due to Crook a nd Gillett founders, Melnyks currentism is indeed an inadequate response to Hempels dile mma. Here is why. In what follows I will grant for the sake of the argument that SR is a correct characterization of the attitude scientists hold toward the best current scientific theories. To recapitulate, for Melnyk, being a physicalist is holding SR toward Pcp, i.e., toward the thesis that everything in the world is reducible to the entities and properties mentioned by current physics. Melnyks background view that 8 Crook and Gillett claim that the level of detail exhibited by (P+) should be considered adequate, for it at least equals, and most likely surpasses, the detail of dualistic or vitalistic hypotheses that are acknowledged by Melnyk as relevant rivals to physicalism (p. 341, fn. 13). This s eems clearly wrong with respect to dualism, as we happen to know about the new properties introduced by the dualist (i .e., the mental properties) in a very intimate way. Arguably, even vitalism fares better than (P+). For one thin g, it specifies the exact natu re of the explanandum that the vital forces are introdu ced to account for. 39


physicalism has the status of a scientific hypothesiswhich is used to lend credibility to the contention that it is illegitimate to demand of th e physicalist a stronger attachment to physicalism than is required of the typical scientist (namely, SR)is likel y to be found congenial by many physicalists. While flattered by the compliments that Melnyks construal of physicalism pays him, a staunch physicalist, I suspect, would nonetheless be unable to suppress the nagging feeling that Melnyks construal fails to capture the spirit of his metaphysical commitments. Melnyks account of what it is to be a physicalist is at best incomplete, as it omits a crucial element. First of all, it seems that what Melnyk offers is an account of what it is to be a physicalist at the present time, as opposed to a physicalist simp liciter. While qualifying as a physicalist at the present time might require holding SR toward Pcp, it certainly couldnt be exhausted by that, as it also plausibly requires willingness to expand the category of physical entities whenever physics does so, except of course for the case in which physicists decide to include irreducibly mental properties among the basic things they countenance. Under such circumstances, it seems that the ri ght thing to do for the physicalist is to admit defeat rather than mimic the physicist and adjust his notion of the physical accordingly. Without a diachronic element which indicates the exact way in which the physicalists use of the the physical is responsive to the development of physical scie nce, Melnyks account of what it is to be a physicalist at the present time is only pa rtial. Among those who hold SR toward Pcp, it fails to discriminate between the camp that would be in clined to extend the term to cover patently mental phenomena, and those who would withhold th e term in such instan ces. Intuitively, only the latter count as physicalists. The need to specify the acceptable attitudes a physicalist can adopt toward various things that future physical theories might end up positing brings to light the 40


inadequacy of currentism, which proposes to dem arcate the physical exclusively by reference to current physics. Melnyk is not unaware of the need for a thes is that captures the spirit of physicalism, which transcends particular formulations (Mel nyk, 1997, p. 633, fn. 21), i.e., a thesis that helps explain what Hobbes, a contemporary physica list and a future physicalist relying on an ontologically richer physics have in common. His response to this difficulty is unfortunately rather disappointing. He offers the conjunction of (1) and (2) as a sort of canonical formulation of physicalism in his sense, and claims that commitment to (1) might be the heart of a good candidate for capturing the spirit of physicalism in the intuitive sense. 1. There is some science, S, distinct from the to tality of all the sciences, such that every entity (property) is either itself mentioned as such in the laws and theories of S or is ultimately constituted (realized) by entities (prope rties) mentioned as such in the laws and theories of S. 2. S is current physics. (Melnyk, 1997, p. 633) It is unclear how much content ther e is to this suggestion, though, if (1) is assert ed to be just the heart of a good candidate. On the one hand, (1) by itself is obviously inadequate, because it is compatible with idealism (if S is psychology), as well as with dualism (if S is the conjunction of physics and psychology supplemented with psychophysical laws). On the other hand, if (1) needs to be conjoined with some other thesis, we are offered no clues as to what that thesis might look like. Another major response to Hempels dilemma, Polands futurism, challenges its second horn. Polands strategy9 is to draw a distinction between sp ecific physical theories developed at various points in time, on the one hand, and the invariant research program of physics, on the other. Essential to the latter is a set of di stinctive descriptive and explanatory goals which 9 See (Poland, 1994). 41


physics always strives to attain, and which specific physical theories m anage to fulfill to various degrees. Because of the unchanging nature of th e research program, the distinction renders intelligible the development of physics over time and it also provides a way of specifying in satisfactory detail what is mean t by an ideal physical theory. According to Polands characterization of phys ics as a research program, physics is the branch of science concerned with identifying a basic class of object s and attributes and a class of principles that are sufficient for an account of space-time and of the composition, dynamics and interactions of all occupants of space-time (Poland, 1994, p. 124). Ideal physics, by reference to which Poland proposes to define the physical, is understood as the physical theory that fully meets these goals. It is by no means clear, however, how this a pproach to defining the physical is expected to overcome the so called downward incor poration problem (to use Polands own happy coinage), which, as we have seen, is one of the notorious perils threatening attempts to delimit the physical by appeal to the ont ology of ideal physics, and is part of the second horn of Hempels dilemma. Rather disappointingly, disc ussion of this objection is postponed by Poland for the penultimate chapter of his book, where it is devoted a brief and rather unconvincing section, even though, on the one hand, solving Hempels dilemma is from the outset one of Polands avowed goals, and, on the other hand, the downward incorporation problem seems to be staring us in the face from the very beginning. Polands answer to the question What is th e physical theory that meets the requirements laid out in the research program? reads: No one knows. But the best guess that can be made is that it is current physics, or so me suitable elaboration, given that portions of current physics are underdeveloped or tentative (Poland, 1994, p. 126). The pressing question is, of course, 42


whether such a suitable elaboration m ight turn out to include straightforwardly mental properties. The trouble is that ideal physics, as Poland understands it, is by definition complete. Therefore, if the domain of non-mental properties turns out not to be causally closed because irreducibly mental properties in fluence the dynamics of some occupants of space-time, then these irreducibly mental prope rties come out physical. In response, Poland grants that his view of the research program of physics allows of such a possibility: [M]y reply to the objection is to bite th e bullet and allow that indeed it is conceivable that physics might be revised to incorporate mental [phenomena] [] into the physical basis (Poland, 1994, p. 331). Polands relatively sanguine at titude in response to this problem is rather puzzling. How could his ad mission fail to spell defeat for futurism? He takes pains to emphasize a couple of poi nts, which are intende d to show that his concession does not render the futurist approach to defining the physical ill-suited to ground an adequate formulation of physicalism. One of the aspects stressed by Poland is that the incorporation of mental phenom ena into the ontology of physics need not be ad hoc, i.e., motivated by the shear desire to save physicalis m, but could very well turn out to be based on solid methodological principles, just as the past incorporation of electromagnetic forces or massless particles was. This misconstrues the objection, however. The problem for Poland is precisely that physics might legitimately come to incorporate unmistakably mental properties, and his account of the physic al predicts that the phys icalist would win under such circumstancesa verdict that sounds intuitively wrong. Another aspect that Poland emphasizes is that a priori substantive constraints upon physical theories are ill-advised (Poland, 1994, p. 330), and he clai ms that forbidding physicists to include mental properties in their ontology would be just such a constraint. All of that sounds 43


right, bu t one wonders how it is supposed to make the predictions of Polands account of the physical any more palatable. Poland writes: if the choice is betw een allowing that the mental could be incorporated into the physical and imposing a priori substantive constraints upon what physicists may permissibly appeal to for theore tical purposes, the physicalist must lean towards the former (Poland, 1994, p. 332). The natural respons e that this defense e licits is that there surely are more choices than the two envisaged. There is the third opti on of allowing physicists to do their job and extend the ontological basis of their discipline as th ey see fit, but avoid turning physicalism into a trivial doctrine by refusi ng to define the physical by reference to ideal physics, or by reference to physics at all for that matter. My own account of the physical is a case in point (as will become clear in the next section). A different sort of consideration which Po land adduces in support of futurism is the following: [A] distinction must be drawn between th e ways in which the physical has been identified in the past and the more theoreti cally-based manner in which it is identified according to contemporary versions of physi calism. The former relied essentially upon a priori criteria, for example location, extension and impenetrability, whereas the latter leaves it to physics to identify the physical domain subject to various general and specific constraints upon physical theory c onstruction [] (Poland, 1994, p. 330) The reasonable thing to say in re sponse to these remarks is that while the notion of the physical which the physicalist operates with may be to a certain degree theoretically based, in the sense of responsive within certai n limits to the discoveries of physi cal theory, it would better not be completely reliant on physics. Poland seems to say that since physicalists completely defer to physics in matters concerning what counts as physical, should physics end up accepting irreducible mental properties, phys icalists would have to adjust their notion of the physical accordingly. What Poland fails to appreciate, it s eems, is that this argument can be much more 44


plausibly read as a reductio of the prem ise th at physicalists complete ly defer to physics on determining the boundaries of the physical. To sum up, I have examined the prospects for two major responses to Hempels dilemma (a problem that afflicts physics-based accounts of the physical), i.e., curre ntism and futurism, and reached a pessimistic conclusion with respect to both views. The Lack of Generality Objection against Chalmers In this section I return to Chalmerss cons trual of the gap, which I gave a reconstruction of in Chapter 2, and I argue that it fails to characterize the gap phenomenon in general enough terms, as it rests on an assumption about the physi cal that is at most contingently true. The assumption in question is that the physica l logically supervenes on the microphysical. It is unfortunate that Chalmers articulates the explanat ory gap problem for physicalism in terms of the relationship be tween the phenomenal and the microphysical A connected problematic move that Chalmers makes is from me re lack of derivability of the phenomenal from the microphysical to its non-physicality and hence to dualis m. As in the previous section we reached no viable response to Hempels dilemma, one cause for concern would be, of course, the analogue of Hempels dilemma for a microphysicsbased account of the ph ysical. That is not, however, the point that I wish to press in the present context. Rather, I will argue that a fully general account of the explanatory gap should not presuppose that there are no emergent physical properties. It is intuitive that, to ascertain whether the mental gives rise to an explanatory gap, and whether as a result we should re gard it as constituting a distin ct ontological realm, it is not necessary to determine whether special scien ces like thermodynamics, chemistry, or biology are entailed by microphysics. The mind-body problem, as well as the gap intuition, considerably 45


predate the putative discovery that microphysics entails all the scien tific theories which describe what one would ordinarily characte rize as non-m ental phenomena. The construal of the explanatory gap as lack of entailment by the microphysical is fully general, i.e., not dependent on assumptions about th e physical that are at best contingently true, only if (*) is not just tr ue, but logically necessary. (*) The physical logically supervenes on the microphysical.10 First, as we will see, in light of what some writers have referred to as the re-emergence of emergence11 in contemporary philosophy of science lite rature, the denial of (*) might even begin to look epistemically possible, given the present state of our knowledge regarding the interface between, for instance, chemistry and quantum mechanics. Second, even if (*) is in fact true about the actual world, if (*) is conceivably false, the mere l ack of logical supervenience of the phenomenal on the microphysical fails to eith er accurately characterize the explanatory gap exhibited by the phenomenal in worlds where (*) fails, or to warrant (w hen taken by itself) any metaphysical conclusions about the nature of phenomenal properties in the actual world. To reach such metaphysical conclusions, the actual tr uth of (*) would have to be established. This task is not only onerous but also unnecessary once the gap problem is formulated properly. Is it conceivable for there to be physical prop erties that fail to l ogically supervene on the microphysical?12 It was Kirk Ludwig who first pointed out to me the relevance to Chalmerss 10 Perhaps framing the entire discussion in terms of material rather than physical properties would have the advantage of avoiding the association with physics whic h the latter term encourages, and which, as will become clear, I think one must resist, if one is to reach a correct understanding of the notio n of the physical that is relevant to the mind-body problem. 11 See (Cunningham, 2001) and (Scerri, 2007). 12 Of course, many physicalists maintain that its not even true, let alone logically necessary, that the physical logically supervenes on the microphysical. They do so on the grounds that physical concepts of a certain sort have non-ascriptive content. However, this is perfectly compa tible with maintaining that in all possible worlds the physical metaphysically supervenes upon the microphysical. If one believe s that there is a substantive distinction to be drawn between metaphysical and logical necessity, the important question for my purposes here is whether its 46


account of the gap of historical em ergentism about the properties studied by various special sciences. Before theories like thermodynamics, ch emistry or biology came to be considered in many circles reducible to theories about lowe r level phenomena, there had been philosophers belonging to the British emergentist tradition initiat ed by Mill who maintained that the properties the central terms of such macrotheories refer to are emergent.13 If the notion of a property that is both emergent and physical is coherent, then (*) is conceivably false. With regard to the historical question about the British emergentists actual stance on the metaphysical status of the emergent chemical a nd biological properties, wh ats certain is that chemical and biological properties were accordin g to them fundamental (because emergent). A cursory look at the literature provides, so far as I can see, no in dication that they also thought of them as in some relevant sense physical. For inst ance, with respect to bi ology, the explicit target of their attack was vitalism, which postulated non-physical entities called entelechies. That leaves it open that, while adamant in their rej ection of non-physical entities, they might have deemed the biological a domain of non-physical properties. Is it genuinely conceivable for there to be physical properties that fail to logically supervene on the microphysical, and are merely nomically necessitated by it? One might feel inclined to wonder how complete information about the structure and dynamics of the world at the microphysical level could possibly fail to fix all the physical facts about it. One might feel the pull to think that, if there are macroscopic fa cts other than phenomenal facts that arent fixed by complete information about the physical features possessed by the most basic physical constituents of reality, then ont ologically speaking they are more akin to phenomenal than to sensible to maintain that there are possible worlds in which the physical fails to metaphysically supervene on the microphysical. Thus, even though many physicalists would reject (*), that isnt pertinent to my discussion here, because they lean on non-ascriptivism about conceptual content, which, in my view, is mistaken. 13 (McLaughlin, 1992) gives a helpful overview of the movement. 47


m icrophysical facts, and its unclea r that they really deserve to be called physical. Maybe one could extend the term physical to cover such phenomena, the thought is, but that would be a mere terminological move designed to conceal a cr ucial ontological similari ty between such facts and phenomenal facts. In recent years there seems to have been among scientists and philosophers of science a kind of revival of emergentism w ith respect to certain special sciences, for instance, somewhat surprisingly, with respect to chemistry. I will focus in what follows on chemistry as an example, but if chemistry turns out to be emergent, biology will likely also be emergent relative to microphysics, given its dependence on chemistry. It is rather unexpected to find contemporary defenses of the epistemic possibili ty of chemical emergence becaus e, with the advent of quantum mechanics, the relationship between chemistry and microphysics had come to be regarded as perhaps the typical example of reduc tion. McLaughlin, for instance, writes14: The quantum mechanical explanation of chemical bonding is a paradigm of a reductive explanation. He claims that the advances made in quantum mechanics, molecular chemistry and genetics render the doctrines of configurational chemical and vital forces enormously implausible (McLaughlin, 1992, p. 89). Configurational forces are the type of forces an emergentist is committed to: they are forces exerted by configurations of entities, an d are not resultant, but ra ther fundamental (i.e., complete information about the component entities, their properties, the forces exerted by them as well as the laws governing those is in sufficient to account for such forces). This view has more recently been challenge d, for instance, by Sce rri, who argues that we are currently unable to in fer from what quantum mechanics tells us about the micro level either what properties an element has or what re sults when two elements react together: 14 In (McLaughlin, 1997). 48


In the case o f elements we can predict part icular properties perhaps such as ionization energies but not chemical behavior. In the case of compounds, what ca n be achieved is an accurate estimate, and in many cases even pr edictions, regarding specific properties in compounds that are known to have formed between the elements in question. Quantum mechanics cannot yet predict what compounds will actually form. Broads complaint about the inability of mechanistic or classical chemis try to predict the prope rties of elements, or the outcome of chemical reac tions between any two given elements, remains unanswered to this day. (Scerri, 2007, p. 924) Scerri consequently finds McLa ughlins verdict about the prospects for emergentism entirely unconvincing (Scerri, 2007, p. 925). Let us next take a closer l ook at some of the evidence adduced in favor of the epistemic possibility of chemical emergentism. Both (B ishop, 2005) and (Hendry, 2008) explain that when it is assumed that chemistry is derivable from mol ecular quantum mechanics, it is in virtue of the following sort of picture about how chemical properties are accounted for. To explain the behavior of a certain chemical system, it is in principle enough to accomplish the following: First, a list is needed of the constituent par ticles (proton, electron, et c.) and their properties (mass, charge, spin, etc.). Then the relevant forces operative at the mi crophysical level must be specified (electromagnetic, strong-nuclear, weak-nuclear and gr avitational), as well as their magnitudes. In the end, the so-called Hamilt onian corresponding to the chemical compound must be calculated, which represents the total ener gy of the system, and is expressible as the sum of two operators, one for kinetic and one for potential energy. The resulting Hamiltonian is then taken to enable one at least in principle to account for the various as pects of the systems chemical behavior. One difficulty is that the true molecular Hamiltonian is mathematically unwieldy and in fact impossible to set down. Fortunately, the st andard story goes, mathematically manageable, though only approximately true, Hamiltonians can be calculated which suffice to achieve most explanatory goals. What both Bishop and Hendry point out, however, is that these Hamiltonians 49


used in practice differ from the true ones not only in the ir superior tractability but also in that they supply in fact additional key information wh ich could not be retrie ved even in principle from the true Hamiltonians, namely, informa tion regarding molecular structure. The true Hamiltonian cannot, for instance, distinguish between isomers, which are distinct substances that share the same elemental composition but differ in molecular structure, and which as a result can have very different chemical pr operties. As Hendry puts it, the problem is not that molecular structure is difficult to recover fr om the exact quantum mechanics, but that it is not there to begin with (Hendry, 2008, p. 527). To illustrate, Hendry discusses the quantum mechanical account of the spectra of a CO2 molecule and characterizes the form of the explanation given as follows: We use quantum mechanics to explain the motions of parts of the molecule within the context of a given structure for the molecule as a whole. The emergentist will see this as a case of downward causation: we did not recove r the CO2 structure from the resultant Hamiltonian, given the charges and masses of th e various electrons and nuclei; rather, we viewed the motions of those electrons and nuc lei as constrained by the molecule of which they are a part. (Hendry, 2008, p. 183) A question worth pondering is whether, if we were to discover that chemistry is emergent, we would automatically come to re gard the fundamental chemical forces with suspicion, as though they were somehow different in nature from the fundamental microphysical forceswhether we would declare them non-physic al. It is far from clear that we would. Interestingly, the philosophers of science who entertain the possibility that chemistry might be emergent do not appear worried th at, should this prove right, chemistry would turn out to deal in funny propertiesfunnier, at an y rate, than the microphysical. For instance, Bishops willingness to consider emergent chemical properties physical is quite clear: [B]ased on our best physical theories, there is no reason to suspect, aside from hidden reductionist metaphysical presuppositions and some misundersta ndings of science, that physics should be exhaustive of the physical []. (Bishop, 2005, p. 717) 50


Even if it is epistemically possible that chemistry is emergent, one possibility, for all we know, is that the story we currently have about the micro level leaves part of the chemical behavior of the system out of account because it is itself incomplete and/or mistaken in some other way, and chemistry is in fact not emerge nt after all. For example, maybe there are microphysical properties or forces that are missing from the explanans, but which a completed microphysics would make mention of. Perhaps the microphysical story that fails to predict the right molecular structures a nd hence the right behavior fo r substances like ethanol (CH3CH2OH) and dimethyl ether (CH3OCH3)which are isomersis in fact incomplete. Perhaps it misses some fundamental aspect of microphysical reality that would be enough to close the explanatory gap. Consider then the following subtraction thought experiment. The important point for my purposes is that a scenario seems conceivable wh ere this story is an exhaustive account of the microphysical, while the chemical beha vior of substances is the same as that in the actual world. This world is just as we, mistakenly, under th e current assumption, thi nk the actual world is. Given that we think of the chemical properties of actual substances as physical, should not the chemical properties in this scen ario qualify as physical as well? The chemical behavior is by hypothesis the same in the two worlds. The only difference is that in the alternative scenario some of the chemical forces responsible for chemi cal interactions are not resultant relative to the microphysical level, but rather basic, configurati onal. (Note that the microphysical element that I am assuming exists in the actu al world but is not known yet can only make an explanatory difference if it is fundamental. An d, arguably, if it is fundamental it is conceivable that it could be missing while the rest of microphysics stay s the same.) A thought experiment along these lines, if successful, shows that (*) is logically possible, and, since the scenario seems perfectly 51


conceivable, the burden of proof rests on the op ponent to make a plausible case that there are hidden problem s with it. To adopt a slightly different tack, bracketi ng any questions about the status of chemistry in the actual world, consider a possible world in which the bonding properties of chemical systems cannot be fully explained by a complete story about what happens at the microphysical level. The predictions microphys ics makes (based on a full description of the microphysical entities, their properties and the laws they obey) about the causal powers of chemical compounds and about what compounds chemical reactions shoul d result in are inaccurate. In this world, the presence of chemical forces has been documente d that are exerted only by configurations of particles. They are fundamental in relationship to the microphysical. Suppose, in addition, that these chemical forces are not mental, in the se nse that they are neither identical with nor constituted by mental properties. If one want ed to pose the mind-body problem relative to a world like this, would not it be natural to include the chemi cal properties on the physical (or material) side of the divide? Should we have any misgivings about including the emergent chemical properties from this s cenario in the physical realm, in the sense of physical that is relevant to the philosophical project of formulating the mindbody problem and articulating the array of proposed solutions to it? We should not, I submit.15 Whether such properties can legitimately be considered physical depends, of course, on which notion of the physical, exactly, is pertin ent to the mind-body problem. Barbara Montero has an interesting approach to framing the mind-body problem.16 She claims that, since attempts to spell out the notion of the physical tend to fall prey to Hempels dilemma, instead of 15 In fact, a further restriction will be included in the next section: the emergent chemical properties should be graspable (in a sense to be spelled out). 16 In (Montero, 1999), and (Montero, 2001). 52


construing the m ind-body problem in terms of th e contrast between the physical and the nonphysical, it is better to rethink it from the perspective of the me ntal vs. non-mental distinction. I think she is quite right when she says that the question that tro ubles philosophers who struggle to solve the mind-body problem is whether mentality is a fundamental feature of the world, which she reformulates as the question whether mentalit y can be accounted for in terms of non-mental phenomenawhether the mental is fundamenta lly non-mental. Montero does not specify how this should be further unpacked. She argues, however, that hopelessly obscure talk of the physical should just be replaced with talk of the non-mental. As will become clear in the next section, I adopt in this respect a different stra tegy: the notion of the nonmental can be pressed into service to spell out whats meant by the physical in a way that overcomes the problems which other attempts to demarcate the physical r un into. In the next se ction, I will revisit the question as to how exactly the mental is fundamentally non-mental should be understood, and I will employ the mental vs non-mental distinctio n understood as applied to vocabularies to give an account of the physical, which will then be instru mental in articulating a superior derivability construal of the gap. For the purpos es of this section, however, th e important insight that needs to be kept in mind is that reflection on the philosophical import of the mind-body problem reveals that, at least in many c ontexts, determining whether ment al properties are physical just comes down to determining whether they ar e reducible to non-mental properties. Our hypothetical scenario is a case in point: if the emergent ch emical properties are neither straightforwardly mental nor reducible to more basic properties that prove to be mental in nature, then in a statement of the mind-body problem, they should be placed on the physical side of the divide. 53


To see that this diagnosis is intuitively ri ght, let us compare the following two scenarios. Suppose that, in the first scenario, chemical properties are emergent but by hypothesis nonmental. Suppose, for simplicity, th at biological properties are s upervenient on the conjunction of microphysics and chemistry, and also that there is a further level of reality, call it the schmental, which is fully explainable in chemi cal and biological terms, and therefore logically supervenes on the microphysical-cum-chemical, but not on the microphysical taken in isolation. We are to assume that schmental concepts (unlike mental concepts) are functional, and the reason the schmental does not logically supervene on the microphysical by itself is that the realizers of these functional roles are not among the states microphysics talks about, but rather belong to the domain of quantification of chemistry or biology. There are obvious similarities between the way I depicted the schmental and ho w the espousers of anal ytic functionalism were (mistakenly) thinking of the mental. It is intuitive to maintain that, if the dualist were told that the mental has in our world the status that the schmental has in this hypothetical scenario, she would consider her view disconf irmed rather than vindicated. Imagine yet another scenario, which is just lik e the first in all respects, except that there are also mental properties instanti ated. Assume further, for the sake of the argument, that here the mental is not logically supervenient on the conjunction of the microphysical and the chemical. Unlike the schmental, the mental is ontologically fundamental, in the sense of not reducible to anything non-ment al in nature, and its being so has nothing to do with the relationship between microphysics, on the one hand, and chemistry and biology, on the other. Should this prove to be the actual world, the dua list would deem her view confirmed and would not care very much that, as she would likely put it the physical has turned out to be two-layered. 54


Let us now pause for a while on the exegeti cal question whether Chalmers holds that (*) is logically necessary. The quest ion whether the arguments he pr ovides adequately support even the actual truth of (*) will also come up in the course of the i nvestigation. Regarding the first question, Chalmers does not seem to discuss the issue fully explicitly anywhere. At least at first, a charitable reader would perhaps be inclined to interpret him as being committed merely to the actual truth of (*). There seems to be, howev er, a certain amount of textual evidence against such an interpretation. There ar e at least two places in which Chalmers provides lengthy arguments whose intended upshot, given the larger context, seems to be the mere truth of (*): (Chalmers, 1996, pp. 71-77) and (Chalmers, 2001, pp. 14-19). The trouble with these arguments, however, is that in the course of developing them, Chalmers does not seem to point out any distinctive features of the actual microphysical and macrophys ical facts and concepts which could account for the truth of (*) at the actual wo rld, if the assumption that (*) is conceivably false had been in the background. If Chalmers had regarded (*) as a mere contingent truth, he would have seen the need to fill in the stor y about the actual microphysical and macrophysical domains so as to rule out the possibility that th e actual world might be one of those where there are emergent physical properties. On the contra ry, one kind of consideration that Chalmers repeatedly emphasizes in these passages is that macrophysical concepts typically express only structural and functional conditions,17 as though that were enough to ensure that macrophysical properties logically superven e on microphysical properties. This assumption is, however, problematic. Consider functional properties. Of course, in any possible world, functional properties do logically supervene on lower level properties, but to take that to establish that, in the actual world, functional pr operties such as those studied in biologyChalmerss favorite 17 See, for instance, (Chalmers, 1996, p. 79). 55


exam plelogically supervene on microphysical propertie s is simply to assume that in the actual world all basic physical prope rties are microphysical and therefore to beg the question. Taking a closer look at one formulation of Chalmerss argument proves worthwhile: [H]igh level facts are enta iled by all the microphysical facts (perhaps along with microphysical laws). [] The logical supervenien ce of most high-level facts is most easily seen by using conceivability as a test for logical possibility. [] A world [micro]physically identical to ou rs but in which these sorts of facts differ is inconceivable. In conceiving of a microphysically identical world, we conceive of a world in which the location of every last particle throughout space and time is the same. [] Once this is fixed, there is simply no room for the facts in question to vary (apart, perhaps, from any variation due to variations in consci ous experience) (Cha lmers, 1996, p. 73). Another statement in a similar vein is: Th[e] biological facts are not the sort of thing that can float free of their physical underpinnings even as a conceptual possibi lity (Chalmers, 1996, p. 73). Before appraising this argume nt, I would like to note that it offers absolutely no details either about the inner workings of the chemistry, biology, etc., which supposedly could not float free of the microphysical, or about the microphysical fundamental particle s, forces and laws themselves. The assumption that this (perhaps co upled with a lay persons vague idea about what the sorts of things are that the special sciences are in the business of explaining, as well as about the types of concepts they typically use) suffices to put one in a position to conduct the relevant thought experiment is the sort of evidence that makes the reader suspect Chalmers of an attachment to more than the actu al truth of (*). In light of th e arguments offered above for the epistemic possibility of emergent chemical propertie s, whether (*) is true or not looks like more of an empirical matter than Chalmers allows. To assess his argument, note first what Chalmers means by a world being microphysically indiscernible from ours: not that it is indiscernibl e at a specific time, t (in the sense of the exact same microphysical facts obtain ing there as here at t, and the exact same microphysical laws being operative), but rather that it has an i ndiscernible microphysical history 56


(perhaps in addition to having th e sam e microphysical laws). It is instructive to examine the force of the resulting argument if he had had in mind the former. In that case, the reasonable rejoinder would be that it may very well be concei vable to have the same microphysical facts at t (and the same microphysical laws) but different chemical propensities, because, for all we know, there might be chemical forces responsible fo r bonding patterns which are not explainable in terms of elementary particles, their properties and the laws governing them. We simply do not know enough to be in a position to perf orm the thought experiment properly. Chalmers has in mind the diachronic option, though. And one begins to wonder whether he might not be right when he says that keepi ng fixed the entire collect ion of microphysical facts throughout history (which include s the facts about the distribution of every last particle and field in every last corner of space-time) leaves very little room for anything else to vary (Chalmers, 1996, p. 72). This is an important argument to address, because it is akin in spirit to the argument I mentioned at the beginning of th is section, according to which, if higher level facts are not fixed by how things are physically at the most fundamental level of analysis, then it is not clear they could be physical. The reason why the version of Chalmerss argument that is based on the diachronic constraint will seem to an emergentist harder to rebut than the version based on the synchronic condition is that an emergentist about chemis try, for instance, upholds downward causation, and therefore claims that some of the microphysical events are caused by chemical events. The first scenario leaves plenty of room for downward causation, because it does not stipulate either the past or the future microphysical events (rela tive to t). It does lay dow n the microphysical laws, but leaves room for interference from above by not specifying exactly how things unfold. The second scenario, by contrast, asks the emergentis t to keep fixed also the microphysical events 57


that on his view are due to fundam ental chem ical causes. This unfairly requests of the emergentist a conceivability feat more demanding than is necessary for a successful defense of his position. Even so, the emergentist is able to rise to the occasion. The only option the emergentist is left is to vary the emergent laws that connect the microphysical to the chemical in a way that do es not affect the dist ribution of microphysical properties. Suppose that in this alternative worl d, when there is a chemical reaction involving two chemical compounds, the config urational forces responsible for it are slightly different than in the actual world, but the differences cancel out, so that the outcome is the same as the actual one. This line of thought suggests that Chalmers has failed to show that it is inconceivable to have a world whose microphysical history and la ws are the same as in our world, but where chemistry is different. Admittedly, the differences in configurational forces and bridge laws between this world and the actual world do not ha ve any opportunity to manifest themselves in different chemical behaviors, but that is only because of the extra constraints gratuitously imposed by Chalmers on the thought experiment. Interestingly, Chalmers further argues th at if biological fact s were not logically supervenient on the microphysical, then there w ould be no way that we [could] know those facts on the basis of external evidence (Chalmers, 1996, p. 74) and concludes that, since biological facts raise no special epistemic challenges, they cannot be emergent: If there were a possible world [micro-]physically identical to ours but biologically distinct, then this would raise radical epis temological problems. [] [I]f I were in the alternative world, it would certainly look the same as this one It instantiates the same distribution of particles found in the plants and animals in this world; indistinguishable patterns of photons are reflected from those en tities; no difference w ould be revealed under even the closest examination. It follows that all the external evidence we possess fails to distinguish the possibiliti es. (Chalmers, 1996, p. 74) Note that, for the reasons indi cated above, there is no worry he re for the emergentist. The differences between the two wo rlds are not salient only on th e diachronic interpretation of 58


m icrophysical indiscernibility, a re sult due solely to th e unjustified further restrictions set by Chalmers. On the alternative, sy nchronic, interpretation, there can be quite striking differences in the chemical interaction between various chemical compounds, a nd consequently, also in the behavior of some microparticles. Whether or not Chalmers would be ready to endorse the logical necessity of (*) (as opposed to its mere truth), it seems fair to say th at the evidence marshaled by him falls short of establishing even the truth of (*). I have argued in this section that it is logically possible for there to be emergent physical properties. Consequently, one disadvantage of the account of the gap as lack of derivability from the microphysical is that it fails to characterize the phenomenon in fully general terms, because it rests on contingent features of the physical, such as being one-layered (if it is indeed so). Further, if the claims of philosophers of science such as Scerri, Hendry and Bishop about the interface between chemistry and microphysics are to be trusted, emergent physical properties are not just logically, but also epistemically possible, given the current state of our knowledge. Even if there are in fact no emergent physical properties establishing that ther e are not proves, upon reflection, to be highly nontrivial. Ascertaining whether phenomenal properties are ontologically fundamental, however, does not requi re establishing that. Therefore, another drawback of the account of the ga p as lack of derivability from the microphysical is that it unhappily entangles these two separate tasks. Delimiting the Physical In the current section I revert to the task of demarcating the phys ical. After critically examining a proposal by Wilson and another on e put forth by Crook and Gillett, I advance a novel account of the physical which is superior to all those discussed. 59


One thing that the logical possi bility of emergent physical properties brings out is that, quite apart from the specific problems that bese t various types of physic s-based accounts of the physical, all of them are also misguided for a common reason: they rule out the logical possibility of emergent physical properties, or, to be more preci se, the logical possibility of the physical outstripping physics. It should be mentioned that, to avoid some of the pitfalls of pure physics-based accounts (such as the downward incorporation problem fa ced by futurism), some philosophers combined reliance on physics with an app eal to the non-mental According to a modified version of futurism, those properties count as physical that are either postul ated as fundamental by an ideal physics and non-mental or reducible to such prop erties. Wilson embraces a similar sort of mixed strategy. She writes: An entity existing at a world w is physical if and only if 1. it is treated, approximately accura tely, by current or future (in the limit of inquiry, ideal) versions of fundamental physics at w; 2. it is not fundamentally mental []. (Wilson, 2006, p. 72)18 It is quite plain that this entire family of hybrid approach es can ultimately fare no better than the pure physics-based accounts, because th ey share the assumption that physics could not fail to exhaust the physical. A construal of the physical th at looks much more promisi ng is due to Crook and Gillett.19 They avoid Hempels dilemma by explicating the notion of the physical independently of physics altogether, whether current or ideal. P hysicalism is formulated by them, roughly, as the 18 Her formulation differs from the one I just gave in that, first, it avoids exclusive reference to ideal physics because of certain rather mysterious and very sketchily presented epistemic concerns about formulating physicalism in terms of ideal physics (doing so would allegedly weaken our epistemic purchase on the thesis, which can only be grounded in the successes of curr ent physics), and, second it is intended to cover any pos sible world, not just ours. 19 (Crook & Gillett, 2001). 60


thesis that all instantiat ed properties are physical.20 Further, a property c ounts as physical iff either it is a basic physical prope rty or it bears a certain relation R to basic physical properties. (R is a place holder for the metaphysical relation ship in virtue of whose holding between x and y, if y is a physical property, x also counts as one, and the account can be filled in appropriately depending on which type of such relation is de emed adequatee.g., realization, metaphysical or logical supervenience, etc.) Crucial is the additional claim that a property is basic physical iff it is ontologically basic and non-mental. This might look like a very appealing account of the physical. Not only does it escape Hempels dilemma, but it has two further advantages: First, it accommodates the logical possibility of emergent physical properties. Secon d, it avails itself of th e attractive suggestion to characterize the physical in terms of the non-mental, while promisi ng to avoid the difficulty that the straightforward way of implementing that sugg estion runs into: if the physical is simply identified with the non-mental, physicalism come s out self-inconsistent. (Montero, for instance, who highly recommends framing the mindbody problem is terms of the non-mental,21 likes to state physicalism as the view that the mental is fundamentally non-mental. Unfortunately, she never unpacks the qualification fundamentally, wh ile noting that it cannot be discarded on pain of contradiction.) Crook and Gillett propose what looks like a nice way of solving this very problem. The collection of physical properties can naturally be divided into basic and no n-basic properties, and while the sweeping equation of the physical a nd the non-mental would be ill-advised, equating 20 There are in fact a couple of differe nces between this formulation and that advanced by Crook and Gillett, but I have disregarded them for the sake of simplicity because they are irrelevant to my discussion. First, Crook and Gillett employ in their statement of physicalism the generi c term entity rather than property, and intend it to cover individuals, properties, states of affairs, events and processes. Second, their formulation of physicalism is restricted explicitly to logically contingent objects. 21 In (Montero, 1999) and (Montero, 2001). 61


the two only with respect to basic properties a ppears harm less. The assu mption this move rests on is that it is simply not open to the phys icalist to claim that mental properties are basic physical properties. Mental properties tend not to look phys ical, and the intuition is that, if the physicalist manages to propose nothing better than that we should accept mental properties as primitive physical properties, he is just cheating and in fact simply dema nding that the extension of the term physical be expanded in an ad-hoc fashion. What are the merits of the account of the physical by Crook and Gille tt? Note that their construal of the physical makes an outright inconsistency of the thesis that (at least some) mental properties are basic physical properties.22 This is an infelicitous re sult. Consider the following two views about the nature of phenomenal properties: 1. Phenomenal terms express func tional descriptions of categorical properties. The categorical properties they refer to via functiona l roles played by them are physical. 2. Phenomenal properties as pres ented by phenomenal concepts cer tainly do not appear physical. However, the reason for this appearance is that deploying phenomenal concepts is bound to obscure from view the nature of the property picked out. It is cons titutive of tokening a phenomenal concept that one inst antiates the property the concept in question refers to. Nothing like that happens when tokening a physi cal concept, which makes it the case that phenomenal concepts look like they are apt to c apture phenomenal properties in a way that no physical concept could. From that it is hast ily and fallaciously concluded that physical concepts could not refer to phenomenal properties. It is undeniable that both 1 and 2 are versions of physicalism is th e so-called realizer functionalism, as opposed to r ole functionalism, while 2 goes along the lines of Papineaus story about the antipathetic fallacyand it would seem that e ach can be conjoined without danger of overt contradiction with the furthe r thesis that the physical properties some phenomenal terms refer to are basic. (It is interest ing that the conjunction of 2 and the thesis that some phenomenal properties are both physical and basic is available in principle even to 22 The hybrid approach that defines the physical as the domain of properties countenanced by an ideal physics and which are not mental, under certain assumptions, appears to have the same consequence. 62


som ebody who grants that phenomenal concepts are hyperstable.) In general, what the physicalist cannot do is propose that phenomenal pr operties are physical and basic and leave it at that. But as long as he acknowledges the associated explanatory burden and seriously attempts to discharge it, he can enter the competition as a legitimate contender. It may very well be that expanding 1 and 2 as suggested results in vi ews that turn out to have a hard time measuring up to the explanatory st andards they are expected to meet, or turn out to hinge on suspect theses about the import of phenomenal concepts. Further, advocates of 1 and 2 would tend to look at empirical research to identify the refere nts of phenomenal terms, and such research might eventually favor metaphysica lly complex properties. What should be kept in mind, however, is that identifyi ng phenomenal properties with basic physical properties is a position in logical space, and therefore a construal of the physical that predicts otherwise is inadequate.23 I turn now to introducing a novel account of the physical, which possesses a number of desirable features, such as the following: First, it is not beholden to physics (or any other science, for that matter), and therefore goes between the horns of Hempels dilemma. Second, it accommodates the logical possibility of emergent physical properties. These two features it shares in common with the construal due to Croo k and Gillett. Third, it avoids all the objections that I leveled against the account by Crook and Gillett. This last virtue is due to my taking a linguistic turn in defining the physical, and a pplying the predicates mental/non-mental to expressions in the language, rather than to properties themselves. 23 If identifying mental properties with basic physical pr operties is a position in logical space, a slight rephrasing of the second horn of Hempels dilemma is in order. The possibility that the list of basic properties countenanced by ideal physics might include mental properties is in itself no cause for concern for the futurist. Futurism can be successful, however, only if phenom enal properties as referred to via phenomenal concepts are barred from the ontology of ideal physics. The verdict I reached about the plausibility of Polands response to Hempels dilemma remains, though, in place. 63


Let L be an ideal language whose expressive powers are such that fo r any concept C that is graspable by some possible cognizer, there is in L a predic ate P which expresses C. Some predicates of L have the distinguis hing feature that it is a priori e ither that everything is such that if it satisfies that predicate, then it is conscious, or that everything is such that if it satisfies the negation of that predicate, it is conscious. I will call those phenomenal predicates. The predicates of L that lack this feature are the non-phenomenal predicates of L. Non-phenomenal predicates that have the further feature that their analysis in L involves no phenomenal predicates will be called phenomenal-free predicates. (The notion of the analysis of a predicate in L builds in the assumption that the analysans is formulated in most basic terms.) A pure phenomenal predicate is a phenomen al predicate such that atomic sentences containing it do not entail that something is P, for any non-phenomenal predicate, P, in Lexcept, I suppose, for the existence and self-identity predicates. A phenomenal predicate that is not pure in this sense will be called an impure phenomenal predicate. Let A be the set of all true sentences in L that contain only pure phenomenal predicates (i.e., contain neither non-phenomenal nor impure phenomenal predicates). Let B be the set of all true sentences in L that contain only phenomen al-free predicates. A natural proposal to make about what it takes for the phenomenal to be fundamental would be the following: (1) The phenomenal is fundamental just in case B fails to either entail or metaphysically necessitate A. (Subsequent discussion will reveal that this prin ciple actually builds in an assumption which may be deemed problematic. Ill return to this point in the last section of Chapter 3.) To give an account of what makes a property physical, it will be help ful to first introduce the notion of a hyperstable term : A hyperstable term is one that is insensitive to context, rigid (in 64


the sense that it picks out the sam e property relati ve to any world considered as counterfactual), and not world-indexed. The first tw o constraints are intended to en sure, in the terminology of the two-dimensional framework, that the primary a nd the secondary intensi on of hyperstable terms are both constant functions. The third requirement is intended to rule out expressions such as being Anas favorite property at t in @, which are otherwise both contex t insensitive and rigid. Intuitively, the concepts expressed by hypersta ble terms reveal to the subject who grasps them the very nature of the properties they pick out. Chalmers discuses in a different context the notion of a semantically neutral term. Being se mantically neutral (which comes down to being insensitive to contextwhere the world of utte rance is understood as a contextual parameter, too) is necessary but not sufficient for bei ng hyperstable. Since any hyperstable term is by definition semantically neutral, the primary inte nsion of a hyperstable term is identical to its secondary intension (modulo centering). Here is a preliminary suggestion about how to characterize the notion of a physical property: (2) A property is physical just in case either there is a hype rstable phenomenal-free predicate in L that picks it out, or it is logically/metaphysically necessitated by such properties. (2) is problematic as it stands for two reasons First, there could be functional terms (i.e., terms that pick out functional prope rties, rather than terms that pick out lower level properties via functional descriptions of them) which are phenomenal-free and hyperstable but pick out functional properties that are re alized by phenomenal properties. Second, it could be objected that (2) fails to rule out intentional propert ies, if one makes a couple of assumptions, among which that intentionality can c onceivably be dissociated from c onsciousness. If it is conceivable 65


for there to be intentionality without consciousness, the objection goes, then (2) classifies inten tional properties under the heading phys ical, which is awkward, because it is counterintuitive to claim that if consciousness we re shown to be reducible to intentionality, it would be thereby proved to be physical. I should say, first, that I am doubtful that intentionality is conceivable without consciousness, and, second, that those who tend to disagree are typically philosophers who do so in order to have an eas ier time making a case that intentionality is reducible to physical properties. As the view that intentionality could exist without consciousness but is nonetheless non-physical looks like a position in logi cal space, I will not incur unnecessary commitments and will qualify (2) as follows. Let us introduce the no tion of an intentionalfree predicate: a pred icate in L counts as intentional-free just in case it is non-intentional and there are no intentional predicates in its analysis in L. Further, a predicate is non-intentional if and only if it is not a priori either that everything that satisfies it has an intentional state, or that everything that satisfies its negation has an intentional state. I propos e to replace (2) with (3): (3) A property is physical just in case ei ther there is a hyperstable phenomenal-free, intentional-free predicate in L that picks it out and which is not functional, or it is logically/metaphysically nece ssitated by such properties.24 After deflecting in the next section some objections that could be raised against my account of the physical, I turn in the last section of Chapter 3 to developing an account of the explanatory gap and of the gap intuition. 24 To ward off further objections, other qualifications might be needed. Arguably, one would have to exclude from the relevant category of predicates those that are modal, those that can be satisfi ed by abstract objects, as well as the normative predicates. 66


Defense of the Proposed Account of the Physical from Objections The main goal of this section is to res pond to certain objections that could be raised against the account of the physical I have delineated in the foregoing. Some attempts in the literature to fra me the mind-body problem in terms of the nonmental (which were chiefly motivated merely by the desire to sidestep Hempels dilemma) have prompted the criticism that, upon closer inspecti on, they fail to successfully handle Hempels dilemma because a structurally similar objecti on surfaces when seeking to offer support for physicalism via the familiar causal argument: (CP) Any physical event that has a caus e at t has a physical cause at t. (CE1) Mental events cause physical events. (NO1) There is no overdetermination (of physical events by physical and mental events). (P1) Therefore, mental events are physical. The problem is alleged to emerge when trying to buttress the causal closure principle stated this time in terms of non-mental properties. This objection is raised by Gillett and Witmer,25 and is directed against a proposal due to Papineau and Spurrett.26 While, as we will see shortly, the target of the objection is substa ntively different from my own approach, the objection is also relevant to my story about how the physical should be understood. If Gillett and Witmer are right, then it might seem that physicalism understood in terms of my notion of the physical is in danger of encountering the specter of Hempels dilemma as soon as we look into the evidence typically offered in its favor. In the rest of this section Ill exam ine the objection and argue that it is not successful. To give a little bit of contex t, in a 1999 article, Papineau and Spurrett focus on the causal closure argument for physicalism and consider a version of Hemp els dilemma that targets the 25 (Gillett & Witmer, 2001) 26 (Papineau & Spurrett, 1999). 67


causal closure principle itself: if the causal closure principle is understood by reference to current physics, then it is very likely fa lse, w hile if it is interpreted in terms of ideal physics, it is a nebulous thesis, unless we get a re latively firm grip on the notion of ideal physics. One might be tempted to circumscribe it as the complete theory that explains a certain domain of phenomena, but then both the causal closure principle and physicalism become trivially true. Therefore, one must avoid building completeness in to the definition of ideal physics, but it is not clear how else to draw its boundaries without presupposing an antecedent notion of the physical. In response, Papineau and Spurrett claim that a fruitful move would be to simply shift focus from the argument formulated in terms of physical properties to a parallel argument phrased in terms of non-mental properties. While the authors do not propose this as a definitional move (out of a desire to avoi d terminological tangles and concen trate on issues of substance), they nonetheless style the thesis that the reformulated argument yi elds a version of physicalism (p. 25). The reformulated argument is labeled by Witmer and Gillet the via negativa: (CNM) Any non-mental event that has a cause at t has a straightforwardly non-mental cause at t. (CE2) Mental events cause straig htforwardly non-mental events. (NO2) There is no overdetermination (of non-mental events by straightfo rwardly non-mental and mental events) (P2) Therefore, mental events must be understood in terms of straig htforwardly non-mental events. The criticism raised by Gillett and Witmer agai nst the via negativa is two-fold. First, they claim that if (CP) falls prey to Hempels dile mma, so does (CNM) (or at least a good prima facie case can be made to that effect, and therefore it is incumbent on th e proponents of the via negativa to show that, despite appearances, (CNM) has better chances of escaping Hempels dilemma than (CP)). Second, Gillett and Witmer argue that there are respects in which (CP) is in fact in a better episte mic shape than (CNM). 68


I will discuss the two criticisms in turn. Regarding the first criticism, a preliminary remark is in order. I have argued that the not ion of the physical is not tied to physics, and therefore on my view (CP) as such is not subj ect to Hempels dilemma. A slightly modified version of the first objection is nonetheless pert inent to my account of the physical. According to it, my account of the physical rende rs one of the theses usually offered in support of physicalism (namely, the principle of the causal closure of the physical) vulnerable to an analogue of Hempels dilemma. As part of the stage setting for their first cr iticism of the via negativa, Gillett and Witmer argue that (CNM) is a tenet that greatly expa nds (CP) (Gillett & Witmer, 2001, p. 304). Interestingly enough, their motivation for holding that is not the desire to leave room for some unusual sort of robust metaphysical trialism (i.e., the view that th ere are three different kinds of propertiesmental, physical, and ne ither mental nor physical). Rather, they interpret physical as tied to physics, and take (CNM) to cover in addition the kinds of events that it is in the business of the special sciences to explain. Of course, the potential fals ifiers of (CNM) that most r eadily spring to mind are bodily movements. (CNM) entails (1). (1) Bodily movements have sufficient straightforwardly non-mental causes. Gillet and Witmer ask how the claim that bodily movements are fully explainable in terms of non-mental events and laws governing them is justified. They argue th at its justification necessarily goes via an appeal to physiological research: It is plausible to maintain that bodily movements are causally explainable in non-mental terms to the extent that it is reasonable to expect bodily movements to be causally explaina ble in terms of physiological events and laws. (2) Every bodily movement is causally determined by physiological events. 69


But then, the charge is, even though the term non-mental is not itself beholden to theory change, the predicate physiological is, and consequently the just ification of (CNM) necessarily rests upon a thesis which falls prey to Hempels dilemma (i.e., (2)). If physiological in (2) is tied to current physiological theory, since the latter is probably incomplete, (2) is likely false. Alternatively, if physiol ogical is spelled out in terms of an ideal physiology, (2) proves vague, and the danger arises that th e ontology of an ideal physiological theory might end up including straightforwardly mental events. Either way, (2) is unsuited for th e role of justifier for (CNM). I do not think that Gillett and Witmers first criticism against the via negativa is successful.27 The advocate of the via negativa could respond to this objection that the second horn of Hempels dilemma can easily be bl unted. Why does (1) tend to look compelling? Arguably, because the physiological information we possess at present seems to suggest that a complete explanatory story for be havior should in principle be available which is formulated only in terms employed by current physiology, perhaps supplemented with terms that are sufficiently similar to those, should it turn out that the ontology of current physiology is incomplete. If we are not yet in possession of su ch a complete account, it is only because of the limited character of our current knowledge about the exact way in which various complex bodily processes (among which are brain processes) wo rk, but a fully worked out theory could in principle be attained. If physiologi cal in (2) is tied to an idea l physiology (which, in turn, could be plausibly characterized as the complete th eory that meets the explanatory goals deemed 27 (Montero & Papineau, 2005) also defends the via ne gativa against the criticism mounted by Gillett and Witmer. Their defense is different from mine and I think fails For reasons of space, I cannot discuss it here. 70


distinctive of physiology, whatever those are), then the proponent of the via negativa could sim ply substitute (3) for (2): (3) Bodily movements have as sufficient causes straightforwardly non-mental physiological events. It is worth emphasizing that, because emergent physical properties are conceivable, it is better to resort to idea l physiology when substantiating the causa l closure principle than to appeal to ideal physics in the very de finition of the physical. One thing that their first criticism of the vi a negativa brings to light is that Gillett and Witmer interpret the causal argument in what I ta ke to be a rather idiosyncratic fashion. They understand the predicate physical as having its extension fixed by physical science and not covering, for instance, the states and events that the special sciences are in the business of explaining. In particular, it is ra ther striking that they use phy sical in a way that excludes behavior from the category of physical events. I ta ke it that this is diffe rent from how the causal closure argument is typically understood. The physicalists who wo rry over whether physical in (CP) should be explicated by reference to current or ideal physics usually think that physics is broad enough to count as the relevant discipline because they assume that the properties within the purview of other natural scie nces are logically/metaphysically supervenient on those studied by physics. Gillett and Witmer imp licitly question the epistemic st anding of that supposition, as they find very problematic even the justificatory status of the weaker claim that the domain of properties studied by the conjunction of physics, chemistry, biology, etc., is causally closed. They deem (CP) easier to establish than (CNM ), and hold that much of the appeal of the [original] causal argument appears to derive fr om this very point ( p. 308). Even though I find their reading of the causal argument rather peculiar, if their view of the relative epistemic merits 71


of (CP) and (CNM) is correct, then barring unexp ected d ifficulties with the other premises, the causal argument as interpreted by them would clearly be superior to the causal argument on what I think of as its us ual interpretation. The gist of the second criticism due to Gille tt and Witmer is expressed in the following passage, worth quoting in full: It seems to us that CP is clearly in a superi or position. Physics supplie s us with a relatively short list of forces and laws that offer a complete explanation of say, why a particular electron accelerated when it did. In contrast, a startling number and variety of factors, usually with ceteris paribus conditions, must be invoked in explaining a bodily movement and these factors reach beyond physiology itself to other sciences such as biochemistry and genetics. [] [T]he special sciences must consta ntly stray into other fields of inquiry and their findings, not least because their objects of st udy are often composed by the entities of such sciences. [] In contrast, given the di fferent character of its entities, physics may study its objects relatively unconcerned with othe r sciences and its prospects for complete explanations are far rosier. (Gillet & Witmer, 2001, p. 307) It is highly doubtful that (CP) could be more secure epistemically that (CNM). When it comes to causal closure, the narrower the domai n of properties declared causally closed, the more difficult to establish the corresponding causal closure thesis. It sounds plausible to claim that physics prospects for complete explanations are rosier than those of special sciences like biology only against the background of some heavy idealizations, which are illegitimate in the context of the causal argument for physicalism. In particular, one must abstract away from the fact that some of the events studied by physics appear to have among their causes events within the purview of the special sciences. Focus for a second on (CE). When we find it intuitive that the me ntal is causally efficacious, it is not because we have evidence that mental events tend to cause specific types of microphysical events, or evidence that they tend to bring about events indivi duated at the level of analysis where physics broadly construed operates. Rather, the sort of evidence that comes to mind consists of cases where, as we typically put it, mental states caus e behavior of various 72


kinds. (It is therefore rather unexpected for Gill e tt and Witmer to examin e precisely the case of behavior in their discussion of the comparative merits of (CP) and (CNM)). If we interpret physical in the causal argument by reference to phys ical science (whether current or ideal is, as we know, a vexed question, and the realization th at the criticism of via negativa proposed by Gillett and Witmer seems to presuppose that a satisfactory physics-based response to Hempels dilemma is available makes one feel a bit uneasy) and let that be called physical*, then (CE) has to be interpreted as the claim that mental states cause physical* events. Does ordinary experience provide evidence for this modified claim? Suppose that, as we like to put it, my desire to move my arm caused me to move my arm. Even though intuitively we talk about desires causing behavior, some of what is going on at the effect end of this causal relationship can be described at a physical* level of analysis: the desire caused a physical object (the arm) to describe a certain trajectory in space. Moreover, the resulting behavior can cause further changes in the environment which are unmistakably physical*, such as a ch air toppling over. Why is it that even though th e physicalist grants that mental states cause behavior, she insists that the physical (in which she typically includes behavior) is causally closed? Because she is confident that a complete causal explan ation of behavior can be given in purely nonmental termsi.e., an explanation that traces back that piece of behavior to various physiological processes in the body and ultimately to neurological happenings. That prospect renders mental states explanatorily expendable, and the further belief that God couldnt have been such a bad engineer as to design a wo rld where overdetermination is systematic and widespread forces the conclusion that, if mental states are really causally efficacious, their underlying nature must be physical. The events that make up this ca usal chain leading from brain goings-on to behavior appear to exceed the bounds of physics, however and if Gillett and 73


W itmer have doubts that the spec ial sciences which study these ev ents and the laws governing them will be able to offer complete causal explanations in the foreseeable future, it seems to me that consistency forces them to accept that (CP), even interpreted in terms of physical* properties, partakes of the same kind of epistemic insecurity as (CNM). To bring into sharper focus the point that I am pressing, higher level physical events appear to cause physical* events. If the physic al logically/metaphysically supervenes on the physical*, all these instances of downward causation can be explaine d as derivative upon instances of same-level causation between physic al* events. The hypothesis that the physical is not causally closed while the physical* is (which seems to be entertained by Gillett and Witmer as an epistemic possibility) en tails that higher level physical events are something over and above the physical* events and, ther efore, an argument is needed to establish that the appearance of downward causation (in this cas e of a quite robust variety, because not underwritten by samelevel causation) is an illusion. The suggestion that when we study physical* events in isolation from any possible influence from chemical, biological and mental events, we discover that complete causal explanations in physical* term s are typically forthcoming goes no way toward showing that the same holds when all these restrictions are dropped. Consequently, there seems little reason to worry that my proposal to spell out the physical in terms of the non-mental (as specifie d in the previous section) either renders physicalism indirectly liable to an analogue of Hempels dilemma, or damages the epistemic standing of the physicalist stance by leading to an unnecessarily stronger, and hence more difficult to establish, causal closure principle. 74


A Generalized Derivability Construal of the Ga p and an Account of the Gap Intuition The purpose of the current section is to eluc idate the nature of bot h the explanatory gap and the gap intuition. Recall that on the construal of the physical that I have put forward, a physical property is characterized as follows: (1) A property is physical just in case either there is a hyperstable phenomenal-free, intentional-free predicate in L that picks it out and which is not functional, or it is logically/metaphysically necessi tated by such properties. Here is how I propose we should understand the notion of a basic physical property: (2) A basic physical property is a property such that there is a hyperstable phenomenal-free, intentional-free term in L that picks it out, and which is not logically/metaphysically necessitated by any other such properties. We are now in a position to formulate a deriva bility construal of the explanatory gap that does not depend on any contingent assumptions about the relationship either between microphysics and physics or between physics and the special na tural sciences: (3) The phenomenal generates an explan atory gap if and only if phenomenal properties do not logically superv ene on the basic physical properties (where the notion of a physical and that of a ba sic physical property are understood as delineated in (1) and (2), respectively.) We can at this point also offer a more specific characterization of what it is for the phenomenal to be ontologically fundamental: (4) Phenomenal properties are ontologically fundamental if and onl y if they are not logically/metaphysically necessitated by the (basic) physical properties. 75


Could it be that (3) still build s in assumptions which, to atta in a fully general account of the gap, have yet to be identified and eliminat ed? Consider the followi ng line of argument. Arguably, purely phenomenal properties are intrinsic and, in addition, satisfy (5) below. (5) For any intrinsic property P, there is in L a predicate, T, such that T is hyperstable and T picks out P. An argument can be put forth, whic h purports to show that the only intrinsic properties that could be expressed by hyperstable terms are phenomenal properties. Suppose all th eoretical terms have their meaning exhausted by the role they play in th e theories in which they are introduced. If that is right, it seems plausible to conclude that natural sciences tell us nothing about the intrinsic nature of the basic properties the existence of which they postulate.28 One possibility is that those basic properties are in f act phenomenal. In that case, since phenomenal properties are expressible by hyperstable terms, and any truth statable using only pheno menal-free (and logical) vocabulary is entailed/metaphysically necessitated by the set of all truths formulated using pure phenomenal predicates, it is only the phe nomenal properties that are fundamen only tal. The alternative case to consid er is that the basic properties postulated by natural sciences are not phenomenal. It seems that there are, th en, two different cases to consider. On the one hand, suppose that the intrinsic properties that co nstitute the subject matter of natural sciences are in principle expressible by hyperstable terms. Even though our ordinary and scientific languages do not include such terms at the moment such terms could in principle be introduced, and the corresponding concepts could be acquired. (Or, even if beyond our cognitive capabilities, this is attainable by some possi ble cognizer.) This option presents no problems for either (1) or (3), as the intrinsic propertie s that (at the moment) we ar e supposed to know about only 28 The view that physics fails to give us any clues about the intrinsic properties of matter has probably been on the table since (Russell, 1927). 76


relationally, but which are in pr incip le graspable, count on my account as physical. On the other hand, however, suppose that the intrinsic properties at is sue are cognitively inaccessible to any possible thinker. As such, they are not expressible via hyperstable terms from the ideal language L and, therefore, do not count as physical accordi ng to (1). If that is the case, phenomenal properties can be ontologically fundamental or not, depending on whether they are something over and above this level of myst erious intrinsic properties. If th e view that the phenomenal is constituted by more fundamental, ungraspable pr operties is coherent, then (4) needs to be amended in the following way, to accommodate such a possibility: (6) Phenomenal properties are ontologically f undamental if and only if either a) there are physical properties, and phenomenal properties are not logically/metaphysically n ecessitated by the (b asic) physical properties, or b) there are no physical properti es (for the reasons delin eated), and the phenomenal properties are not logically/metaphysi cally necessitated by the intrinsic ungraspable properties that the natural sciences pick out merely relationally. Even though, as I will argue in Chapter 5, I do not think that the phenomenal could be constituted by something that is fundamentally non-phenomenal, whether physical in nature or mysteriously non-graspable, the view that the phenomenal is reducible to intrinsic unknowable properties (which reminds one of the position la beled neutral monism) should not be precluded by the very definition of what it is for the phenomenal to be ontologically fundamental. If phenomenal properties are the only intrinsi c properties, or if the central terms of natural sciences in fact pick out intrinsic ungraspable pr operties, it follows that nothing satisfies (1), and hence my account of the physical entails that there are, strictly speaking, no physical properties. One interesting question that this raises is whether, if there are no physical properties, 77


it would still m ake sense to talk about consciousness generating an explanatory gap. Arguably, it would. It is noteworthy that the ga p intuition arises independently of whether the properties that the fundamental terms of physics, chemistry or biology turn out to pick out are intrinsic nonphenomenal properties, or phenomenal propert ies, or intrinsic unknowable properties. Accordingly, if the explanatory gap is underst ood as closely tied to the gap intuition (as it probably ought to be understood), then it should be construed in such a way as to accommodate the absence of physical properties. What this discussion brings out is the need for a further generalization in our search for the right construal of the gap. The explanatory ga p consists in the lack of a certain type of explanation of the phenomenal in non-phenomenal te rms, and the presence of an explanatory gap is compatible with the absence of physical properties in the strict sense of the term. In what follows I will provide a sketch of a general account of the explanatory gap, which is applicable whether or not there are physical properties. If th ere are physical properties, then the explanatory gap can be characterized in a more specific way in terms of failure of logical supervenience on the domain of basic physical properties, as I have suggested at the beginni ng of this section: (7) Let C be the collection of phenomenal-free predicates in L such that they pick out properties that are not logically necessitated by mo re basic properties which are also picked out by phenome nal-free predicates in L. Let D include all the true atomic sentences in L that specify exactly which objects satisfy the predicates in C plus statements of all the laws of nature that are e xpressible using only predicates from C (in addition to logical and modal terms). For any phenomenalfree predicate not in C, any truth to the e ffect that some object satisfies it at t is entailed by D. Likewise, any true id entity statement/necessity statement 78


for mulated only using phenomenal-free terms is deducible from D. The explanatory gap consists in the fact that parallel statements involving phenomenal terms such as S is in pain at t, pain is the firing of C-fibers or the relevant necessity statements are by contrast not so deducible. If there are physical properties, a more spec ific characterization can be given of what the gap phenomenon comes down to, as the derivation base will consist this time of all the basic physical properties: (8) Let A be the collection of phenomena l-free predicates in L that pick out basic physical properties. (This will be a specific subset of the set of hyperstable phenomenal-free, intentional-free terms from L.) Let B include all the atomic sentences in L that indicate which objects sa tisfy the predicates in A, as well as all the laws of nature that can be formul ated only using predicates from A (in addition to logical and modal terms). For any phenomenal-free predicate not in A, any truth to the effect that some object satisfies it at t is entailed by B. Likewise, any true identity statement/necessity statement formulated only using phenomenal-free terms is deducible from B. The explanatory gap c onsists in the fact that corresponding statements involving phenomenal terms are not derivable from B. Let us now address ourselves to the question regarding the sources of the gap intuition. It is quite clear that mere lack of derivability from D is not enough to elicit a gap intuition. False identity statements that contai n only phenomenal-free terms (such as the liquid in rivers and lakes is acetic acid), or false statements about particulars satisfying a certain phenomenal-free 79


predicate (e.g., a is a sp here, when a is in fact a cube), even though not deriva ble from D, are not a sim ilar source of puzzlement. Phenomenal statements (e.g., a is in pain at t) have the interesting feature that they are not deducible from any conceivably true D in L. One could propose, then, that the gap intuition is grounded in the special property of phenomenal c oncepts that is responsib le for this lack of derivability from any conceivabl y true D. Arguably, that property is that phenomenal concepts have basic constituents that are not among the at omic building blocks of any of the concepts expressible by predicates in C.29 In particular, the concept of consciousness looks like an ingredient of all phenomenal concepts, and though perhaps not itself basi c, it is clearly not homogeneous with any of the basic constituents of the concepts expressible by predicates from C. Nevertheless, this suggestion is not adequate either because it predicts that phenomena like water and temperature should present the same sort of puzzle as consciousness. The explanatory gap and the gap intuition are grounded in distinct but closely connected properties of phenomenal terms. What generate s the former is that phenomenal terms go beyond the conceptual resources of D. That s something that holds for water or temperature too, to the extent that the concepts expressed by them contain phenomenal constituents. What is responsible for the gap intuition is that factor in conjunction with the further feature that phenomenal terms are hyperstable. In particular being hyperstable rules out being a natural kind term. Unlike the concept of water or temperat ure, phenomenal concepts are not natural kind concepts. That is to say, by contrast with water or temperature they do not purport to pick out 29 I am assuming here that P fails to entail Q only if Q contains a term such that the concept expressed by it cannot be analyzed in terms of the basic building blocks of the concepts expressed by the terms from P. One might worry that the following is a counterexample to this assumption: a is an experience as of red entails a is not an experience as of green, and yet, arguably, the concept of green cannot be analyzed in terms of the basic components of the concept of red. I suspect, however, that the use of negation is key here. While there appear to exist pairs of basic concepts that cannot apply to the same thing (at the same time) on pain of contradiction, I doubt that there can be two distinct basic concepts such that it is analytic that if an object falls under one, then it falls under the other. 80


som e underlying nature via a set of superficial propert ies. That is the reason why, when the physicalist tells us, for example, that pain is identical to the firing of C-fibers, we do not understand how that could possibly tu rn out to be true. Just on th e basis of grasping the concept of pain, we are confident that there is no hidden na ture of pain that only empirical investigation could uncover. The nature of pain seems to be re vealed to us simply by virtue of possessing the concept of pain, and therefore we strongly beli eve that no surprises c ould be awaiting us. One could extrapolate these considerations and offer the following account of what it is in general for a set of properties, X, as referred to via a certain set of pr edicates, Z, to generate an explanatory gap, as well as a gap intuition, relativ e to a set of properties, Y, as referred to via a set of predicates, U: (9) A set of properties, X, referred to via a set of predicates, Z, generates an explanatory gap relative to a set of properties, Y, referred to via a set of predicates, U if and only if a complete Zdescription of the distribution of the X properties in conjunction w ith the laws of nature whose formulation does not require using any non-logical, non-modal terms not in Z fa ils to entail a complete U-description of the distribution of the Y properties. Further, the X properties, as referred to via the terms in Z, also elicit a gap intuition if and only if the predicates in Z are hyperstable. 81


82 CHAPTER 4 EXPLANATORY POWER CONSTRUALS OF THE GAP In Chapter 4, I present and exam ine the pros pects for two construals of the gap that belong to the explanatory power family and which are suggested by certain remarks on the gap that J. Levine makes at two different stages in his thinking about this problem. In the first section, I distinguish between an early no-causal-role-concept account of the gap (NCC), and a more recent thick-concept account (TC). NCC is a version of what Levine calls in Purple Haze ascriptivist non-exceptionalism, wh ile TC is a type of non-a scriptivist non-exceptionalism.1 (The two labels will be explained in due course.) In the second section of the chapter, the two Levine inspired accounts are examined critically. A key criticism that I level against NCC is that, even if we grant the (highly dubious) conception of scientific reduction that NCC rests on, and, in particular, the underlying view of the justifi cation of theoretical identifications, NCC is a highly unstable position on the nature of the gap. TC will prove to rest on non-ascriptivism about the content of macrophysical concepts, and my ma in objection against TC is that conceptual nonascriptivism is a highly problematic view. While in the second section I ar ticulate this objection and indicate the broad outlines of my argument in s upport of it, it will be the object of the last section to substantiate it. Two Accounts of the Gap Inspired by J. Levine: NCC and TC Since Levine isnt a particular ly rigorous writer, and the pi cture of the gap that em erges from some of his papers might even be suspected of incoherence, I wish to evade the question whether these are really accounts of the gap that at some time or other he would have recognized as his own. Instead, for these as well as other re asons, it is better to ultimately think of them as construals of the gap inspired by Levine. 1 (Levine, 2001, pp. 50-52).


Levines conception of the explanatory gap seems to have undergone significant changes over the years. While there are a few important recurring themes in his thinking about the gap, there are also important respects in which he changed his mind. Among the common elements there are the idea that a theoreti cal identification is justified by its explanatory potential, the emphasis on the link between explanation of macroscopic phenomena and deduction, and an attempt to draw a sharp contrast between ordina ry theoretical identifica tions and psychophysical ones by developing a certain picture about the justificati on of the former, and then arguing that this picture could not fit the psychophysical cas e because of a marked dissimilarity between phenomenal concepts and ordinary macrosc opic concepts like water or heat. One significant change in his thinking has to do with the fact that in more recent writings he subscribes to a version of non-ascriptivism about conceptual content. This is bound up with an important difference in his conception of th e contrast between phe nomenal concepts and macroscopic concepts that unproblematica lly support theoretical identifications. Let us take a closer look at the two accounts of the gap insp ired by Levine, i.e., NCC and TC. If we focus on the justificati on of ordinary theoretical identific ations, and we bracket the fact that the macrophysical concepts involved may have phenomenal components, the justification consists, according to a derivability view of the ga p such as Chalmerss, in the deducibility of the identity statement, roughly, from a full descri ption of the microphysical. It is tempting to interpret some of the claims Levine makes (at least from (L evine, 1993) on) as an implicit rejection of this model, and as suggesting inst ead that familiar theoretic al identifications are justified by their explanatory power. Even if water is H2O is not itself derivable from the microphysical, it is warranted in virtue of its ex planatory potential: assuming it is true enables us to explain the macrophysical properties of water. The fact that these macroscopic properties are 83


instantiated under such and such ci rcum stances is explained or rende red intelligible because it is derivable from a full description of the microphysical (P) plus bridge principles like water is H2O. The bridge principles are required in orde r to connect the micro and macro vocabularies, and their acceptance is justified to the extent that they license this type of derivation of macroscopic truths.2 This much both NCC and TC share in co mmon, as well as the general tenet that one cannot tell a similar story about psychophysical identities becaus e theres a crucial difference between phenomenal concep ts and concepts like water or heat One important difference between NCC and TC is that NCC rests on ascr iptivism about the macroscopic concepts that support scientific reductionsuch as water, heat, etc., while TC is based on non-ascriptivism about these concepts. Another, not unrelated, di fference is that according to NCC, the main source of the epistemic asymmetry between ordinary theoretical identifications and psychophysical ones is that phenomenal concepts are not causal role conc epts, while TC has it that the epistemic asymmetry derives from the peculiar thickness of phenomenal concepts. Certain remarks (Levine, 1993) makes a bout the gap suggest he might have been attracted to NCC at the time.3 TC is inspired by Levines reflections on the gap in Purple Haze. It should be noted that NCC is a version of what Levine calls in Purple Haze ascriptivist nonexceptionalism, while TC is a brand of non-ascriptivist non-exceptionalism. 2 One disconcerting aspect of discussions of the explanatory gap is that it seems not all participants mean the same by microphysics. I will work under the assumption that Levine would take the gist of what he says about the micro-macro relationship as he understands the micro level to hold true even if one thought of microphysics as Chalmers does. 3 It should be noted that some claims Levine makes in On Leaving out What Its Like are not compatible with NCC, nor more generally with an explanatory power model of scientific reduction, and sound more in line with a derivability account of the gap. The prob lematic passages are those where Levine spells out what he deems to be the intimate link between the existence in the psychophysical case of an explanatory gap and the persistence of a certain type of conceivability intuitions. In fact, the relationshi p between his conception of the gap and the knowledge and the zombie arguments is, I think, systematically mischaracterized throughout his writings, except for some places in Purple Haze which is not, however, completely free of confusing remarks on this topic. 84


Levine presents in Purple Haze two possible responses on behalf of the type B materialist to the argument about the conceivability of zombies (i.e., creatu res that are physically indiscernible from us but lack consciousness): exceptionalism (E) and non-exceptionalism (NE). The Exceptionalist4 grants that there is a di sanalogy between water and pa in in that truths about water are derivable fr om microphysical truths5 while truths about ones pain are not derivable from truths about oneself expressed in physic al terms, and so accepts that zombies are conceivable, and denies the conceivability of zombie-H2O (i.e., H2O that is not water assuming that all the microphysical facts are the same as in actual world). According to NE, zombies are indeed conceivable, but so also is zombie-H2O. There is an ascriptiv ist and a non-ascriptivist brand of NE. In the book, Levine professes his allegiance to the non-ascriptivist version, but he emphasizes that a non-exceptionalist can still op pose the conceivability argument even if she would rather resist non-ascriptiv ism, and characterizes the ascr iptivist variety of NE as a plausible position (Levine, 2001, p. 66). Ascriptivist NE entails that macroscopic concepts do have ascriptive conten ts (e.g., it is a priori that water is the watery stuff around here), but despite that, familiar theoretic identifications are not derivable from the micr ophysical because there is a radical disconnect between the macrophysical and the microphysical vocabulary. If a po ssible world is described in macroscopic terms excluding water and its cognat es, one is able to determine what the English word water as used by a speaker in that world picks out there. However, the same is supposed not to be the case if the world is descri bed in microphysical terms. water is H2O is justified by 4 In (Levine, 1998), the philo sopher sounds exceptionalist. 5 Following Levine, I am idealizing away from the fact that macroscopic c oncepts like water have phenomenal constituents. 85


its exp lanatory power, i.e., because it enables us to explain the macrophysical properties of water in a special fashion. A Closer Look at the Two Acco unts and their Shortcomings Heres the sense in which the phenomenal generates an explanatory gap according to NCC. While ordinary theoretical identifications are fully explanatory (e.g., supposing that water is H2O explains all that needs to be explained), there is a ga p in the explanatory import of psychophysical identifications6: the assumption that qualia are identical with such and such physical properties leaves unexplained the ve ry qualitative character of experience. A natural way to try to flesh out this thought is the following. Let a be a state of pain, and F a phenomenal feature of a. The complaint would then be that from a full description of the microphysical plus the claim that pa in is the firing of C-fibers, one cannot deduce that a has F. To assess this proposal it is useful to take a closer look at the reduction of water to H2O. It is important to note that ty pically other bridge principles in addition to water is H2O are needed, according to Levine, to get from micr ophysics to statements about the macroscopic properties of water. For example, that water boils at 212F is only derivable from P in conjunction with water is H2O and some parallel statement about boiling: P & water is H2O & boiling is being G water boils at 212 F (G belongs to the vocabulary of microphysics). boiling is being G is ju stified in a parallel manner, i.e., by its explanatory power: together with other suitable bridge principles, it enables us to get from the microphysical to true statem ents about the macroscopic properties of boiling. Going back to the conditiona l about pain, one will wonder w hy this is not just a case where we simply need another bridge principle to arrive at a has F? P & (pain is the firing of C6 This formulation is from (Levine, 1983, p. 357). 86


fibers) only im plies that a is a pain state. To be able to derive a has F, another bridge principle of the form being F is being H is required, wh ere H belongs to the vocabulary of microphysics. Here is what Levine says about what is at bottom responsible for the explanatory asymmetry: The picture of theoretical re duction and explanation that emerges is of roughly the following form. Our concepts of substances a nd properties like water and liquidity can be thought of as representations of nodes in a ne twork of causal relati ons, each node itself capable of further reduction to yet another ne twork, until we get down to the fundamental causal determinants of nature. We get bottom-up necessity, and thereby explanatory force, from the identification of the macroproperties with the microproperties because the network of causal relations cons titutive of the micro level realizes the network of causal relations constitutive of the macro level. Any concept that can be analyzed in this way will yield to explanatory reduc tion. (Levine, 1993, p. 551) Levine argues that what renders phenomenal concepts not suitable for explanatory reduction and so engenders an explanatory gap in the case of psychophysical identifica tions is the fact that phenomenal concepts do not expres s causal roles (i.e., that they do not refer to the property they pick out as the property that is causally relate d in specific ways to certain other properties): What seems to be responsible for the explanat ory gap is the fact that our concepts of qualitative character do not repr esent, at least in terms of their psychological contents, causal roles. () [T]o the extent that there is an element in our concept of qualitative character that is not captured by features of its causal role, to that ex tent it will escape the explanatory net of a physicalistic reduction. (Levine, 1993, p. 553) What these remarks suggest is that if a micromacro bridge principle is to be adequately justified (i.e., fully explanatory), the macr oscopic concept it contains must express a causal role. The important question that arises is whether th e view that scientific reduction is justified not by deducing the identity statement from a lower level description of the facts but rather by explanatory fruitfulness (understood as above) is really compatible with the claim that adequately warranted bridge pr inciples are bound to employ causa l role concepts. This question must be answered in the negative. An explan atory-power model of sc ientific reduction that 87


incorporates ascriptivism about m acroscopic concepts must re st on the assumption that there are macroscopic concepts that are not causal role concepts, lack c onceptual ties to microphysical concepts, but still turn out to pick out lower-le vel state. The tenet that ascriptive macroscopic concepts which neither expr ess causal roles nor are conceptually homogeneous with microphysical concepts can nevertheless pick out lower-level states must be backed up by an espouser of NCC through an argument to the best explanation: supposing that such a concept picks out a certain microphysical state is justifie d because, when added to a full description of the microphysical, it enables us to derive the re levant collection of m acroscopic truths. That NCC does indeed entail that there must be macroscopic concepts possess ing all the features I have indicated is made clear by the following argument: Suppose all macrophysical concepts are ascriptive, and express causal roles. If all the basic concepts that these concepts are built out of were also among the building blocks of micr ophysical concepts, scientific reduction would always be underwritten by entailment, for me rely building causal ro le concepts out of constituents of microphysical con cepts could not possibly generate a derivability failure, if the relevant macro-micro identity statements are true in the first place. Thus, NCC proves to be a highly unstable c onception of the gap. It combines a certain view about scientific reduction with an alle ged explanation of why phenomenal terms do not support scientific reduction. However, it turns ou t that the underlying view about scientific reduction can only be true if the purported acco unt of the exceptional character of phenomenal concepts is mistaken. Since the truth of the form er requires that there be basic concepts that support reduction, it cannot be that the reason why phenomenal states are not amenable to reduction is that phenomenal concepts are not causal role concepts. 88


Consequently, even granting the adequacy of the ascriptivist explan atory power model of scientific reduction (which is already too much to grant), NC C proves unable to capture the epistemic asymmetry it was intended to. An additional problem with NCC stems from the fact that the picture of scientific reduction entailed by ascriptivist non-exceptionalism is implausible. There do not seem to be any good reasons for an ascriptiv ist to resist the view that true th eoretical identifica tions are entailed by a complete description of the distribution of basic physical prope rties together with statements that express the laws governing thos e, of course, if we abstract away from the fact that some macrophysical concepts have phenomenal constitu ents. Levine suggests it is implausible to suppose that macrophysical truths are deri vable from a full description of the microphysical The view that exhaustive information about the behavior of microphysic al entities conclusively fixes all consciousness-free macroscopic goings-on is upheld in (Chalmers, 2001). If it might seem problematic to maintain that a full description of the world in quantum mechanical terms entails, for instance, theoretical identifications such as water is H2O, the weaker claim that water is H2O is deducible from a full description of the universe couched in quantum mechanical and chemical vocabulary looks hard to object to, if conceptual ascriptivis m lies in the background. Suppose the concept of water is th e concept of the watery stuff around here, i.e, roughly of the clear, drinkable liquid to be found in the rivers, lakes, etc., from our environment. Given that, and bracketing the phenomenal aspects the concep t of water specifies, it sounds quite plausible to say that complete information about the distribution and chemical properties of the compound H2O, along with information about the chemical effects of H2O on our bodies entails that H2O is water.7 Pointing out that, in fact, on a more ge neral conception of scie ntific reduction, the 7 If biological aspects are relevant, an d it is suspected that they are not dedu cible from microphysics plus chemistry, they can be simply added to the supervenience base. 89


supervenience base consists in all the funda ment al physical properties, rather than just the microphysical ones, makes it clear that ascriptivist non-exceptiona lism is highly undermotivated. Turning now to TC, it entails th at the reduction of water to H2O is justified by its enabling us to explain the superf icial properties of water (such as for instance, its boiling point) via a priori conditionals of the form: P & water = H2O & boiling = being F water boils at 212 F The other bridge principle, Boiling is being F, is justified in a similar manner. Eventually, the reduction of macrophysical to microphysical properti es is justified holistically, in the sense that there is assumed to be a very co mprehensive set of bridge principl es which is justified as a whole in virtue of being the unique such set that enables us to account for the entire network of privileged beliefs about the macrophysical (or th e entire network of true beliefs about the macrophysical) via a priori conditionals of the form specified. This much TC shares with NCC. One difference between them is that, unlike NCC, TC rests on non-ascriptivism about macrophysical concepts. One objection that could be raised against TC is that, on a non-ascriptivist semantics of macrophysical terms, there is no gua rantee that the conditions laid out by Levine are sufficient to single out a unique set of bridge principles. The only constraint imposed by this model on what macro-micro mapping counts as adequate is that the micro phenomena that one identifies the macro phenomena with must be cau sally interconnected in certain ways, while nothing is said about the intrinsic nature of the states that macro-physical term s can turn out to pick out. This suggests that there is no principl ed reason why we should expect to find only one set of bridge principles that satisfies the requirements imposed, which in turn casts doubt on the model of reductive explanation that TC relies on. 90


Be that as it may, the major liability of TC is that it appe als to conceptual nonascriptivism about macrophysical co ncepts to account for the epistemic asymmetry between the macro-physical and consciousness. I believe its unfortunate that Levine does not draw a sharp distinction between ascriptivism about linguistic expressions and ascriptivism about conceptual content, but rather gives the impr ession that they are a package deal. It is hardly the case that ascriptivism about conceptual content and as criptivism about linguistic expressions are inseparable. In fact, as soon as it becomes clear that ascriptivism versus non-ascriptivism about items in the language is a false dichotomy, much of the motivation for embracing nonascriptivism about conceptual content vanishes. The considerations typically adduced in favor of non-ascriptivismboth linguistic and conceptualpertain to the work of Kripke and Pu tnam. In the next secti on, I will argue that the arguments provided by Kripke and Putnam about the semantics of expressions like proper names and natural kind terms fail to establish linguist ic non-ascriptivism as Levine understands it. (Searles criticisms of Kripke from (Searle, 1997) and (Searle, 1999) are quite illuminating in this connection.) The claim that there are no modes of presentation associated with proper names and natural kind terms as a matter of semantics is compatible with rejecting the causal historical account. As Searle ar gues with regard to proper names, pace Kripke, what fixes their extension is not a matter of cer tain brute causal relationships obtaining, but rather some intentional content in the speaker s mind, i.e., a mode of presentation the speaker associates with the term and which uniquely identifies the object(s ) picked out. The mode of presentation need not be the same across the entire linguistic comm unity, or single out the object via a property of it that is in any sense remarkable. 91


In fairness to Levine, it shoul d be mentioned that, while im pressed with what he term s the Kripke-Putnam revolution (Levine, 2001, p. 54), he emphasizes also Quinean scruples as part of his motivation to endor se non-ascriptivism. It is qui te mysterious, however, why he deems such scruples perfectly compatible with ascriptivism about pheno menal concepts. It is worth emphasizing that, as interpreted, TC e ssentially relies on the claim that phenomenal concepts, in contrast to macrophysical concepts, ar e ascriptive, for that is what is intended to explain the epistemic asymmetry between consciousness and the macrophysical. Another curious aspect is that Levine maintains he cannot accept that the only a priori truths are the logical ones, because de re necessities come out brute on that view. Therefore, he is willing to make the concession that certain concepts bu ild in the proviso that whatever falls under them does so in all the possible worlds in which it exists (Levine, 20 01, p. 43). It is far from clear why he deems this concession compatible with the Quinean scruples that have contribute d to his shunning fullblown ascriptivism. Recall that TC purports to e xplain the epistemic asymmetry by appealing to a contrast between thick and thin concepts, an d I have interpreted the thick-thin distinction as equivalent to the contrast ascriptive vs non-ascriptive. At times Levine does sound as though that is precisely the sense in which phenomenal conc epts are thick: the y, unlike ordinary macroscopic concepts, have ascriptive content. It is fair to menti on, however, that some rath er cryptic remarks that Levine makes about the connection between th e thickness of phenomenal concepts and the subjectivity of phenomenal expe rience or the special epistemic access we have to our conscious states engender the suspicion that the thickness of phenomenal concepts might in his view go beyond possessing ascriptive content, having to do with th e special epistemic properties of qualia. Levine sometimes sounds as though consciousness genera ted in a sense two 92


explanatory gaps, one of which m ight turn out to be more fundamen tal than the other: there is a first-order gap for which it is phenomenal features themselves that are responsible, but there also seems to be a second order gap, as it were, whic h is generated by the special nature of our epistemic access to our phenomenal states, or as he sometimes puts it, by the subjectivity of conscious experience. While Levines positi on on the relationship between the gap and the special epistemic access we have to our current me ntal states remains hard to pin down, Sturgeon developed in some detail an approach to the ga p that postulates an esse ntial connection between the two. Sturgeons proposal to und erstand the gap as a consequence of the special epistemic role of phenomenal concepts, which he employs in an attempt to block the argu ment from the gap to dualism, will be examined and evaluated in Chapter 5. The Epistemic Credentials of Conceptual Non-Ascriptivism Physicalists are not exactly famous for pain stakingly arguing for the semantic views on which the plausibility of their metaphysics often depends. For instance, it is not unusual for them to embrace a version of NA about linguistic expr essions and/or conceptual content on the grounds that Kripke and Putnam have shown it to be right. Recall that, according to linguistic A, whats responsible for fixing the extension of a term is a mode of presentation awareness of which is guaranteed by semantic competence. (The mode of presentation must thus be the same for all competent speakers.) Conceptual ascriptiv ism is the view that what determines the extension of a concept is a mode of presentation thats cognitively accessible to the subject merely in virtue of his grasping the concept. Linguistic NA entails th at the extension fixing mechanism for linguistic expressions essentially involves external factor s that the competent speaker need know nothing about, such as causal re lations with the environment. Conceptual NA is the parallel thesis about what fi xes the extension of concepts. 93


The relevance of NA to the explanatory gap is two-fold. First, as we have seen, there are construals of the gap that attempt to account fo r the epistemic asymmetry between consciousness and the macro-physical by claiming that, whil e phenomenal concepts are ascriptive, macrophysical concepts which support reductive explanation (such as water gold temperature etc.) are non-ascriptive. (This is the case of what I labeled the thick c oncept construal of the gap inspired by Levine.) Second, the discussion of physicalist res ponses to the gap problem from Chapter 5 will reveal that some physicalists ar e inclined to counter the gap argument by arguing that, although consciousness may not be derivable from basic physical facts, that has no metaphysical significance because neither are familiar macrophysical facts about water, temperature, etc. The lack of derivability cl aim about macrophysical phenomena appears to be justified by an appeal to NA. In bot h cases, it is conceptual, rather than linguistic, A that is relied upon.8 It appears that the effect of the arguments proposed by Kripke and Putnam was to create a presumption against the otherwise quite intuitive stance that the mechanisms of reference fixing for both linguistic expressions a nd concepts involve possession of a uniquely identifying mode of presentation of the referent, and in favor of the less intuitive view that reference fixing essentially depends on factors that may be beyon d the individuals ken. Levine, for instance, writes: [e]ver since the Kripke-Putnam revolutio n, its been taken for granted that internal meaning alone cannot determine extension [] [W]het her it still has a substant ive role to play in determining extension is still a matter of debat e (Levine, 2001, p. 54). The legitimacy of this 8 Block and Stalnaker, for instance, would seem to belong here, though what they explicitly reject is linguistic, rather than conceptual A. It is clear, however, that the conceptual variety of A is presupposed by their goals. 94


presum ption seems rarely questioned. My main goal in this section is to make a case that it is in fact ill-founded.9 In the present section, I take a l ook at the arguments advanced by Kripke10 and Putnam11 and argue that they fail to establish the ki nd of NA needed by physicalists to counter the challenges posed by the gap argument, namely c onceptual NA. Kripkes discussion of natural kind terms is modeled on his account of proper name s, under the assumption that there is a deep semantic similarity between the two kinds of expressions. Even though Searle12 offered what I take to be a successful rebuttal of Kripkes causal theory of proper names, both his arguments against Kripke and the version of descriptivism about proper names that Searle defends seem to have received little notice in the literature. In this section, I first review the Kripke-Searle debate about proper names, and then argue that in light of Searles insights, th e objections leveled by Kripke and Putnam against descriptivism about natu ral kind terms can likewise be laid to rest. To bring into sharper focus the import of Kr ipkes semantic views, their bearing on the theses of linguistic and conceptual A and NA, as well as the strate gies available to a conceptual ascriptivist faced with Kripkes arguments, let us distinguish between the following three questions. For simplicity, both the questions and the answers to them that I lay out are phrased in 9 A full-scale defense of conceptual ascriptivism goes beyond the scope of the present study. In this section I focus on the views advanced by Kripke and Putnam, which constitute a major development in natural language semantics with a sweeping influence in many ph ilosophical circles. Considerations of space will prevent the discussion of other views which could also be used to support NA, such as indicator semantics (espoused, for instance, by Dretske and Fodor) and teleological theories (upheld, for example, by Millikan). 10 (Kripke, 1980). 11 (Putnam, 1975). 12 (Searle, 1982). 95


term s of proper names, but parallel questions and answers can be form ulated for natural kind terms.13 1. What does a name contribute to the proposit ion expressed by an utterance containing it? 2. What does competence in the use of a name cons ist in? (In virtue of what do I manage to use the name N as a name for x?) 3. In virtue of what is my thought a thought about object x? Question 3 is the most basic. One possible an swer to 3 is that my thought is about object x because I am in possession of a uniquely identif ying mode of presentation of x. This is what I labeled conceptual ascriptivism. As Searle points out, the mode of presentation in question need not be expressible by any definite descri ption in the language: i n some cases the only identifying description a speaker might have that he associates with the name is simply the ability to recognize the ob ject (Searle, 1999, p. 233). A very di fferent answer would be that my thought is about x in virtue of some external, causal, relatio nship holding between me and x14, which given our Levine inspired nomenclatu re, counts as a species of conceptual nonascriptivism. In response to question 1, one could maintain that the name contri butes to the proposition expressed the referent, or, alterna tively, some mode of presentation of the referent (i.e., a sense). Given how sense is traditionall y understood, if one opts for the latter alternative, semantic competence will require knowing what the associated sense is. As for question 2, on the assumption that one en dorses a direct reference theory of proper names (i.e., the first option mentioned in respons e to question 1), one ca n claim that semantic 13 In what follows I go along with Kripkes treatment of na tural kind terms as directly referring terms that pick out natural kinds. 14 There is room for variation here, as there may be r easons to impose instead the requirement that a causal relationship hold between the speaker and the initial baptism, or the gradual process by which the name got introduced, or the event of introducing an entire system of names. 96


com petence typically requires in the case of prope r names that the speaker have an associated mode of presentation that picks out the referent uniquely. Anothe r possibility is to say that semantic competence is rather a matter of a certain causal relationship holding between the speaker and the referent.15 If to the former option the proviso is added that the mode of presentation is the same across the entire comm unity of competent speakers, we get linguistic ascriptivism. The second option is a version of linguistic non-ascriptivism that features a socalled causal-historical chain. It is crucial to no te that linguistic ascriptivism and non-ascriptivism, while mutually exclusive, are not contradictories. There is a third position in logical space according to which, even though a uniquely identifying mode of presentation is require d for semantic competence, it need not be the same for all competent users of the name. Searle can be interpreted as attempting to steer just such a middle c ourse between linguistic A and NA. Note also that conceptual ascriptivism and non-ascriptivism are, by contrast, genuine contradictories. Either possessing a concept requires grasping a uniquely identifying mode of presentation or it does not. (In the case of proper names, on a direct refe rence theory that rejects both linguistic A and linguistic NA, it would be natural to maintain that different co mpetent users of a name may associate different concepts with it.) The existence of a middle ground position betw een linguistic A and NA is of the utmost importance for our goal of evaluating physicalism, because this position is compatible with conceptual A. Therefore, conceptual and ling uistic A are not a package deal, as Levine assumes.16 More specifically, one can be a conceptual ascriptivist while denying the claim built 15 See previous footnote. 16 See, for instance, (Levine, 2001, pp 53-55), where he introduces the term s ascriptivism and non-ascriptivism, and where he writes as though the conceptual and the linguistic variety were interchangeable. 97


into lingu istic A that the uniquely identifying m ode of presentation must be the same for all competent speakers. Searle clearly sees his disa greement with Kripke regarding proper names as one over how to answer question 2. With regard to question 3, S earle explicitly endorses conceptual A. As for Kripke, some of what he says might suggest that he takes himself to ha ve answered not just 2, but also 3, and that he upholds a corresponding non-ascriptivist view of thought content. More importantly, however, quite apart from Kripkes own intentions, his arguments are deemed by many to have established a versi on of conceptual NA as well. Since my ultimate goal is to ascertain whether the physicalists who resort in their response to the gap problem to conceptual NA are right in supposing that its truth was established by Kripke, the following question presents itself. Rega rdless of the intentions that Kripke himself might have had, is it the case that, if we ha d been persuaded by Kripkes arguments for linguistic NA, we would have ha d to thereby adopt conceptual NA as well? It generally looks like a good idea to give parallel answers to questions 2 and 3. For one thing, it is plausible that sema ntic reference is derivative upon mental reference. Couldnt one perhaps argue, though, that even if Kripkes de fense of linguistic NA had been successful, an additional argument would have been needed to bridge the gap between language and thought? A view that requires modes of presentation for refe rence in thought but not for linguistic reference makes it much easier, for instance, to make statements about tige rs than to th ink about them. Therefore, on a mixed view of this sort, it w ould be possible for a speaker to talk about tigers using tiger, while being unable to think about th em, on the grounds that he stands in the right causal relationship to tigers, but associates no uniquely identifying mode of presen tation with the token of tiger that is part of his statement. (To his statements there corresponds just an 98


incom plete thought.) As we have here a hard ly desirable consequence, a combination of linguistic NA and conceptual A does not look tenable. This is a significant result because it means that, irrespective of wh ether Kripke intended his causal account of reference to apply not just to language, but also to thought, if we do not want to be stuck with conceptual NA, we have no other option but to re but Kripkes arguments for linguistic NA. To anticipate, th e gist of the criticism Kripke w ill be faced with is that, even though successful in refuting linguis tic A, his arguments fail to establish linguistic NA, because they fail to block the middle ground position we have identified between the two. I turn now to examining the controversy betw een Kripke and Searle over the semantics of proper names. That will enable us to draw some instructive lessons which will shed new light on the case of natural kind terms. The descriptive17 theories of names that Kri pke sets out to dismantle in Naming and Necessity fall into two categories: on the one ha nd, the synonymy view, according to which a proper name is synonymous with a definite description (Frege and Russell are often thought to belong here), or with a cluster of descriptions,18 and on the other hand, the reference fixing view (endorsed by the early Searle19), according to which a proper name has its reference fixed by a definite description or by a family of descriptions. Kripkes official statement of the synonymy view includes the following theses: 17 Descriptivism is typically understood as a view of the semantics of linguistic expressions. It comes in a synonymy and a reference fixing variety and traditionally bu ilds in two assumptions: that the mode of presentation is unique across the community of competent sp eakers, and that it can be expressed vi a a definite description. It is the latter assumption that distinguishes it from linguistic NA, as I am using the term. 18 To demystify talk of the name expressing a cluster concept, or being synonymous with a family of descriptions, we can take that to mean that the name is synonymous with a definite description, only of a slightly different kind than it was assumed traditionally: a definite description of the form the object that possesses most/enough of the following properties ... 19 (Searle, 1997). 99

PAGE 100

(1) To every nam e or designating expression X there corresponds a cl uster of properties, namely the family of those properties such that A believes X. (2) One of the properties, or some conjoint ly, are believed by A to pick out some individual uniquely. (3) If most, or a weighted most, of the s are satisfied by one unique object y, then y is the referent of X. (4) If the vote yields no unique object, X does not refer. (5) The statement, If X exis ts, then X has most of the s is known a priori by the speaker. (6) The statement, If X exis ts, then X has most of the s expresses a n ecessary truth (in the idiolect of the speaker).20 (Kripke, 1980, p. 609) If one discards thesis (6), what remains is the reference fixing view. One of the major objections that Kripke le vels against the synonymy view is the socalled modal argument, which turns on the claim th at names are rigid designators, in the sense that they pick out the same individual with re spect to any counterfactua l scenario (where it exists), while the descriptions usually proposed as synonymous with them express contingent properties of the denotatum and are, therefore, non-rigid. For example, intuitively, Aristotle was fond of dogs is true relative to a counterfact ual circumstance just in case this individual we refer to by Aristotle is fond of dogs in that situation, whethe r he is a teacher there or not. By contrast, if Aristotle is treat ed as synonymous with the teach er of Alexander the Great, the truth of the same statement relative to a counterfactual circumstan ce depends on whether whoever in that world happens to be the teacher of Alexander the Great is fond of dogs, which seems wrong. 20 Kripke includes in fact the further claim that a sema ntic account of proper names passes muster only if the descriptive contents associated with the names are not conducive to vicious circularity. 100

PAGE 101

The cluster version of the synonymy view does not seem to fare any better, as it is contingent that Aristotle had th e inclusive disjunction of the pr operties commonly attributed to him. In response to the modal argument, one can dr op thesis (6) from the list, which results in the reference fixing view of proper names.21 In regard to the referen ce fixing view, Kripke finds thesis (1) harmless, but endeavors to rebut theses (2) through (5). (2) is considered false on the grounds that very often there is no set of properties that could possibly be assumed by the speaker to pick out a unique individual. For inst ance, one can still use Feynman as a name for Feynman, even though all one knows about him is that he is a famous physicist. Kripke offers the following putative counterex ample to thesis (3). The only thing many people know, or think they know, about Gdel is that he discovered the incompleteness of arithmetic. Suppose that in fact Schmidt turns out to be the real discovere r of the incompleteness of arithmetic, while Gdel simply took hold of the manuscript and murdered Schmidt. It is a consequence of thesis (3) that all along it is Schmidt that we have been talking about when we employed the name Gdel, and th at surely sounds implausible. How about the suggestion that the reference fixi ng description for Gdel is in fact the man to whom the incompleteness of arithmetic is commonly attributed or the man who most people think proved it? Kripke rais es two objections against this. The first is that the proposal may give the wrong result. Suppose that the Gdel-Sc hmidt story above is tr ue, and that in fact most people know about this, though the speaker does not. If the reference fixing description for that speaker is the man who mo st people think proved the incomple teness of arithmetic, then he should be referring to Schmidt by his use of Gdel, while we tend to think that he may still be 21 Another possible move is of course to rigidify the original description, but the objections raised by Kripke against the reference fixing view seem to be just as effective against this move. 101

PAGE 102

referring to Gdel. The second criticism is that the suggestion viol ates a non-circularity requirement: it cannot be the case that everybody in the community fixes the referent of Gdel via this description, for if we all attempted to fix it in this way, no referent would be secured. An attempt to avoid circularity that Kripke considers is the passing the buck suggestion: by my use of Gdel I intend to refer to whomever J ohn, whom I picked this name from, intended to refer to, John passes the buck to somebody else, etc. Kripke finds this idea unsatisfactory for two reasons. The first is that one may very well forget whom one got the name from. An interesting question which should be kept in mind for it will be revisited when discussing Searles response to Kripke is whethe r that bit of information is really necessary. Would it not suffice in many cases to just have th e intention to refer to, say, the mathematician (or even the person) whom whoever I got the name Gdel from referred to? Kripkes second cause of dissatisfac tion appears to be more obscure: One has to be very careful that this doesnt co me round in a circle. Is one really sure that this wont happen? If you could be sure yourself of knowing such a chain, and that everyone else in the chain is using the proper co nditions and so isnt getting out of it, then maybe you could get back to the man by referring to such a chain in that way, borrowing the references one by one. However, although in general such chains do exist for a living man, you wont know what the chain is. You wont be sure what descriptions the other man is using, so the thing wont go into a circ le, or whether by appealing to Joe you wont get back to the right man at al l. So you cannot use this as you r identifying description with any confidence. (Kripke, 1980, pp. 621-2) It seems right that on this proposal, if the chain is circular, you have failed to refer, at least if nobody in the chain associates any additiona l descriptive material with the name, but this outcome does not sound especially implausible, not to mention that it coincides with the prediction made by Kripkes own causal view. Why should it be a problem that in many cases you are not able to ensure that the chain is not ci rcular? It is not as though you could be certain in those cases that you managed to use the name successfully. Let us note for now that Kripkes dismissal of the passing the buc k proposal looks questionable. 102

PAGE 103

Other alleged counterexam ples to thesis (3 ) are actual cases in which we use names successfully on the basis of a considerable am ount of misinformation. For instance, the only description that many people who are competent in the use of the name Peano associate with it is the discoverer of such and such axioms in arithmetic. Even though those axioms have actually been discovered by Dedekind, their use of the name Peano does not refer to Dedekind. Against (4) Kripke argues that there can be cases where no object satisfies any significant number of the descriptions in the associated clus ter, and yet we do not think the name is vacuous. For example, if it turns out that the definite description the discoverer of the incompleteness of arithmetic is non-denoting because in fact the p roof that we have involves a mistake, it need not thereby turn out that Gdel is a vacuous name. Kripke rejects thesis (5) because, even wh en If X exists, then X has most of the s is true, it is not known a priori by the speaker.22 We do not know a priori, for example, that the macabre story told above about the putative murd er of the discoverer of the incompleteness of arithmetic is false. According to Kripke, the picture of naming that theses (1)-(5) suggest is that one always introduces a name via a mental ceremony on the basis of some uniquely identifying description. Kripke claims that such cases are in fact very rare. That may apply to the way the police introduced Jack the Ripper to stand for whoever committed most of those murders, or to introducing Hesperus to stand fo r the celestial body visible over there, or to the case of learning somebodys name by meeting him, but this picture is in genera l mistaken. On the alternative picture of naming proposed by Kripke (Kripke 1980, pp. 625-26), a name is introduced in the public language in an initial baptis m, when the reference of the na me is fixed by either ostension 22 I take it that X has most of the s counts as known to be true a priori by the speaker (at time t) just in case it is known to be true by the speaker (at time t) only in virtue of his being able (at t) to use X as a name for X. 103

PAGE 104

or descriptio n. Then a certain speaker is a competent user of that name just in case she is part of a chain of communication of the right sort that reaches back to the referent itself. Thus, she is able to use Feynman as a name for Feynman not because she is in possession of any uniquely identifying description, but because shes at the end of a causal communication chain which reaches back to the referent itself via a number of other speakers, each of whom has the intention to refer to the same thing as th e person he got the name from.23 A number of objections have been raised ag ainst the causal-historical chain account of proper names. While many can be deflected by su ccessive refinements of the theory, one of them, due to Searle, strikes a deadly blow. Be fore moving on to Sear les central objection, a quick look at two of the difficu lties that Kripkes account has the potential to overcome and the amendments they motivate will prove helpful in ou r subsequent discussion of natural kind terms. It was objected that the standards for sema ntic competence set by the causal account are too low (Searle, 1999). Somebody who hears a brief exchange about the philosopher Socrates but confusedly gets the impression that Socrates refers to a number has not thereby become a competent user of Socrates, even though he meets the constrai nts of the unqualified causal account. Arguably, even deferential uses of na mes require some knowledge about the kind of thing referred to. As Kripke co uld in response grant that intenti onal content may be relevant to the fixing of reference, but insist that an exte rnal, causal, element is nonetheless ineliminable, such considerations are not suffi cient to rebut Kripke s causal account. What they do bring to light, however, is that intentional content is at least necessary for semantic competence, even for deferential uses. 23 If there was nothing like an initial baptism, then the role assigned in Kripkes original story to the baptism is going to be played instead by the gradual proce ss whereby the word became the name of an object. 104

PAGE 105

Evans argued24 that the causal communication chain account fails to give sufficient conditions for reference for the following reason. Originally, the name M adagascar referred to a portion of the Afri can mainland. However, Marco Polo, mis understanding the interlocutors that he got the name from, intended to use it to appl y to an island, and today the name Madagascar does refer to that island. Since he also had the inte ntion to refer to the same thing as those he got the name from, as we can suppose anyone else did in the chain leading up to our current uses of the name, the causal account wrongly predicts that our use of Madagascar refers to a portion of the African mainland. To account for the possibility of shifts in reference despite each speaker in the chain having the intention to use the name in the same way as the previous speaker, advocates of the causal account have introduced the notion of multiple grounding of a name (Devitt & Sterelny, 1999, pp. 75-6). While subseque ntly to the initial dubbing many uses of a name are indeed deferential, often some are not. Ra ther, they resemble the initial baptism in that they essentially rely on perceptual encounters with the referent. Under normal circumstances, such uses reinforce the original grounding of the name in its referent. Speakers confusion, however, can cause the grounding pattern to shift, and if the shift gains prevalence in the linguistic community, as we ma y assume happened for Madagascar, reference follows suit. I turn now to succinctly presenting Searles version of a descrip tive theory of proper names. Then, from Searles perspective, I will de velop a criticism of Kripkes arguments against theses (2)-(4). According to Searle, the central cases of names are those for thi ngs we are acquainted with (family members, friends, pl aces one visits often, etc). We usually learn their names from other speakers, but come to associate them with a very rich intentional content. At the other 24 (Evans, 1985). 105

PAGE 106

extrem e there are cases of parasitic (i.e., defere ntial) uses of names, where the speaker may have some associated intentional content, but it is very meager (it specifies, for example, in very general terms what sort of object the referent is). And in between there ar e all sorts of cases in which the referent has not been the object of ac quaintance, but the speaker does associate with the name a more or less rich intentional conten t derived from other speaker s. In general, with respect with names of objects we have never b een acquainted with, the deferential intention is going to be overriding, if it turn s out that there are conflicts among the bits of descriptive material contained in the cluste r the speaker associates with th e name. Even so, Searle suggests that reference may very well fail if the deferentia l intention turns out to be in conflict with the speakers belief that the referent is a certain type of object (for instance, a human being, as opposed to a frog).25 According to Searle, in all these cases, app earances to the contrary notwithstanding, the total intentional content in the speakers mind does uniquely identify the object referred to. Lets consider how Searle would defend theses (2)-(4) from Kripkes criticism. Against thesis (2) Kripke argued that one can use the name Feynman as a name for Feynman even though all one knows about Feynman is that he is some fam ous physicist. Searle would argue, though, that this is an oversimplifica tion. The speaker is even in cases like this in possession of uniquely identifying information, as is revealed by examining the speakers total intentional content, although th e uniquely identifying information may turn out to be less 25 On Searles account of proper names, for some uses of names perceptual modes of presentation are essential to reference fixing. For instance, if a physical thing is given a name via a baptism, those present at the baptism will typically be able to use the name competently in virtue of their possessing a perceptual mode of presentation of the object. It could be objected that this introduces an external elem ent into Searles picture of naming, for there seems to be an important causal factor in determining what a perceptual experience is an experience of. However, according to Searles theory of perceptio n, this is in fact an instance of in tentional causation, because perceptual content builds in a representation of the relevant causal relation. 106

PAGE 107

exciting than what the causal theorist was unj ustifiedly looking for. In the example under consideration we have what looks like a deferen tial use of Feynm an, and surely the speaker must have the intention to refer to the object that the person he got the name from referred to. This suggestion is somewhat similar to the pa ssing-the-buck proposal Kr ipke did consider, only he assumed the proposal would have to require th at the speaker remember whom he got the name from (Kripke, 1980, p. 622). However, there is no reason why that should be so. Kripke mentions the case of somebody who got the name Cicero from Smith, who used it to refer to the famous Roman orator, but later comes to mistakenly believe he got it from Jones. Jones, however, unbeknownst to the speaker uses Cicero to refer to a German spy. In this case its not how the speaker thinks he got the reference, but the actual chain of communication which is relevant (Kripke, 1980, p. 623). That does not ra ise any problems for the descriptivist, though. The speaker may have the in tention to refer to what Jones refers to with Cicero, but he also has the more general intention to refer to th e object that the person he got the name from referred to, and this latter intent ion is overriding, as evidenced by how he would react, were he to discover a conflict between the two. Similar considerations are effective agains t Kripkes criticism of (3) and (4). For instance, it is not at all clear that in the Gdel-Schmidt example a weighted most of the descriptions in the cluster really fit an object diffe rent from the referent. If one pays attention to the total intentional content (including the deferen tial component, which is here crucial), as well as to the actual weighting of its various components, one sees that the descripti on the discoverer of the incompleteness of arithmetic is outranked by the deferential element, and we get the intuitive result. 107

PAGE 108

To summarize, Searle argues that the cau sal account is rooted in exaggerating the importance of parasitic uses of names, and rega rding those from a purely external perspective, which makes one miss the fact that what is cruc ial to reference is not the mere holding of a causal relationship, but rather the holding of it being represented by the mind. It is symptomatic in this respect that, as Searle perceptively puts it, causal theorists generally say suspiciously little about the causal relationship be tween speakers which is supposed to be the central element of their view. Kripke registers a resemblance between proper names and natural kind terms and briefly argues that his results about the prospects of the descrip tive views of proper names and the success of the alternative, causal-historical account extend to natural kind terms. It might be tempting to think that there mu st be a description/cluster of descriptions semantically associated with a term like tiger which specifies the superficial properties that tigers are often identified by and is either synonymous with the term or at least fixes its referent (perhaps something along the lines of large carnivorous quadrupedal fe line, tawny yellow in color with black stripes). Kripke argues agains t a version of (4) regarding natural kind terms that it might turn out that tigers have none of the features enumerated, because all of them are due to optical illusions. Therefore, contrary to (5),26 it is not a priori that tigers have any of these properties. Likewise, since cats might turn out to be robots or dem ons, it is not a priori that cats are animals, and since gold might look yellow to us only because of strange atmospheric conditions, while being in fact blue it is not a priori that gold is yellow. An analogous argument is assumed to work for pretty much any of the supe rficial properties of cats or gold that might be submitted by the opponent. Against (6), even if gold is as a matter of fact yellow, it is certainly 26 In what follows, I will use the designations (1) through ( 6) to refer to versions of those theses formulated for natural kind terms. 108

PAGE 109

not neces sary that it should be so, as there are po ssible worlds in which gold, i.e., the metal with atomic number 79, has the propensity to cause in normal perceivers under normal circumstances a very different type of color experience. What is by contrast, necessary, if true at all, is that gold has the atomic number 79 (hence natural kind terms are also rigid designators), but that is something that could hardly be semantically associated with gold because, first, it was discovered long after the term was introduced into the language, and, second, there is no reason to believe that the discovery determined any change in the terms meaning. In addition, Kripke notes that all the superficial properties we associate with gold are possessed by iron pyrites, too, as we very well know, and yet irons pyri tes does not count as a type of gold. From this he concludes that the fe atures we associate with gold are not sufficient for membership in the kind. Since we are quite fa miliar with the existence of fools gold, this could also be seen as an implicit criticism of (2 ). (A more striking counterexample to (2) is, of course, Putnams famous beech-elm case: not bei ng able to tell an elm from a beech is a very common occurrence among persons whom we nonetheless consider competent users of these terms. More will be said about that below.) As regards (3), although Kripke does not address it, we can imagine a scenario where, while gold is in fact bluealthough it looks to us yellow due to an optical illusioniron pyrites is indeed yellow. In this s cenario, it might look like iron pyrites should count not only as gold but as the only kind of gold. But, intu itively, that is the wrong verdict. These are arguments against theses (2) th rough (6) formulated for natural kind terms which closely parallel those directed against desc riptive theories of prop er names, and the moral Kripke draws from them is that descriptiv e accounts of natural kind terms are beset by difficulties and should give way to a causal-histori cal account. A natural kind term like gold is 109

PAGE 110

introdu ced into the language through an initial baptism in which its re ference is fixed via a description of the form the kind that most of these samples instantiate. Then the term is passed from speaker to speaker along a causal chain of communicati on, which enables persons who have never been presented with members of th e kind and perhaps do not even know much about it to use the term competently. Let us now address the question whether Kr ipkes arguments are a genuine threat to conceptual A about natural kind terms. Causal-historical accounts usually have two parts, which address the following two aspects: on the one hand, the initial fixi ng of reference for the term, when it is first introduced into the language (or, to be more precise, given the need for multiple groundings of the term in its referent, all the in stances of grounding), an d, on the other hand, the borrowing of reference by one speaker from anot her. It is noteworthy that, given how causal theorists approach the task of explaining the two as pects, it is only the latter that appears to cause trouble for conceptual ascriptivism.27 (Causal theorists themselves seem to be looking at the matter this way.28) The initial reference fixing is char acterized as accomplished through either 27 One difficulty that accounts of grounding proposed by causal theorists en counter is the so-called qua problem. The initial reference fixing description the kind that most of these individuals instantiate is not uniquely identifying because those individuals often instantiate several different natural kinds (e .g., mammal, feline, tiger). Interestingly, causal theorists reaction to the problem is to try to build more descriptive content into the mode of presentation, though often, as a resu lt, tension may arise with the upshot of the argument from the epistemic possibility of extensive error. I think they are right to expect the problem to be solvable along descriptivist lines, for it is quite unclear how recourse to causal relations with the samples could discriminate between the different kinds the samples instantiate. (I also suspect that the argument from error may have been carried further than is reasonable. There are some things that could not turn out to be false. For instance, stri ctly speaking, cats couldnt turn out to be robots. For one thing, I think it is built into the concept cat that it refers to a natural kind. Kripke and Putnam argue otherwise, but it seems to me that they are stuck with an implau sible consequence. If cat is a term that picks out some underlying nature that explains a set of superficial propertie s, and it is literally true that cats could turn out to be robots, then we are committed to the fo llowing: if the things we call cats do turn out to be robots, we should call cats also things that share the same robot-like underlying structure but, because of various factors, look nothing like cats, or even like animals for that matter. I believe the correct way to describe the situation is this: if the things we call cats turned out to be ro bots, then we might say things like cats turned out to be robots, and continue to call those thin gs cats, but a shift in meaning would occur, whereby the term ceases to be a natural kind term and begins to track a cluster of superficial properties.) 28 See, for instance, Sterelny (1983). 110

PAGE 111

description or ostension, and if the latter, a per ceptua l mode of presentation is what is actually doing the job. The regrounding of the term in the referent usually occurs through perception, too. Of course, the picture about wh at is going on at the initial dubbing and subsequent regroundings is, for proper names at least, clearly incomp atible with linguistic A because the uniquely specifying mode of presentation operative un der those circumstances looks nothing like the descriptions that linguistic NA might propose as semantically associated with the name. The important point, though, is that as far as conceptual A goes, it is not so much the grounding as the borrowing stage that appears challenging. We saw that the borrowing of referen ce for the case of proper names has been compellingly accounted for along con ceptual ascriptivist lines by Searle. Whenever it might look like the speaker simply knows too little to be in possession of a uniquely identifying mode of presentation of the referent, appearances are de ceptive because she is nonetheless in possession of the description the thing referred to by the person from whom I got th is term. Consider a parallel move for natural kind terms. To use the term tiger competently even if one has never been perceptually presented with tigers (or realistic tiger im ages), it is enough to have the intention to refer to the kind that the person one got the term tiger from referred to (provided of course that that person was herself a competent speaker). When I discussed the semantics of proper names, I noted that semantic compet ence often requires knowing about the referent something a little more substantiv e than that (such as what kind of object the referent is). For natural kind terms this kind of argument seems at least as persua sive, and therefore the proposal under scrutiny sets the bar for semantic competence too low because it ascribes too little content to the reference fixing description. While th at is clearly a problem and the Kripke-Putnam argument for the epistemic possibility of extensiv e error about kinds sugge sts that finding a more 111

PAGE 112

robust, counterexam ple free proposal may be no eas y task for natural kind terms, this is no reason for the conceptual ascriptivist about inst ances of borrowed reference to lose heart. The intuition that, even though the description offered is enough to fix reference, competence requires more descriptive content than that certain ly could hardly motivate abandoning conceptual ascriptivism for some form of causal-historical account. I wish to stress that it is not my goal he re to articulate a full-blown account of the semantics of natural kind terms and show it to be definitive. Rather, what I have attempted to show is just that Kripke offers no good reas ons to abandon conceptual ascriptivism about the concepts expressed by natural kind terms. My objections agai nst Kripke are compatible with a variety of conceptually ascriptivist accounts of the semantics of natural kind terms, all of which differ from Kripkes own largely in that re ference borrowing is always accomplished via a uniquely identifying mode of presentation of th e referent. Depending on how much communitywide error about the kind referred to is deemed epistemically possible, th e modes of presentation associated with uses of borrowed natural kind terms will be construed accordingly as less or more robust. Putnams goal in The Meaning of Meaning is to show that the intrinsic properties of a speaker are not sufficient to fix meaning (sem antic externalism). Just as with Kripkes arguments, Putnams are often taken to establis h a similar point about thought, namely, that the intrinsic properties of a subject do not suffice to determine mental content (content externalism). In Putnams case the move from language to tho ught appears even more straightforward, as one of his famous arguments regarding language is that it is conceivable for there to be two individuals whose psychological states (narrowly construed29) are type-identical but who are 29 A psychological state narrowly construed is a psychological state such that an individual xs being in that state does not entail the existence of any object different from x or any of xs parts. 112

PAGE 113

talking about two different substances, from wh ich he famously concludes that meanings are not in the head. It is very te mpting to go one step further and al so conclude that an individuals psychological state (narrowly construed) fa ils to determine his thought contents. Putnams Twin Earth thought experiment alle gedly shows that causal relations with the environment essentially contribute to reference fixi ng, while his discussion of what he labels the division of linguistic labor purpo rtedly establishes that causal re lations to other speakers from ones linguistic community play a similar role. To briefly rehearse the Twin Earth thought experiment, suppos e that in the actual world there is another planet, Twin Eart h, which differs from Earth as litt le as possible compatibly with accommodating the following difference: the liquid that Twin Earthlings call water has the chemical composition XYZ instead of H2O. XYZ shares absolutely all the superficial properties of wateris clear, odorless, drinkable, is typical ly found on Twin Earth in the same kinds of places as on Earth, etc. The two planets are so similar that every individual on Earth has a qualitatively identical copy on Twin Earth (or something as close as is allowed by the differences between H2O and XYZ). Suppose that Oscar1 is a traveler from Earth who pays a brief visit to Twin Earth at a time when neither Earthlings nor Twin Earthlings ha ve any knowledge of chemistry. Oscar2 is Oscar1s duplicate on Twin Earth (modulo th e chemical composition of their bodies). In particular, le t us imagine that Oscar1 and Oscar2 occupy type-identical (narrow) psychological states. The intuition, nonetheless, is that the extension of water as used by Oscar1 is the compound H2O, while the extension of water as used by Oscar2 is the compound XYZ, even though, due to their lack of chemical knowledge, they are not in a position to detect this difference and will have the impression their words mean the same thing. Oscar1, for instance, will assent to a lot of mistaken sentences such as There is water in rivers and lakes on Twin 113

PAGE 114

Earth. Putnam s conclusion is that narrow psychol ogical state does not fix extension, and hence, since he takes meaning to be by definition enough to fix extension, that meaning is not determined by narrow psychological state. We can extrapolate and propose that thought content is not so determined either. According to the doctrine of the division of linguistic labor, for a natural kind term like elm, a distinction can be dr awn between a limited number of experts, who possess a way of recognizing elm trees (Putnam, 1975, p. 145), and the rest, who can nonetheless use elm competently simply in virtue of their social links within the larger linguistic community which includes experts. More sp ecifically, Putnam writes: Whenever a term is subject to the division of linguistic labor, the average speaker who acquires it does not acquire an ything that fixes its extension. In particular, his individual psychological state certainly does no t fix its extension; it is only the sociolinguistic state of the collective linguistic body to which the speaker belongs that fixes the extension. (Putnam, 1975, p. 146). I suppose that in the last sentence Putnam is idealizing away from the contribution to meaning of the non-social environm ent. Note that it is not fully clear how we are to think of expert speakers, though. In light of his own Twin Earth thought experiment, by expert speaker he cannot mean somebody in possession of an infallible test for membership to the kind, or else it would follow that prior to the scientific discovery of the kinds nature, no t even the state of the community as a whole (together with the physical environment) suffices to fix the referent.30 Then the question arises about exactly how to understand an experts recognitional abilities. Whether ultimately justified or not, Putnams distrust in the helpfulness of superficial properties in fixing the referent suggests such abilities mi ght have to be grounded in perceptual encounters with members of the kind. 30 Putnam does, though, make the puzzling remark that water, for example, did not exhibit [this kind of division of linguistic labor] at all prior to the rise of chemistry (Putnam, 1975, p. 145). 114

PAGE 115

Leaving such obscurities aside, it is plausibl e that Putnam s notion of a non-expert user of a natural kind term is intended to cover speak ers who use the term deferentially. As we have seen in the critical examination of Kripke abov e, deferential uses do not in any way support the hypothesis that speakers often la ck any reference fixing mode of presentation. Therefore, we have been offered no compelling reasons to embrace Putnams conclusion that extension is determined socially, not individually, owing to the division of linguis tic labor (Putnam, 1975, p. 165). Should the Twin Earth thought experiment convince us that knowledge of meaning does not consist in the ability to deploy a certain inte ntional content, and that meaning is largely a matter of how the individual is placed in the wo rld, of what causal rela tionships hold between her and the environment? Ironically, Putnam offers the following diagnosi s of why natural kind terms exhibit the interesting behavior that his thought experiment brought to light: Words like now, this, here have long b een recognized to be i ndexical [] For these words no one has ever suggested the traditional theory that intension determines extension. [] Our theory can be summarized as saying that words like water have an unnoticed indexical component: wat er is stuff that bears a ce rtain similarity relation to the water around here. (Putnam, 1975, pp. 152-53) This looks to me like a potentially self-def eating diagnosis for an externalist to make because it immediately suggests a way of cont aining the problem posed by the Twin Earth thought experiment and drastically reducing the number of things th at may be needed in addition to intentional content to fix the extension of linguistic expressions. I think it is right that, in the last analysis, the problem Putnams though experi ment calls attention to is derivative upon the difficulties that indexicals raise for internalism. I think, however, that this is also why the thought experiment fails to establish the robust, wi de-ranging form of externalism that Putnam champions. Even though narrowly construed thought contents ma y not by themselves yield an extension, all that is needed is to supplement them with a spec ification of the thinker and the 115

PAGE 116

present tim e, as that is sufficient to ha ndle the problem raised by indexicals and demonstratives.31 Therefore, the most that Putnams argument can establish is that, if meaning and thought content are by definition sufficient to determine extension, then wide thought contents must be postulated (if we want to call th em that), which may not be strictly speaking in your head, but whose object depe ndence regards exclusively the subject and the present time. This is a very attenuated variety of externa lism (note that only reference to the present time contravenes the usual characteriza tion of internalism), and one that would be of no help to the physicalist because in the two-dimensional framew ork, the extension of a concept token given a scenario considered as actual is always determin ed relative to the specification of a speaker and a time. To sum up, contrary to what is often assumed, the Kripke-Putnam arguments do not succeed in challenging conceptual ascriptivisms st atus of default position on how concepts latch onto the world, a status that is due to its in tuitive appeal. Therefor e, if somebody deploys conceptual NA, in an argument to defend physicalism, for instance, it is incumbent on him to make a case that, although highly c ounterintuitive, conceptual NA is to be preferred, rather than take its truth for granted. 31 See (Ludwig, 1996) for such a minimalist approach to singular thoughts. 116

PAGE 117

CHAPTER 5 THE METAPHYSICAL IMPLICATIONS OF THE EXPLANATORY GAP From an Explanatory Gap to an Ontological Gap In light of the generalized derivability constr ual of the explanatory ga p that I defended in Chapter 3, in this se ction, I will first provide a sketch of a version of the gap argument that doesnt rest on the two-dimensional framewor k. This is a generalization of an argument suggested by Chalmers, which hinges on his conception of the gap as lack of derivability from the microphysical, and which I have given a reconstruction of in the sec ond section of Chapter 2. I will subsequently offer some background on two-dimensional semantics and present Chalmerss reconstruction of the version of the gap argument that makes use of two-dimensional semantics. I will end by reformulating Chalmerss argument in more fundamental terms in view of my generalized derivability account of the explanatory gap, a nd identifying the places in the gap argument where it might make se nse for the physicalist to object. First, the sketch that doe s not require the two-dimensi onal framework. If C is the collection of non-phenomenal predicates in L whic h pick out properties th at arent logically necessitated by more basic properties that are al so picked out by phenomenal-free predicates in L, let D comprise all the true atomic sentences in L that specify exactly which objects satisfy the predicates in C together with the laws of nature that can be formulated using only predicates in C, and logical and modal terms. Postulating psychophysical id entities or claiming that the phenomenal metaphysically supervenes upon the physical is ad-hoc because it is tantamount to postulating epistemically primitive identities or necessities, and there dont seem to be any such things outside the context of the mind-body problem. So, if primitive psychophys ical or metaphysical supervenience is the physicalists only option, we should reject physica lism. An identity statement is epistemically 117

PAGE 118

prim itive just in case it is not deducible from statements that dont involve identities. A metaphysical necessity statement is epistemically primitive iff its neither logically necessary nor deducible from non-modal statements (which dont involve identities) plus statements of mere nomic necessity. Any true identity statement or metaphysical necessity statement formulated in L that contains only phenomenal-free pred icates is deducible from D, and consequently it is not epistemically primitive. For identity and necessi ty statements that include terms similar to water, the explanatory story is a bit more complex, but they arent epistemically primitive either because they are entailed by D in conjunction with certain statements that express causal connections between states pi cked out by non-phenomenal terms and phenomenal states. However, psychophysical identity statements and metaphysical necessity statements are epistemically primitive, because they are not deducible from the conjunction of D and a complete description of the phenomenal and stat ements of mere nomic necessity of any kind. Thus, we should reject physicalism. Now the argument formulated in the twodimensional framework. In Chalmerss twodimensional framework, there are two intens ions associated with every expressiontoken/concept: the primary intension is a func tion from centered logically possible worlds considered as actual to extensions, while the s econdary intension is a function from logically possible worlds considered as c ounterfactual to extensions. A cente red world is an ordered pair of a world and a center in that world, which is in turn an ordered pair of a subject and a time. It is important that the two intensions att ach to expression-tokens rather than expressiontypes. In particular, the primary intension associated with a certain speakers use of a predicate is a function from centered worlds considered as ac tual to extensions which is grounded in the 118

PAGE 119

speaker s conditional ability to id entify the extension of the concept she associates with the predicate, given relevant information about how the world turns out to be, and ideal rational reflection. Consequently, it is not exactly linguistic meaning that the primary intention is designed to capture, but rather th e conceptual content that the sp eaker associates with a certain expression-token. This has crucial implications as far as the import of the gap argument goes: when articulated in two-dimensional terms, th e gap argument proves to rest on ascriptivism about conceptual content, rather than linguistic ascriptivism. A statement, S, is primarily possible if its primary intension yields the value true for some centered world, and it is secondarily possibl e if its secondary intension takes the value true for some possible world. S is said to be ideally primarily negativ ely conceivable if and only if ~S is not a priori. S is ideally primarily positively conceivable if and only if given ideal rational reflection, one can coherently modally im agine a situation, and one judges that if the actual world turns out to be that way, then it is incoherent to s uppose at the same time that S is false. Chalmers offers the following reconstructi on of the knowledge argument and the zombie argument, which makes use of his two-dimensional framework.32 (P is a full description of the microphysical, Q is a phenomenal statement to the effect that a par ticular instantiates a phenomenal property; if Q is not semantically neutral, let Q be a statement whose secondary intension is identical to the prim ary intension of Q modulo centering33; Chalmers omits T and I, 32 (Chalmers, 2002). 33How are we to understand the claim that Q has the sa me secondary intension as the primary intension of Q (modulo centering)? As far as I can see, this must have to do with de-rigidifying the mental terms in Q, so to speak. (That may still leave us with context sens itivity.) If pain is interpreted as synonymous with the state that actually plays such and such a causal role, we just remove the adverb actually. If pai n is thought as having its reference fixed via some description, then we just substitute for pain in Q that description (which is unactualized). 119

PAGE 120

and so will I for the sake of si mplicity). The ve rsion about negative conceivability captures the import of the knowledge argument, while that app ealing to positive conceivability corresponds to the zombie argument. Because of the way Chalme rs is thinking of the explanatory gap, his version of the gap argument actually coincide s with the knowledge argument as reconstructed below. 1. P&~Q is ideally primarily posit ively/negatively conceivable. 2. If P&~Q is ideally primarily positively/nega tively conceivable, then P&~Q is primarily possible. 3. If P&~Q is primarily possibl e, and both P and Q are semantic ally neutral, then P&~Q is secondarily possible, and he nce physicalism is false. 4. If P&~Q is primarily possible, and P is sema ntically neutral but Q is not, then P&~Q is secondarily possible, and he nce physicalism is false. 5. If P&~Q is primarily possible, and P is not semantically neutral but Q is, then either P&~Q is secondarily possible (and hence physica lism is false) or panprotopsychism34 is true. 6. If P&~Q is primarily possible, and neither P nor Q is semantically neutral, then either P&~Q is secondarily possible (and hen ce physicalism is false), or th e causal roles expressed by the primary intensions of phenomenal terms are play ed in the actual world is necessitated by the actual nature of the players of the microphysical roles (plus perhaps the macrophysical causal roles themselves), and hence panprotopsychism is true. 7. So, either physicalism is false, or panprotopsychism is true. It is important to clarify the import of the no tion of semantic neutra lity, as it plays a key role in the arguments above. It turns out that a statement S is semantically neutral just in case the primary intension of S is identical to it s secondary intension modulo centering, which comes down to the requirement that S display no context sensitivity whatsoever. In other words, for every possible world w, when we evaluate S as uttered in w (in any cen ter in w, actually) and relative to w we get the same result as when we evaluate S as uttered in <@, c> relative to w. 34 That is, the view that fundamental particles ha ve intrinsic properties which are protophenomenal. 120

PAGE 121

It turns out that semantic ne utrality, however, is not the only feature we must assume P has at stages 3 and 4 of the reconstructions given above. Since the arguments rest on the assumption that physicalism can be true only if th e secondary intension of P necessitates both the primary and the secondary inte nsion of any true statement, it is assumed throughout that microphysical concepts have a cons tant secondary intension (i.e., ar e rigid). It should be noted that what the combination of semantic neutrality and rigidity yields is precisely my notion of being hyperstable. Consequently, at the steps where Chalmers maintains he supposes that microphysical predicates are semantically neutral, what we need in fact to assume is that they are hyperstable. I turn now to presenting the argument from the explanatory gap to the falsity of physicalism according to my generalized derivabi lity account of the gap. If there are hyperstable phenomenal-free non-functional predicates in L (i.e ., if there are physical properties), then let G be a set of statements that provides complete information about which objects satisfy these predicates and also specif ies what laws govern the properties picked out by them. 1. D&~Q is negatively conceivable. (Lack of de rivability from D is what the explanatory gap consists in.) 2. Either there are physical pr operties, or there are not. 3. If there are no physical propertie s, then physicalism is false. 4. If there are physical properties, G&~Q is negatively conceivable too. (since D entails G) 5. Therefore, G&~Q is primarily possible. 6. So, if the phenomenal predicate Q includes is hype rstable, then the property determined by its secondary intension is non-physical. 7. So, if the phenomenal predicate Q includes is not hyperstable, th en it is the property determined by its primary intens ion that is non-physical. 8. Consequently, physicalism is false. I have argued that the gap intu ition is a phenomenon grounded in certain semantic features of phenomenal terms, one of which is their being hyperstable. That makes step 7 unnecessary. It is natural to expect the gap argument to remain closely connected to the zombie and the knowledge argument, when those are likewise pr operly understood. Accord ingly, the knowledge 121

PAGE 122

argum ent has the same reconstruction as the ga p argument, while the zombie argument is quite similar except that it relies on positive instead of negative conceivability. It is important to isolate the steps in the gap argument above that a physicalist who grants that the phenomenal is not entailed by D c ould in principle resist. Being clear on what the options are for the physicalist will be essentia l in the rest of the chapter, where I give reconstructions of various physicalist responses to the gap problem and evaluate them. There appear to be two main strategies such a physicalis t could make use of to block the move from the negative conceivability of G&~Q to the falsity of physicalism: On the one hand, he could deny that the primary possibility of a statement, wh ich amounts by definition to the existence of a logically possible world relative to which its primary intension is true, guarantees the existence of a metaphysically possible world in which the pr imary intension of the statement is true. On the other hand, he could reject as criptivism about conceptual content, which the two-dimensional framework rests on, which would have the consequen ce that conceivability intuitions have in fact no bearing on questions about po ssibility. The first move commits one to what Chalmers calls strong necessities, i.e., statements that are a posteriori but have a necessary primary intension. Denying the Epistemic Asymmetry The purpose of sections 2-5 is to offer a crit ical exam ination of some prominent physicalist responses to the gap argument. The strategies ad opted by physicalists fall into two categories. Some philosophers reject the conception of sc ientific reduction that the explanatory gap argument presupposes, and claim that from an ex planatory point of view, there is really no difference between identity statements that the phenomenal is not conceptually constitutive of and psychophysical identity statemen ts. Against this first type of strategy (which is discussed in the current section), I argue that it involves co mmitment to a mistaken picture of scientific reduction, and that it seems unable to account for the gap intuition. The alternative approach is to 122

PAGE 123

grant that th ere is an explanatory asymmetry betw een the two sorts of theoretical identifications, but to maintain that it can be accounted for in terms of special features of phenomenal concepts possession of which is perfectly compatible with physicalism. Most versions of what is often labeled in the literature the phenomenal concepts strategy in dealing with the gap problem belong here. The best way to think of the gap on a view of this type is as an epistemic phenomenon that is fully grounded in harmless feat ures of our concepts, and which consequently warrants no metaphysical conclusi ons: the existence of the explanatory gap provides no grounds to postulate a corresponding metaphysical gap. I exam ine and criticize prominent versions of this approach in sections 3-5, while the last section of the chapter is devoted to examining in general the prospects for a posteriori phys icalism and neutral monism. One possible reaction to the challenge posed by the explanatory gap is to deny that the phenomenal generates an explanatory gap by re sisting the underlying conception of reductive explanation. Some type B materialists deny that there is a ny epistemic asymmetry between ordinary theoretical identific ations and psychophysical ones.35 This is to endor se a version of what Levine calls non-exceptionalism (NE), acco rding to which not only zombies but also zombie H2O is conceivable. They claim that psychophysic al identities are just ified in exactly the same manner as the identification of water with H2O is, i.e., via an argument to the best explanation. Block and Stalnaker and Papineau appear to fall in to this category. Another feature their views seem to share in common is a commitm ent to some form or other of non-ascriptivism about conceptual content. That is something appa rent in Papineau, not qu ite so obvious for Block and Stalnaker, though, for the reason that what they explicitly reject is just linguistic nonascriptivism. 35 Their motivation has nothing to do with the view that many natural kind concepts have phenomenal constituents. 123

PAGE 124

Combining NE with ascriptivism about macrophysical concepts results in a highly implausible position. As I argued in Chapter 4, one advantage of my generalized derivability construal of the gap over Chalmerss is that the underling account of scien tific reduction is a lot harder to object to. If it might seem doubtful that ordinary theore tical identifications such as water is H2O are entailed by microphysics alone, it woul d be highly implausibl e to insist that the conditional from a full description of the distribution of microphysical and chemical properties to water is H2O is conceivably false. The only option left for an espouser of NE is, therefore, to embrace non-ascriptivism about macrophysical concepts, a view which I examined in the last section of Chapter 4 and found unmotivated.36 Consequently, denying that there is an explanatory gap proves to rest on a mistaken conception of scientif ic reduction. Moreover, even some body who denies that there is an explanatory gap must discharge the obligati on of explaining how the gap intuition arises. Block and Stalnaker do not attempt to address th is issue. Even though Papineaus quotational model of phenomenal concepts is designed to ac complish precisely that, a criticism of this putative explanation that was first advanced by Levine can be further developed employed to show that Papineaus attempt to account for the gap intuition fails. According to Papineau, using a phenomenal concept involves instantiating the very property the concept refers to (or a relevantly similar oneI will ignore this qualification in what follows). The corresponding material concept is nothing like that. Hence there is a sense in which the material concepts leave out the pheno menal properties. And from this its very easy to slide, fallaciously, into the conclusion that material concepts cannot refer to phenomenal properties (Papineau, 2002, p. 105). 36 In fact, Papineau endorses the radical variety of non -ascriptivism about all concepts, including phenomenal concepts, which he quite implausibly maintains are nakedly referential, too. 124

PAGE 125

An objection along the following lines is sugge sted in (Levine, 2006). Suppose there is a concept C with the following features: it picks out physical property P, and tokenings of it necessarily involve being in a physical state that has P exactly in the sense in which on Papineaus view tokenings of a phenomenal concept necessarily i nvolve instantiating the phenomenal property at issue. Suppose also that states with property P are typically not associated with any phenomenology (there is nothing it is like to be in such a state.) Let C be another concept that picks out the same property P, but whose tokenings do not involve be ing in a state that has P. Do we have the intuition that a being that posse ssed both concepts woul d be bound to be puzzled about how C and C could pick out the same thing? Would such a being automatically be inclined to suppose that the property C picks out is non-physical? It seems the intuition is that it would not, while Papineaus account of the gap intuition commits him to saying that a parallel distinctness intuition should arise here as well. I think that what the thought experiment pr oposed by Levine nicely brings out is that what may be doing the explanatory work is not the quotational model of phenomenal concepts exactly, but rather something brought into the pict ure illicitly. Ironically, Papineaus antipathetic fallacy story sounds prima facie pl ausible (if it does) only because it in fact tacitly exploits the very dualist intuition it was suppos ed to explain. His explanation has any grip on us only to the extent that it exploits the fact that phenomenal states are such that there is something that it is like to be in them. If having the property is constitutive of deploying th e phenomenal concept, since there is something that it is like to ha ve the property, thinking of that property via a phenomenal concept is going to feel very differe nt from thinking of it via a material concept, hence the temptation to conclude that a material concept leaves out the feeling not only in the 125

PAGE 126

sense that the feeling is not i nvolved in the tokening of the con cept, but in the sense that the feeling is not even referred to (Papineau, 2002, p. 170). For instance, he em phasizes how in deploying phenomenal concepts, the mind is some how in possession of an instance, or version, of the property being referred to (Papineau, 2002, p. 106). But if you think of the phenomenal property as a property of the brai n, then it is not at all obvious th at the mind will be somehow in possession of the property merely by virtue of the brain instantiating th e property, at least not with the strength of logical necessity. Here is a suggestion on Papineaus behalf. Papineau could supplement the story about the phenomenon of quasi-self-referen tiality with some other devices that a physicalist can avail himself of. For instance, suppose one thought that phenomenal states are physical states, but that they have the peculiarity that one can only be in such a state if one knows non-inferentially (or is in a position to know non-inferentiall y) that one is in it. In that case, whenever the creature above deploys the concept C, not only does it instantia te physical property P, but it also knows noninferentially that it instantiates property P. By contrast, when it deploys the concept C, nothing like that happens. Would it be more plausible to claim that, in this case, the creature might become puzzled and inclined to suppose the feat ure picked out by C could not be the same as that picked out by C? The thought-experiment deployed earlier coul d be suitably modified, however, to accommodate this extra element, and so to show that this addition does not render the explanation any more convincing. On the contrar y, one might worry that Papineaus story about how phenomenal concepts work is not playing a ny explanatory role on the modified suggestion anymore. What motivated the modification I prop osed was the observation that his explanation can only seem plausible to the exte nt that it exploits the intuition that phenomenal states are in 126

PAGE 127

som e relevant sense present to the mind. And then I proposed that he coul d try to cash out this presence to the mind in cognitive terms. However, one gets then the impression that Papineaus suggestion that de ploying a phenomenal concept mu st involve instantiating the property referred to (or one very similar to it) is not doing any explanatory work anymore. Another, equally compelling, objection that ca n be raised is that the kind of confusion Papineau blames for the gap intuition does not really seem of the irreparable kind. It is not clear he can explain why, even though the confusion has just been exposed, its grip on us is as forceful as ever. The Epistemic View of Subjectivity (Sturgeon) Scott Sturgeon argues37 that even though qualia cannot be explained in non-phenomenal terms, that does not warrant any metaphysical co nclusions about their nature. He maintains that he has offered an explanation of the principled lack of a materialist acco unt of qualia that is ontologically neutral and hence compatible with physicalism. More specifically, Sturgeon claims that it is because of the special epistemic ro le of phenomenal concepts that a materialist explanation of qualia is not fo rthcoming. The fact that qualia are not explainable in physical terms cannot be used to refu te physicalism about phenomenal properties because it merely tells us something about the special epistemic features of the concepts we typically employ to think about those properties. In his view, the de bate between physicalism and dualism should eventually be settled on the basis of overarc hing theoretical consider ations (Sturgeon, 1994, p. 235), such as the principle of parsimony. In the present section, I ar gue that Sturgeons attempt to defend physicalism against the explanatory gap problem by accounting for the unava ilability of a physica listic explanation of 37 (Sturgeon, 1994). 127

PAGE 128

qualia in te rms of the special epistemic role of phenomenal concepts is not successful. After offering a reconstruction of Sturgeons view of the special epistemi c role of phenomenal concepts and examining Sturgeons preliminary ch aracterization of types of property explanation that are relevant to the physic alist project, I take a look at hi s main argument, and offer my diagnosis of where hi s strategy goes wrong. To assess Sturgeons defense of physicalism, we will need to first clarify what the special epistemic role of phenomenal concepts consists in according to him. In Sturgeons terminology, a concept is canonically linked to a type of mental state if the epistemic connection between them is at le ast partly individuative of the concept in question (Sturgeon, 1994, p. 227), i.e., if it is at least partly individuative of the con cept that applications of it are warranted by mental states of that kind. For instance, Sturgeon claims that the concept of red is canonically linked to visual impressi ons as of red objects. He calls the mental states that are of the type which is canonically linked to a certain concept canonical evidence for applications of that concept.38 We need to distinguish between the concept red which applies to external objects, and a different concept, let us say, Q-red, which applies instead to the distinctive qualitative character of experiences as of red. It is Sturgeons claim that both red and Q-red are canonically linked to the same set of mental states, namely, Q-red experi ences. Both the belief that there is something red over there and the belief th at Im having a Q-red experience are in his view justified noninferentially. What plays the role of evidence in both cases is the very same phenomenal state 38 That experiences as of red are canoni cal evidence for applications of red does not mean that there could not be applications of red that are warranted in a different way. For example, I could very well come to believe while blindfolded that there are red objects in the room on the grounds that somebody tells me so. It is just that this would not be applying the concept red on the basis of canonical evidence. 128

PAGE 129

nam ely, an experience as of something red.39 However, he points out, the relationship between the evidence for the belief and what the belief is about is very different in the two cases (see Table 5-1). In the case of beliefs about the color of external objects, the evidence is distinct from the reality the belief points to, or as we might call it, the state of affairs that makes the belief true. There is, on the one hand, ones having an experience as of red a nd, on the other hand, the external object being red. This gap between appearance and reality leaves room for the possibility of conservative defeaters for the evidence, i.e., defeaters that break the link between a concept and its canoni cal evidence while leaving that evidence intact (Sturgeon, 1994, p. 229). For example, if one discovers that the objects around one are bathed in red light, even though one is still in possession of canonical evidence for applying red (i.e., one has a Qred experience), one is not warranted anymore in believing that the objects themselves are red. Table 5-1. Types of belief justification Belief Evidence/Appearance Property the belief is about/reality There is something red in front of me. Q-redness of the experience Re dness of the external object I have a Q-red experience. Q-redness of the experience Q-redness of the experience In the case of phenomenal beliefs, Sturgeon claims that there is no parallel appearancereality distinction: the ca nonical evidence for applying Q-red and the property the belief that I have a Q-red experience is about are one and the same, namely th e Q-redness of my experience. What warrants my belief that I have a Q-red expe rience is simply my having a Q-red experience. Therefore, it looks as tho ugh canonical applications of Q-red simply do not rely on evidential intermediaries, and hence the possibility of conservative defeaters is ruled out for them. 39 On occasion, Sturgeon also talks about Q-redness itsel f (rather than the Q-red st ate) as the evidence. 129

PAGE 130

To return to Sturgeons main tenet, he hol ds that, even if qualia were physical, an explanation of them in physical terms would not be available because of the special way in which they are conceptualized by phenomenal concepts. As we have seen, what makes phenomenal concepts sui-generis in his view is their special epistemic role, which consists in the fact that canonical applicatio ns of phenomenal concepts do not rest on any evidential intermediarywhat serves as the ev idence for the belief is the very state that the belief is about. Sturgeons approach and terminology are ra ther idiosyncratic, and it is therefore sometimes difficult to relate his remarks on phys icalism and the gap to the manner in which these topics are typically dealt with the litera ture. For example, it is somewhat unclear how Sturgeon is thinking of the sort of explanation of qualia in physical terms which the physicalist is faulted for being unable to offer. Clearly, it mu st be an explanation that begins in nonqualitative terms and ends with the instantiat ion of qualia (Sturgeon, 1994, p. 221), but exactly what form would it have on Stur geons view? A brief examina tion of his classification of property explanations (represented in Fi gure 5-1) is in order at this point. Homogeneous Heterogeneous Nondispositional Dispositional Nondispositional Dispositional 1 2 3 4 Figure 5-1. Generative property explanation Sturgeon introduces a distinction between cau sal and generative property explanations (C-explanations and G-explanations henceforth). A C-explanation of a property specifies when an object might gain, lose, or re tain that property, and also what causally follows from this (Sturgeon, 1994, p. 223). This is contrasted with a G-explanation, which tells us how a target 130

PAGE 131

property is instantiated in an obj ect, [] how it is grounded or non-causally generated in an object (Sturgeon, 1994, p. 223). For any property explanation, let us call the explained property the target property. Gexplanations can be homogeneous or heterogeneous depending on whether the concepts used to introduce the explaining propert ies (i.e., the explaining concepts) are homogeneous with the concept used to pick out the target property (i.e., the target concept). For instance, an explanation of the shape of a complex object in terms of the shape of its parts, or of its weight in terms of the weight of the parts is a homogene ous G-explanation. Sturgeon assu mes that it is a heterogeneous G-explanation of qualia that we would normally expect from the physicalist. Sturgeons explicit characterization of he terogeneous G-explanation is somewhat puzzling. As clarifying the notion of heterogene ous G-explanation is important in order to properly assess Sturgeons story about why such an explanation cannot be given for qualia, his remarks on the mechanisms underlying heterogene ous G-explanation are worth examining with some care. To return to the quote above about what is distinctive of a G-expl anation, given his talk of non-causal grounding of a property, it would seem that functional properties are very good candidates for properties susceptible of a hetero geneous G-explanation. Indeed, Sturgeon claims that a heterogeneous G-explanation can have as target property either a dispositional or a nondispositional property, and when the target is di spositional, we try to understand how the target is instantiated by looking for underlying propertie s that realize the disposition (Sturgeon, 1994, p. 224). For instance: When we explain the fragility of glass by appeal to its molecular structure, or the bonding properties of chemicals by appeal to their microphysical constitution, we have heterogeneous G-explanation. In these cases, we explain how a dispositional target is 131

PAGE 132

instantiated by appeal to the instantiation of lower le vel properties. (Sturgeon, 1994, p. 224) This sounds as though what Sturgeon means by d ispositional property is simply functional property, in which case a heteroge neous G-explanation of a disposi tional target property is just a realization theory. One confusing feature of Sturgeons discussion is, however, his occasional talk of the causal role of a dispositional target property. For instance, he claims that when the target is dispositional, its G-explanation will essentially rely upon its (emphasis mine) causal role. When the target is non-di spositional, it will not (Sturg eon, 1994, p. 224). For instance, he says quite explicitly that, when we explain fragility in terms of molecular structure, or chemical bonding in terms of microphysical structure, we e ssentially rely upon the targets causal role (Sturgeon, 1994, p. 224). His language here might sugge st that he treats te rms like fragility not exactly as functional terms, but rather as functi onal descriptions of cat egorical properties. I think that Sturgeons taxonomy of heteroge neous G-explanations can be characterized more perspicuously if a distinct ion is introduced between functi onal terms/concepts and what I will call functional description terms/concepts, where the latter are terms/concepts that pick out categorical properties via functional descripti ons of them. To illustrate the distinction, let us focus on the predicate being fragile and examin e how it would work, if it were interpreted in each of the two ways. If considered a functional term, being frag ile would be analyzable along the lines of having some property or other which plays such and such a causal role. On this view of the predicates meaning, being fragile picks out a functional prope rty, which can presumably be multiply realized by various lower-level properties. Consequently, the predicate can be true of objects which are fragile in virtue of different cat egorical properties that fill the relevant causal role, because what the predicate picks out is th e causal role itself, not some role player. By 132

PAGE 133

contrast, if deem ed a functiona l description term, being fragil e would be analyzable along the lines of having the property that fills such a nd such a causal role, a nd it would refer to what actually plays the causal role. While it lies beyond the scope of my discussion to decide which of the two approaches is more plausible, note that either way being fragile is clearly amenable to a heterogeneous G explanation. Both functional pr operties and categorical properties picked out by functional description concepts are great candidates for heterogeneous G explanation. In the former case, the explanation assumes the form of a realization th eory, while in the latter, it yields an identity claim: the target property as conceived under the target concept is identified with some conceptually heterogeneous explai ning property on the grounds that it is the latter that plays the causal role expressed by the target concept. Therefore, when th e target concept is a functional description concept, the heterogeneous Gexplanation does rely upon the causal role of the target categorical property. When the target is a functional property (conceived via a functional concept), the G-explanation relies upon the causal role that is integral to the target property (rather than being the causal role of the target property), and cite s the realizer of the target property as the explanans. It is not fully clear whether Sturgeon intends to include in the cat egory of dispositional properties only functional properties or perhaps also categorical properties conceived of under functional descriptions. One might be tempted to be charitable and assume that he consistently has in mind the former, and that by the causal role of a dispositional property he means (rather non-standardly) the causal role that is constitutive of a functional property, rather than the causal role played by a categorical property. On this interpretation, categorical properties conceptualized via functional desc ription concepts count as nondispositional properties and to 133

PAGE 134

give a heterogeneous G explanation for them is to follow path 3 in Figure 5-1. On the alternative interpretation, they fall rather under category 4. While it is not crucial which of the two inte rpretations was in fact intended by Sturgeon, it is important that, even t hough phenomenal concepts seem st ructurally dissimilar from both functional and functional descriptio n concepts, all the examples of target concepts that Sturgeon discusses look like either functiona l or functional description concep ts. What one wants to ask at this juncture is whether heterogeneous G explanation can in fact be offered for any target concepts that share the structur e of qualia concepts and so are neither functional nor functional description concepts. If the answer is affirmativ e, one would naturally be very curious to hear how such an explanation goes, since the model fo r that must be rather different from what Sturgeon has offered so far. This issue will be revisited toward the end of this chapter. In his attempt to show that it is the peculi ar epistemic role of phe nomenal concepts that renders qualia not explainable in physical terms, Sturgeon ad vances the following argument: 1. A compelling heterogeneous G-explanation requires that the explaini ng concepts connect sensibly with the canonical evidence a ssociated with the target concept. 2. In the case of a heterogeneous G-explanation, explaining concepts can connect sensibly with the canonical evidence associated with the targ et concept only if the canonical evidence is distinct from the ta rget property. 3. But the canonical evidence associ ated with phenomenal concepts coincides with the target property. 4. Therefore, qualia are not amenable to a heterogeneous G-explanation. Before we examine how the premises of the argument are substantiated, we must try first to clarify the notion of canonical evidence, especially in re lationship to concepts whose application is typically justified inferentially. On Sturgeons foundationalist picture of epistemic justification, if a belief is justified inferentially, there must be an end to the chain of justifiers, which consists in non-inferentially justified belief s. Their justification rests in turn on various kinds of non-doxastic mental stat es (e.g., perceptions). By canonical evidence for a concepts 134

PAGE 135

applications Sturgeon appears to m ean the collec tion of the non-doxastic me ntal states that the inferential chain associated with a standa rd application of the concept ends in. To return to the premises of the above ar gument, Sturgeon suggests in support of (1) that the only way for the connection between the targ et property and the he terogeneous explaining property not to look arbitrary is for the explaining property to be shown to be what is causally responsible for the canonical evidence that is asso ciated with the target concept. Therefore, it seems that the explaining concept connecting sensibly with the targ et concepts canonical evidence comes down to the property singled ou t by the explaining concept being what causes the canonical evidence for the target concept. St urgeon writes: Imagine, for example, trying to explain colors in terms of wave lengths but failing to detail how wave lengths might look like colors. [] No properties can sensibly be thought to realize color unless those properties can sensibly be thought to look like colors. [] [I]mag ine trying to explain th e solidity of objects in terms of component electron clouds but failing to detail how elec tron clouds can feel and look solid (Sturgeon, 1994, p. 232). In connection with (2), St urgeon holds that it is precisely the distinctness between canonical evidence and target property which makes it possible to forge a theoretical link from explaining properties to a con ceptually dissimilar target pr operty (Sturgeon, 1994, p. 233). The thought is that, if the canonical evidence is not distinct from the target property, then the identification between target a nd a conceptually heterogeneous explaining property cannot be justified anymore on the grounds that the explaining property is what is causally responsible for canonical evidence; the relationshi p between evidence and target property is not causal, but rather identity. Consequently, th e conceptual heterogeneity of ta rget and explaining property is bound to make their identific ation seem arbitrary. 135

PAGE 136

One should keep in mind that Sturgeons avowed goal in this article is not to show that qualia cannot be explained in physical terms. That is supposed to be accepted by all parties to the debate. Rather, his aim is to offer an explanat ion of the gap that shows that the gap poses no problems for physicalism. The explanation o ffered goes as follows. For heterogeneous G explanation to be available, th e target concept must meet a cer tain requirement: there must be distinctness between canonical ev idence and target property. Qua lia concepts, however, do not meet this requirement owing to their special epistemic role. Therefore, the absence of a heterogeneous G explanation for qualia raises no problems for physicalism. Even if qualia were physical, the standard mechanisms at work in justifying property reduction would not be applicable in their case becau se of the unique way in which they are conceptualized. Does not this look like a promising strategy in responding to the gap challenge? One could argue that there is a structural problem with Sturgeons line of thought. What Sturgeon has accomplished is just to isolate a property of phenomenal concepts (i.e., their having a special epistemic role) which is sufficient for the gap to arise. Howeve r, if some other feature of phenomenal concepts, P, can be found such that P is also sufficient for the gap, but P looks problematic for physicalism (in the sense that it is at least prima facie plau sible to think that, if phenomenal concepts have P and perhaps some ot her, uncontroversial feat ure, then what they pick out is not physical), then it is incumbent on Sturgeon to s how that P is only apparently problematic. I think that phenomenal concepts do have a special feature that sharply distinguishes them from concepts that express causal roles (whether functional or functional description concepts), though they shar e it with concepts like being spherical : grasping the concept appears to give one a special sort of access to the pr operty picked out, which makes it difficult to 136

PAGE 137

understand how a property like being in pain, for exam ple, could turn out to be physical. To give a rough characterization of th e semantic roots of this feature of concepts like being in pain and being spherical one could say that a concept appears to afford such access to the property it singles out just in case it is rigid (i.e., th e property picked out does not vary with the counterfactual scenario relative to which the concept is evaluate d) and context insensitive (i.e., its semantic value does not depend on which scenario we assume it is tokened in). In Chapter 3, I called this property being hyperstable.40 Being hyperstable can be given a rigorous characterization in Chalmerss two-dimensional framework: a concept c ounts as hyperstable if and only if its 1-intension is both identical to its 2-intension and a constant function. Note that some functional c oncepts can also be hyperstabl e, but phenomenal concepts do not seem to be functional, so it is intuitive to a ssume that no realization theory is available for them. Being hyperstable is also sufficient to preclude heteroge neous G explanation.41 Recall that we wondered at an earlier point in the chapter whether heteroge neous G explanation is 40 Being hyperstable can be given a rigorous characteriza tion in the two-dimensional framework due to Chalemrs. Roughly, the idea is that a concept counts as hyperstable iff its 1-intension is both identical to its 2-intension and a constant function. 41 One might worry that Sturgeon appears inclined to reject this claim. He su ggests that if the causal connection between target property and canonical evidence is built into th e target concept, then we can be certain that the target property is identical to the property that induces the ca nonical evidence (Sturgeon, 1994, p. 233). Even if the causal link is not built into the target concept, however, we are said to still have good empirical reasons (Sturgeon, 1994, p. 223) to uphold the identification. According to Sturgeon, in such a case empirically establishing that despite their conceptual dissimilarity, explaining prop erties look, feel, taste and so und exactly as the target canonically looks feels, tastes and sounds provides strong empiri cal reason for supposing th at the target property is realized by the conceptually dissimilar explaining properties (Sturgeon, 1994, p. 233). This thought could be used to attack my sufficiency claim. Of course, strictly speaking, Sturgeon is not committed to rejecting the claim that being hyperstable is sufficient for the gap, because he might contend that, while the causal connection between target property and canonical evidence doesnt have to be built into the target concept for heterogeneous G explanation to work, there is somethi ng special about hyperstable concepts that defeats what would otherwise be solid empirical reasons for property identification. Since Sturgeon says absolutely nothing about what sorts of considerations might count as defeaters, though, it seems only fair in light of what he does explicitly say to consider the objection that being hyperstable isnt sufficient to preclude heterogeneous G explanation. Let us consider the hyperstable concept of being spherical Could it turn out that the property of being spherical has a hidden nature just as water, temperature or lightning do? The property of being spherical is the 137

PAGE 138

available for any concept that is structurally like ph enom enal concepts, which comes down to any concept that is hyperstable. The question should be answered in the negative. It might seem that the fact that being hypersta ble is a more general feature of phenomenal concepts, which they share with certain other concep ts, could be used against the physicalist. No hyperstable nonphenomenal concepts are amenable to heteroge neous G explanation. We have, then, at least prima facie good reasons to think that, just as for these other concepts, phenomenal concepts afford insight into the nature of the property picked out. Why should we think that phenomenal concepts are the exception? (Even if Sturgeon were inclined to deny that phenomenal concepts are hyperstable, his argument would still seem to be underm ined, for we have a strong intuition that phenomenal concepts have the behavior char acteristic of hyperstable concepts, and therefore the burden of proof would rest on Sturgeon to show that this intuit ion is confused.) property of being shaped like a three dimensional closed surface consisting of all and only the points that are equidistant from a given point, or of being of a shape sufficiently similar to this. Could being spherical turn out to be identical with some conceptually heterogeneous non-shape-property? And could we come to such an identification on the grounds that it is this non-shape-property that u ltimately causes the canonical evidence for applications of the concept of a sphere (i.e., the experiences as of something spherical)? Consider the following thought experiment. Suppose there is a possible world in which the laws of nature are such that all spherical objects must be composed exclusively of molecules havi ng a certain property X (which though physical, is not a shape property) and all aggregates of molecules enjoying X must be spherical Arguably, in such a worl d, being an aggregate of molecules each of which has X is by hypothesis what relia bly generates in perceivers sphericality experiences, and it looks like by Sturgeons reasoning, we would have strong empirical reasons to identify this property with being spherical. I think our intuition about this case, though, is that such an identification is utterly unjustified. In this scenario, sphericality and being composed of molecules ha ving X are clearly two distinct properties between which there is a nomically necessary correlation. On a more general note, of course, talk of what causes an experience does not really make sense unless a particular element of some causal chain that ends in the experience has been raised to salience. In the case of compelling heterogeneous G explanation, it is the target concept that renders one such element prominent. For example, when we try to uncove r the hidden nature of water, we know we are looking for a chemical compound that has a certain causal role, and in particular, that causes expe riences as of a colorless, odorless liquid located in rivers, lakes, etc. A concept like spherical is neither a natural kind concept nor a functional concept, and the only thing it points to is the spherical shape of the object. That simply does not leave room for any discovery of a hidden nature of the target property. The fact that the spherical shape is necessitated with nomic strength by the conceptually heterogeneous microphysical property of being X seems irrelevant to the question about what property spherical picks out, and this seems generalizable to any other hyperstable concept. Consequently, it seems safe to conclude that Sturgeon offers no reasons to doubt that being hyperstable is sufficient to preclude heterogeneous G explanation. 138

PAGE 139

In response to the objection under considera tion, Sturgeon might argue that, even though the familiar mechanism of scientific reduction cannot apply to any hyperstable concept, the phenomenal case is peculiar in that only for phenomenal concep ts there is strong inductive evidence for identifying their referents w ith conceptually heterogeneous properties.42 And perhaps it is here that the special epistemic role of phenomenal concepts really enters the scene, as it might seem to help account for the status of psyhophysical identity statements as strong necessities: Maintaining that psychophysical identities are strong necessities is not ad-hoc. We have good inductive reasons to believe such statements are true. If true, they are strong necessities, but that should not come as a great surprise, given the special epistemic role of phenomenal concepts. In the standard examples of property reduction, ther e is a conceptual link between reducing and reduced property establis hed via canonical evidence. For phenomenal properties, canonical evidence and reduced prope rty are one and the same, and therefore the canonical evidence cannot serve as an epistemic intermediary between reduced and reducing property, which remain conceptually fully disparat e, hence the status of psychophysical identities as strong necessities and their intuitive implausibility. To see that this is just a pseudo-explanation of strong necessities, one needs to be mindful of the distinction betw een the following two explanatory proj ects: explaining why the standard reduction mechanisms are not applicable to c onsciousness, on the one hand, and explaining why postulating strong necessities in the psychophysical case is not after all ad -hoc, so that one can plausibly claim that the gap poses no insurm ountable problems for physicalism. The second project is that of explaining w hy one should grant that the phenomenal, although not amenable to reduction understood in the usual way, is suscep tible to a type of reduction carried out in 42 For more details on the inductive evidence at issue, see the last section of Chapter 5. 139

PAGE 140

accordan ce with a radically different methodology. The difference between these two explanatory projects is crucial, a nd I suspect that the impression th at the special epistemic role of phenomenal concepts is helpful in responding to the gap problem might stem from a failure to appreciate that considerably more is needed to at tain the latter explanatory goal than to reach the former. The assumption physicalists often make is that achieving the former aim and then simply citing the inductive evidence in favor of physical ism is enough to accomplish the latter objective. I think, however, that this assumption is mistaken. There is a residual sense of ad-hocness about strong necessities that Sturgeons appeal to special epistemic pr operties of phenomenal concepts is unable to dispel, and the reas on is that a posteriori physicalism is comm itted to two wildly different reduction methodologies. Standard proper ty reduction has a substantive a priori step which ultimately grounds the identification betwee n the reduced and the reducing property in the content and structure of the target concept. In the case of phenomenal co ncepts, the reduction is based exclusively on a posteriori considerations, and is entirely disconnecte d from the content of phenomenal concepts. Even worse, not only is it not grounded in the content of the target concept, but it appears effected in spite of what the target concept s uggests regarding the nature of its referent. Being hyperstable, a phenomen al concept refers to a certain property by expressing the very same property, and therefor e grasping such a con cept would normally put one in a position to know what property it refers to. Moreover, having a sp ecial epistemic role, a phenomenal concept affords a privileged access to its referent by virtue of not being applied on the basis of distinct canonical evidence. Ironically, a posteriori phys icalism entails that grasp of a phenomenal concept gravely misleads one a bout the nature of its referent. Different Kinds of Imagining/Co nceiving (Hill and McLaughlin) Hills main goal in his 1997 paper is to challenge the claim that what he calls separability intuitions (i.e., the intuition that, for example, there could be pain without C-fiber 140

PAGE 141

f iring and vice versa) constitute a good guide to possi bility. He maintains that such intuitions are due to a particular type of exercise of either the faculty of imagination or the faculty of conception. He argues that either way, because of some special features of this exercise, the separability intuitions it supplies are highly unr eliable. The distinction between imagining and conceiving that Hill makes use of is not explained very clearly. The idea is that the faculty of conception constructs representations that ar e largely conceptual and propositional, while those supplied by imagination are largely qu alitative (Hill, 1997, p. 72), in the sense of perception like. (He holds that when we attempt to form representations of possible situations, we often make use of both faculties.) Suppose that when one tries to construct a re presentation of an instance of pain without C-fiber firings, the source of ones representations of the presence of pa in and the absence of cC-fiber firings is the imagination. Why should we be wary of such representations and doubt that they latch onto a genuine possibility? Hills answer builds upon a suggestion made by Nagel. According to Nagel, this type of representati on should not be trusted because it combines the products of two very different types of imag ination. Conscious states are imagined via sympathetic imagination, whose exercise c onsists in putting onese lf in a state whose phenomenal character resembles that of the conscious state imagine d, while states like that of there being C-fibers firing are imagined via percep tual imagination, which is putting oneself in a state whose phenomenal character resembles in certain respects th at of perceiving the phenomenon in question. Hill amends Nagels story to accommodate the objection that brain states are not possible objections of percep tion. According to Hill, phenomena like brain processes can only be the object of indirect percepti on (I will label it perception*), by which he means either perceiving a certain kind of appa ratus that indicates the presence of C-fibers 141

PAGE 142

firing, and b eing aware that thats what it indicates (this is, by the way, clearly propositional!), or perceiving a model of that brain process as a model of it. The exercise of perceptual* imagination consists in putting oneself in a stat e which is relevantly similar to the state of perceiving* the phenomenon in question. One might wonder at this point what sympathe tically imagining the lack of pain amounts to, and likewise for perceptually* imagining th e lack of some brain process. Regarding the former, perhaps sympathetically imagining a phenomenal state opposed to pain in nature would do for some purposes. However, it is not clear how one could sympathetically imagine the complete absence of phenomenal states, whic h is part of imagining a zombie. As for perceptually* imagining the lack of a brain process, Hill says it is putting oneself in a state that fails to satisfy the list of condi tions laid out for perceptually imagining the presence of it, plus perhaps having the thought that there are no brain processes in the situation imagined. It is Hills view that whats responsible for the imaginability of pain without C-fibers firing and vice versa is the concerted operation of sympathetic and perceptual* imagination. It is important to note that in this paper Hill acknowledges that the mere discovery that separability intuitions are the re sult of splicing together the pr oducts of two different types of imagination does not suffice to call those intuitions into question. He argues that the reason we should not trust separability in tuitions is that a similar ki nd of "cooperation" between two different kinds of imagination--perceptual* and, this time, perceptual (instead of sympathetic)--is responsible for the notoriously unr eliable possibility intuitions elicited by the Kripkean cases. Therefore, we have good reasons to be suspicio us in the psychophysical case too. According to Hill, when we're imagining that water could fail to be H2O, and conclude such a thing is 142

PAGE 143

possible, we are perceptually im agining a case where something is water, and we're perceptually* imagining that that is al so a case of something that is not H2O. The joint operation of perceptual and "perceptual* imagination" certainly looks unreliable. The fault, however, appears to lie with "percep tual* imagination" itself. Given Hill's gloss on what "perceptually* imagining a th eoretical state" comes to, 'perceptually* imagining that state x obtains' looks more like a misnomer for imagini ng something quite different from state x itself namely, that certain facts obtai n that provide some (highly de feasible) evidence for supposing that x obtains. As an explanation of the mistaken possibili ty intuitions associated with the Kripkean statements, I find Hills story quite implausible. He pr esents the intuitions as rooted in a type of confusion that is neither likely nor interesting. I am also quite skeptical of the relevance of perceptual and perceptu al* imagining to conceivability ar guments against phys icalism. While it could be the case that somebody fi nds the statement that pain is not C-fiber fi ring conceivably true because of taking "p erceptually* imagining that p" to be a genuine case of imagining that p, that would be an uninteresting mistake, which no respectable dualist could possibly be suspected of. One could try to fix the underlying account of imagining that C-fibe r firing occurs/fails to occur, discard the putative anal ogy with the Kripkean examples, and claim instead that what makes intuitions about the psychophysical case suspicious is just their partial relia nce on the atypical "sympathetic" imagination. Such an approach looks similar to Papineaus argument about the antipathetic fallacy, as it too proposes to exploit a putative phenom enological similarity between a phenomenal state and the state of im agining that state. Given the overall dialectic, however, for such a strategy to engage the gap ar gument, it would have to be pursued with the 143

PAGE 144

goal of defending either non-ascriptivism or strong necessities, and it is unclear how either aim could be accomplished successfully.43 With respect to separability intuitions produced by our faculty of conception, Hill similarly argues that there are certain psychologi cal mechanisms that are responsible for them, and which should not be trusted because they ar e also the source of th e mistaken possibility intuitions about the Kripkean examples. Here is how these psychological mechanisms are supposed to operate: whenever there are two concepts that are not a priori connected, a separability intuition is generated regarding the properties picked out by the two concepts, provided that the subject is not in possession of any a posteriori evidence that the two concepts are coextensive. Hill says that is what is res ponsible for the mistaken intuitions attaching to the Kripkean examples, and maintains that we thus have a strong inductive argument for the unreliability of separability intuitions. (Actually, the proviso about lack of a posteriori evidence to the contrary should be eliminated for I don t think possessing such evidence is enough to prevent the relevant intuition a bout either pain or water.) This argument is by no means convincing. Quite curiously, Hill grants a little later in the paper that, by contrast with wat er, pain and the firing of C-fibers dont pick out what they do via contingent modes of presentation. Howe ver, since the reason why the psychological mechanism described by Hill yields the wrong in tuition about water is precisely that water picks out the chemical compound H2O via a contingent property of it (which invites a conflation of the perfectly possible s cenario in which the watery stuff around here is not H2O with the 43 Although the involvement of sympathetic imagination may be relevant to the gap intuition (though we have seen in the discussion of Papineau that its explanatory potential is even in that area very limited), it is most unclear how it could help explain the existence of strong necessities, i.e., how it could justify the discounting of some possible worlds when identifying the property determined by the primary intention of a phenomenal concept. 144

PAGE 145

im possible one in which water is not H2O), we have been provided no good reason to believe the separability intuitions shoul d likewise be distrusted. So, the important question Hill must answer is why the separability intuition is unreliable given that pain and the firi ng of C-fibers are not natural kind terms. Some things he says suggest he thinks that is because phenomenal and non-phenomenal concepts have very different associated justification conditions. Paradigmatic applications of phenomenal concepts are justified in virtue of the very state that they pick out (or self justifying, or beyond justification). By contrast, the experiences that are relevant to th e justification of applicat ions of neuroscientific concepts are of two kinds: experiences that confirm the theories in which those concepts are introduced and experiences that one has when one engages in indirectly perceiving neuroscientific states. The claim is that this di fference in justification conditions suffices to put into doubt separability intuitions. However, it is by no means clear why this should be deemed a satisfactory reason to doubt separability intuitions. First, Hill himself a ssumes in the paper that to call into doubt the reliability of the psychological mechanism that is responsible for the separability intuitions one must point out other circumstances in which the same mechanism is unreliable. But this is a mechanism that by definition applies only to the psychophysical case. Second, and more important, if both the concept of pain and the neur ological concept are ascriptive, and neither is a natural kind concept (both of which Hill appears to grant), it simply follows that its possible for each state to exist without the other, as long as that is conceivable, unless metaphysical possibility is narrower than logi cal possibility. Therefore, if H ill neither rejects ascriptivism nor endorses strong necessities, I do not see how he can avoid the conclusion that the separability intuitions are correct, no r how appeal to the different episte mic features of phenomenal concepts 145

PAGE 146

could help h im explain them away. On the ot her hand, if he is willing to either uphold nonascriptivism or embrace strong necessities, Hill owes us a story about the exact explanatory function fulfilled by the special epistemic role of phenomenal concepts with in the larger project of justifying either non-ascriptiv ism or belief in strong necessiti es. (The discussion of Sturgeons response to the gap problem from the previous se ction suggests, however, that one should not be too optimistic in this regard.) The 1999 paper Hill co-wrote with McLaugh lin provides a somewhat different response to the conceivability arguments, even though the authors refer the reader to Hills 1997 piece as if nothing of substance maintained there would be implicitly retracted. The tenet that separability intuitions are either the result of the collaboration of two widely different types of imagination or the product of concurrently empl oying two kinds of concepts with markedly different epistemic properties is still intended to pl ay an important role in thei r response to the conceivability arguments. However, the answer to the questi on why we should deem such a collaboration suspicious is different. This time there is no attempt to offer any examples of mistaken possibility intuitions the same (or a relevantly similar) kind of collaboration is responsible for. Rather, Hill and McLaughlin claim first that th e mere fact that the psychological mechanisms involved are different as indicated makes it clear that there should be no presumption in favor of the reliability of their joint exercise. However, I am not sure why that should be so. Presumably, they do not mean to imply that any modal intu ition grounded in the joint exercise of different kinds of imagination or in the employment of c oncepts with different as sociated justification conditions is doubtful. As Chalmers notes44, this strategy would help one explain away pretty much any modal intuitionthere are different epistemic constraints associated with color 44 (Chalmers, 1999, p. 486). 146

PAGE 147

concepts an d shape concepts, therefore, the intui tion that red squares are possible is unreliable. But then what kind of differences should ra ise ones suspicions exactly and why? It appears that it is the involvement in the alliance of sympathetic imagination or phenomenal concepts, with their peculiar episte mic properties, that is supposed to raise suspicions, and the idea seems to be that those are such that phenomenal properties would tend to appear to us distinct from phys ical properties even if they were not. This, however, brings us back to the strategies pursued by Papineau and Sturgeon, which have already been examined and evaluated in earlier secti ons of this chapter. From a dialectical angle, it is hard to pin down the exact way in which the Hill and McLaughlin strategy engages the gap argument. They argue that separability intuitions and hence the general conceivability-p ossibility principle should be discarded on the grounds that they are in radical conflict with common sense beliefs and the continuing commitments of rational enquiry (Hill & McLaughlin, 1998, p. 450) They maintain that denying physicalism leads to epiphenomenalism, which is highl y counterintuitive. Accepting psychophysical identities by contrast best expl ains the existence of psychophysical correlations, and is therefore commended by our longstanding methodological commitments to simplicity and overall coherence. There appear to be two options he re. On the one hand, this claim might rest on a general picture of the justification of theoretic al identifications which I have rejected as misguided, as on my generalized derivability acc ount of the explanatory ga p, ordinary theoretical identifications are justified by vi rtue of being deducible from D, rather than on the basis of methodological considerations pertaining to thei r explanatory potential. On the other hand, one could suggest that while ordinary theoretical identifications ar e justified via entailments, for psychophysical identity statemen ts the special nature of ps ychophysical concepts precludes 147

PAGE 148

entailm ent, and they are justified instead via an argument to the best explanation. (Hill and McLaughlin do not say enough to make it clear how they understand scientific reduction in the uncontroversial cases, and therefore it is not fully clear which of the two options they find congenial.)45 In this latter case, it seems that Hi ll and McLaughlin do not carry the discussion much further than Papineau and Sturgeon. Phenomenal Concepts as Recognitional (Loar) Loar rejects what he calls the semantic premise (SP), which he deems to be a common assumption of the knowledge argument and the gap argument: (SP) Two conceptually independent concepts can pick out the same property only if at least one of them expresses a contingent mode of presentation. Note that, if ascriptivism is a given, rejecting (SP) amounts to postulating strong necessities. Loars explicitly formulated goal is to give an account of phenomenal con cepts that helps explain how (SP) could be erroneous. In effect, Loar ap pears to think that the only counterexamples to (SP) are certain pairs of concepts one member of which is always phenomenal. Therefore, it is incumbent upon him to supply an account of phenomenal concepts that explains why phenomenal concepts are the only ones well suited to cause trouble for (SP). Loar claims that phenomenal concepts be long to the larger ca tegory of recognitional concepts. Recognitional concepts have the form x is one of that kind; they are type demonstratives. These type-demonstratives are gr ounded in dispositions to classify, by way of perceptual discriminations, certain objects, events, situations (Loar, 1990, p. 600). The recognitional abilities that such concepts are grounded in depend on no consciously accessible 45 In section II of their paper, Hill and McLaughlin declare that they are not committed to strong necessities, but it is rather clear that they misunderstand the notion. They claim that metaphysical and logical possibility are one and the same thing, and reject instead the princi ple that everything conceivable is logically possible, which is in fact trivial, when properly understood. That suggests that Hill and McLa ughlin misdescribe their position and might actually be stuck with strong necessities, unless, of cour se, they would rather reject ascriptivism. 148

PAGE 149

analysis into com ponent features; they can be irreducibly ge stalt (Loar, 1990, p. 601). He argues that, in addition, phenomenal concep ts pick out the propert ies that they do dir ectly, that is, not by way of contingent modes of presentations (L oar, 1990, p. 602). They are claimed to differ in this respect from all other recognitional concep ts, which connote contingent modes of presentation. The crucial question that arises is how referring dire ctly, or not via a co ntingent mode of presentation, should be understood. Two ways of unpacking this come to mind. Phenomenal concepts do not express contingent modes of pr esentation because they express necessary modes of presentation (i.e., are expres sible by hyperstable terms in L), or because they do not express modes of presentation at all. The latter propos al can be understood in two ways: phenomenal concepts do not have any content (radical non-ascriptivism), or, while they do have some content, it fails to determine a full-blown m ode of presentation of the property picked out (moderate non-ascriptivism). Note that the Krip kean intuitions about the semantics of pain which Loar cites in an attempt to mitigate the objection that it is ad hoc to maintain that phenomenal concepts are the only recognitional c oncepts that do not express contingent modes of presentationonly support the firs t proposal. While it is not entire ly clear which one of these one should attribute to Loar (rad ical non-ascriptivism appears not to be an option), (Loar, 1990) gives the impression it might be the first proposal that he is leaning toward. (Loar, 1999) suggests that even more strongly. This interpreta tion is also supported by Loars claim that not connoting contingent modes of presentati on makes phenomenal concepts unique among recognitional concepts, for, presumably, if he had thought that phenomenal concepts did not connote contingent modes of presentation in the se nse that they lacked (full-blown) ascriptive content, this would have been precisely on the grounds that they are recognitional. 149

PAGE 150

How exactly is the tenet that phenomenal concepts are recognitional and, in addition, special among recognitional concepts in one of these senses supposed to undermine (SP)? If phenomenal concepts are non-ascriptive (either in th e weak, or the strong sense), then (SP) is clearly undermined since phenomenal terms simp ly do not express (full-blown) modes of presentation. However, it is hard to see how the claim that phenomenal concepts are recognitional could lend support to non-ascriptivism about phenomen al concepts, whether of the radical or the moderate variety. On the cont rary, if phenomenal concepts are grounded in recognitional abilities of the sort Loar alludes to, then one w ould expect for there to be primary intensions a priori associ ated with such concepts. What if phenomenal concepts are thought of as ascriptive? Well, Lo ar would then claim that they express the same property as they pick out (as the dualist hims elf typically insists) which in my terminology comes down to sayi ng that phenomenal terms are hyperstable. MoreoverLoar seems to be arguingsince physica lism entails that phenomenal terms pick out physical properties, it follows that a phenomenal term and the physical term that is co-referential with it both pick out and express the same phys ical-functional property (Loar assumes that the relevant physical term is also hyperstable). At this juncture it might seem as though Loars strategy runs up against the following objection, due to Chalmers: What reason do we have to expect a phenomenal term to pick out a physical -functional property? Chalmers expects Loar to cite in response to this cha llenge the recognitional nature of phenomenal concepts and to note that it is quite typical of rec ognitional concepts to turn out to pick out the same property as certain theoretical concepts. The problem, accordi ng to Chalmers, is that typically recognitional concepts do not express the same property as the one they pick out, and that looks exactly like the feature that enables a recognitional concept to refer to the same property as some theoretical 150

PAGE 151

concept. For exam ple, that kind of plant picks out the type of cactus that happens to cause a perceptual representation of a specific type. Ther efore, Chalmers concludes that Loars strategy is self-defeating as it conjoins two tenets that undermine each other: if phenomenal concepts are special among recognitional concepts in that they are hyperstable, then their being recognitional cannot be used anymore to support the view that they refer to physical properties. I suspect, however, that Loar invokes the putative recognitional nature of phenomenal concepts to attain a more modest explanatory goal than Chalmers supposes. Moreover, quite apart from Loars particular intentions, a physic alist might well be inclined to construe the explanatory function of the hypothesis that phen omenal concepts are recognitional in such a more modest fashion as I will now delineate. Th e claim that phenomenal properties are physical is often viewed by pysicalists as considerably well supported via an argument to the best explanation. From this methodological angle, it s eems that all that recour se to the recognitional view of phenomenal concepts is intended to accomplish is expl aining why the gap problem does not succeed in overturning the inductive evidence for physicalis m. More specifically, as physicalists tend to understand the dialectic, what is in need of explanation is (assuming ascriptivism) the unique status of psychophysical identities as s trong necessities, or, in other words, why, even though accepting physicalism about the mind has the consequence that phenomenal concepts supply the only counterexam ples to (SP), that does not undermine the epistemic credentials of the physicalist position. If Loars explanatory strategy is interpreted along these lines, the charge leveled by Chalme rs against it appears to miss the mark. Suppose that Loar resorts to the theory of recognitional concepts to develop an argument of the following sort. Outside the psychophysical cas e, independent concepts pick out the same property only if at least one of them expresse s a contingent mode of presentation of that 151

PAGE 152

property. One should not, however, expect the sam e thing to hol d for psychophysical identities. Consider true identity statements that empl oy ordinary recognitional concepts and corresponding coextensive non-recognitional concepts, such as that type of tree is an oa k. In sharp contrast with such ordinary identity statements, in whic h at least the recognitional concept expresses a contingent mode of presentation of the property picked out (if not both concepts, as is the case in the example at hand), a psychoph ysical identity statement like p ain is the firing of C-fibers contains two predicates that express hyperstable concepts. (Loa r assumes that the theoretical concept is hyperstable, which I will conced e for the sake of the argument. Regarding phenomenal concepts, as we saw, he grants that they are hyperstable, a hi ghly intuitive semantic point which he regards as ontologically neutral, a nd one the physicalist pict ure is perfectly able to accommodate.) Phenomenal concepts are un ique among recognitional concepts in being hyperstable, and, assuming physicalism, that is wh at accounts for the nece ssity of the primary intention of pain is the firing of C fibers. On the other hand, psychophysical identities appear to be the only counterexamples to (SP) among theore tical identifications, bu t this observation does not carry much weight for it ignores the recogn itional character of phenomenal concepts. Despite their expressing the same property, pain and the firing of C fibers are conceptually independent (and therefore the identity is a posteriori) because recognitional concepts ar e always conceptually independent from theoretical concepts; they do not describe their referent in theoretical terms, but are grounded in sheer recognitional abilities a nd can be pure Gestalt.46 Is this explanation successful in removing the ad-hoc air of the physicalist tenet that psychophysical identities, even t hough a posteriori, have a necessary primary intension? I submit 46 On the interpretation I am proposing, what the hypothesis that phenomenal concepts are recognitional is intended to help explain is not so much their corefere ntiality with theoretical concepts, as their conceptual independence from such. 152

PAGE 153

that it is not, because it does not even begin to dispel the mystery surrounding the putative strong necessities. Recognitional concepts may very well be in general conceptually independent from theoretical concepts, but if they are grounded in recognitional ab ilities, then they quite clearly are associated with a mode of presentation which yi elds for each logically possible (centered) world considered as actual the objects (i f any) that trigger the recognitio nal abilities in question. Since the relevant domain of the primar y intension is th e collection of logically possible (centered) worlds, the value of the function for some argument s will consist in a state that feels like pain even though the subject at the center of the world is not simultaneously in a C-fiber firing kind of state. Has the physicalist given us any good reason for discounting such worlds as not genuinely possible in a metaphysical sense? It seems not. To sum up, the result established in the presen t section is that Loa rs view of phenomenal concepts as recognitional does not seem apt to either support non-ascriptivism about phenomenal concepts or help explain why phsychophysical identity statements should be expected to enjoy the unique status of strong necessities. The Prospects for A Posteriori Physicalism and Neutral Monism Cha mpions of physicalism like to emphasize that the issue between them and dualists can only be settled on a posteriori grounds (irreducibly mental states would be merely contingent existents after all, which makes it unlikely that a ny purely a priori argument can be offered either for or against them), and, when asked about the evidential support in favor of physicalism, they typically tend to cite either an argument from th e causal closure of the phy sical or an argument to the best explanation, where the explanandum is constituted by the pervasive psychophysical correlations we all agree obtain. Without in a ny way aiming at an exhaustive examination of these arguments, briefly rehearsing them will help us bring into sharper focus the dialectic situation. 153

PAGE 154

The f irst prominent argument for physicalism47 starts from the uncontroversial premise that mental phenomena and brain phenomena are sy stematically correlated. Physicalism offers a very elegant and economical explanation of psyc hophysical correlations in that it postulates that phenomenal states are either identi cal with or realized by brain st ates. For instance, if being in pain just is a functional state re alized, say, in humans, by a certain pattern of neural firings, then there is no wonder that whenever a human e xperiences pain, the pattern in question is identifiable in her brain, and vi ce versa. Physicalism is superior to dualism as an explanatory hypothesis because, first of all, it posits fewer th ings. To explain why the mental and the physical are always correlated in specific ways, dualism postulates the existence of a separate ontological layer of mental properties, which is someth ing over and above the physical domain. The physicalist, by contrast, identi fies each phenomenal property with some physical-functional property. Second, to accommodate the observed ps ychophysical correlations, the dualist needs to countenance fundamental laws of nature that co nnect the phenomenal to the physical realm. The physicalist, however, has no need for psychophysical laws that hold with mere nomic necessity. Therefore, it seems that when Ockhams razor is applied, physicalism turns out to be more economical as regards both the types of propert ies posited and the basic laws countenanced. The two explanatory virtues delineated s eem to apply quite clearly to a priori physicalism, but the question of inte rest to us is whether they au tomatically extend to a posteriori physicalism, as that has been the brand of phys icalism under scrutiny in the current chapter. Before answering the question, let us take a quick look at the ot her major argument for physicalism. 47 See, for instance, (Melnyk, 2003) for a detailed exposition of this argument. 154

PAGE 155

The argum ent from the causal closure of th e physical, which wa s already succinctly discussed in Chapter 3, goes something like this. The impressive success of the physical sciences is taken to substantiate the so-c alled principle of the causal cl osure of the physical, which is notoriously difficult to state, but for our limited purposes in this section, the following formulation will do: any physical event that has a cause at t has a phys ical cause at t. To that the highly intuitive claim is added that the mental is causally efficacious, in the sense that some mental events cause physical events, as well as the tenet that there is no overdetermination (e.g., of physical events by physical and mental events). The conclusion derived from these premises is that mental events are physical. The options available to the dualist, when faced with the argument under discussion, are rather clear: either embrace epiphenomenalism, or reject the causal closure premise, or uphold overdetermination. Each of the first two options raises complex questions of its own, which I will set aside in what follows, however. If we gr ant for the sake of the argument that the mental is causally efficacious and the phys ical is causally closed, the causal closure argument can be reframed as an inference to the best explan ation along the following lines: physicalism about phenomenal states is the best explanation of phenomenal states causi ng physical phenomena, under the assumption that the physica l is causally closed, for the onl y relevant alternative is the hypothesis that some physical events are overdeter mined, and that would constitute an inferior explanation. Not only would it pos it more properties and more basi c laws than physicalism, but it would also entail that there is pervasive redundancy in how the world works. Both arguments to the best explanation that I have sketched on behalf of the physicalist put a high premium on parsimony. A theoretical hypothesis that postulates fewer things and fewer basic laws of nature and avoids positing causal mechanisms that account for events for 155

PAGE 156

which we already have alternativ e uncontroversial causa l explanations counts, other things being equal, as a better explanation. This is the re spectable m ethodology that science is built on, and the philosopher of mind prides herself on followi ng in the scientists footsteps when trying to solve the mind-body problem. However, the crucial question is, of course, whether other things really are equal. What the disc ussion of the explanatory gap pr oblem has brought to light is, I think, is precisely that they are not. There are se veral different criteria th at govern theory choice when scientists resort to infe rence to the best explanation,48 such as accounting for more data, managing to compellingly unify domains previously regarded as disparate and being simpler. Physicalism may be able to account for the highly intuitive claim that the mental is causally efficacious without paying the price of positing su i-generis mental properties and a surplus of causal mechanisms, but, while commendable in it self, ontological parsimony comes at a high cost in the case at hand, and, ir onically, the cost is one in und erstanding. By postulating that the phenomenal is not something over and above the physical, physicalism renders the relationship between the mind and the body epistemically opaque The unification of the mental and the physical domains which physicalism effects has an air of artificiality about it, and the higher ontological simplicity of the phys icalist hypothesis requires learning how to live with a certain amount of irremovable mystery. Suppose that both the causal closure of the phys ical and the causal efficacy of the mental are shown to be unassailable theses. They quite cl early are not that, as th ere are, on the one hand, (at least prima facie) reasons to resist the inference by enumerat ive induction taken to support the former, and, on the other hand, ways of arguing th at epiphenomenalism can account for all the data suggesting that the mental is causally e fficacious, but let us imagine that we were in 48 See, for instance, (Lipton, 2004). 156

PAGE 157

possession of irrefutable evidence for both theses W e would be left then with two major explanatory hypotheses of the causal efficacy of the mental: a dualism that upholds overdetermination and a physicalism that identifies conscious states with physical states but is unable to dispel the intuitive puzzlement about how consciousness could possibly be physical. Which one looks like the better explanation? It se ems to me implausible to opt for the latter. What the gap problem brings out is that the range of explananda that need to be accounted for by the physicalist is broader than one might have tho ught. In particular, it turns out that one of the things that needs to be explai ned is the strong necessity status of psychophysical identities, and when that is fully appreciated, it becomes doubtful that the physic alist hypothesis really ranks as the best explanation. The pr ice the physicalist pays for a un itary ontology is a bifurcated methodology. Ontological parsimony comes at the cost of methodological extravagance, for physicalism postulates two radica lly different reduction mechanis ms: one for all the standard cases of property reduction, which is grounded in the content of the concept via which the reduced property is grasped, and a very diffe rent mechanism applicable only to mental properties, which is entirely disconnected from the target concept and, because of that, is bound to seem ad-hoc. If we take seriously the methodological principles brandished by the physicalist, and hold him accountable to the explanatory sta ndards he upholds, physicalism seems to fail (on its own terms, as it were). In the remainder of the section, I will invest igate the implications of the gap problem for neutral monism. Consider the conjun ction of the following four theses: a. Physical theory only provides information a bout functional properties of things, and reveals absolutely nothing about their intrinsic nature. b. Metaphysically, consciousness is not something over and above these intrinsic properties of what we call matter. c. The intrinsic properties of matte r are not themselves phenomenal. d. The intrinsic non-phenomenal properties ar e not graspable by any possible thinker. 157

PAGE 158

How does the conjunction of a) through d) fare which on my account of the physical qualifies as neutral m onism, and which meets with a lot of praise from philosophers such as Chalmers? First of all, one might worry that the view appears highly undermotivated. What are the reasons why one should favor neutral monism over idealism? Why accept that the intrinsic nature of matter is inscrutable not just to us but to any possible thinking being? At this point one is typically offered some perfunctory argument th at the only intrinsic prope rties that we have a direct grasp of are phenomenal. But how does it follow that the only intrinsic properties that any possible cognizer could form a non-relational conception of must be phenomenal? And if assuming that the intrinsic nature of matter is no n-phenomenal really leads to the conclusion that it is in principle unknowable, why is not it more ra tional to embrace idealism? After all, it is not like neutral monism, which appe als to mysterious, ungraspable basic non-phenomenal properties can shed any light on the explanatory question as to how there can be consciousness in the world. Since neutral monism is explanatorily silent with respect to consciousness, we might as well take consciousness as ontologically ba sic (especially sin ce there is nothi ng about phenomenal concepts themselves that could force us to treat phenomenal properti es as ontologically derivative), and try to explain othe r things in terms of it, which ar e quite plainly not ontologically fundamental and so must be built out of something else. What Chalmers, who is otherwise the fiercest enemy of property reduct ion claims that are explanatorily opaque, says in defe nse of neutral monism is that its lack of explanatory power when it comes to accounting for consciousness cannot really be held agai nst it because of the special nature of the intrinsic properties of matter. One might also object that we do not have any conception of what protophenomenal properties might be like, or of how they c ould constitute phenomenal properties. This is true, but one could suggest that this is merely a product of our ignorance. In the case of familiar physical properties, there were principled reasons (based on the character of 158

PAGE 159

physical con cepts) for denying a constitutive c onnection to phenomenal properties. Here, there are no such principled reasons. (Chalmers, 2003a, p. 139) Since one cannot form any positive non-relational conception of protophenomenal properties it is illegitimate to expect the way in which c onsciousness is constituted by them to become explanatorily transparent. Once you accept the my sterious intrinsic ungraspable features of matter, which you are supposed to have good reasons to do, all bets are off. Chalmers seems to think that since, accordi ng to this position, the intrinsic nature of matter is in principle ungraspable, and hence my sterious, the tenet that the intrinsic ungraspable properties of matter constitute not only the physi cal but also consciousness comes at no further explanatory cost. However, it seems confused to assume that this extra tenet brings in no additional mystery. We do not ha ve any model for how intrinsic phenomenal properties could be constituted by more basic conceptu ally heterogeneous in trinsic properties, and it doesnt quite do to just point in respon se to the ungraspable nature of the propertie s that are supposed to constitute the supervenience base. It is impor tant to remember what the motivation was for neutral monism in the first place: namely, a clai m about the content of sc ientific terms, which allegedly capture only disposi tional properties (and th erefore, by contrast with phenomenal concepts, are not hyperstable). Consequently, ne utral monism postulates a radical explanatory asymmetry between the physical an d the phenomenal. Despite the inconceivable nature of the categorical non-phenomenal properties, the claim that the physical is constituted by them is fully intelligible. It brings along no further mystery b ecause the structure of physical concepts makes it perfectly clear how it might turn out that the physical was made up of these ungraspable properties: physical concepts are functional/fun ctional description co ncepts. The phenomenal case is much murkier quite apart from the nebulous nature of the supervenience base. We have here, just as we do in the case of a posteriori physicalism, a rela tionship of constitution that is 159

PAGE 160

unique in its explanatory opacity, and it is not clear what justifies Chalmerss different treatment of the two views. What this reveals is that neutral monism is not simply undermotivated when compared to idealism. Rather, it is in a much worse condition, as it faces se rious intelligibility problems. Chalmerss stance on the relative strength of ne utral monism and a posteriori physicalism seems rather irrational. If psychophysic al identities/supervenience cl aims are deemed problematic because of their unique status as strong necessities, how could the neutral monist conception of what makes up the phenomenal, which is also uni que in its inscrutabil ity, fare any better? 160

PAGE 161

161 LIST OF REFERENCES Balog, K. (1999). Conceivability, Po ssi bility, and the Mind-Body Problem. Philosophical Review, 108, 497-528. Bealer, G. (2002). Modal Epistemology and the Ra tionalist Renaissance. In T. Gendler & J. Hawthorne (Eds.), Conceivability and Possibility Oxford University Press. Beckermann, A, Flohr, H. & Kim, J. (Eds.) (1992). Emergence or Reduction? Prospects for Nonreductive Physicalism De Gruyter. Biro, J. I. (1991). Consciousness and Subjectivity. In E. Villanueva (Ed.), Consciousness. Ridgeview. Biro, J. I. (1993). Consciousness and Objectiv ity. In M. Davies and G Humphreys (Eds.), Consciousness: Psychologic al and Philosophical Essays Blackwell. Bishop, R. (2005). Patching Physics and Chemistry together. Philosophy of Science 72, 710-22. Blackburn, S. (1992). Filling in Space. Analysis, 52, 62-3. Block, N. (2006). Max Black's Objection to Mind-Bra in Identity. In T. Alter & S. Walter (Eds.) Phenomenal Concepts and Phenomenal K nowledge: New Essays on Consciousness and Physicalism Oxford University Press. Block, N. & Stalnaker, R. (1999). Conceptual Analysis, Dualism, and the Explanatory Gap. Philosophical Review 108, 1-46. Block, N., Flanagan, O. & Gu zeldere, G. (Eds.) (1997). The Nature of Consciousness MIT Press. Carruthers, P. (2004). Reductive Expl anation and the "Explanatory Gap". Canadian Journal of Philosophy, 34(2), 153-74. Chalmers, D.J. (1996). The Conscious Mind: In Search of a Fundamental Theory. Oxford University Press. Chalmers, D.J. (1999). Materialism and the Metaphysics of Modality. Philosophy and Phenomenological Research, 59, 473-93. Chalmers, D.J. & Jackson, F. (2001). Concep tual Analysis and Re ductive Explanation. Philosophical Review, 110(3), 315-61. Chalmers, D.J. (2002). Does Conceivability enta il Possibility? In T. Gendler & J. Hawthorne (Eds.) Conceivability and Possibility Oxford University Press. Chalmers, D.J. (2003a). Consciousness and its Place in Nature. In S. Stich & F. Warfield (Eds.) The Blackwell Guide to the Philosophy of Mind Blackwell.

PAGE 162

Chalm ers, D.J. (2003b). The Content and Epistemol ogy of Phenomenal Belief. In Q. Smith & A. Jokic (Eds.) Consciousness: New Philosophical Essays Oxford University Press. Chalmers, D.J. (2004). Epistemic Two-Dimensional Semantics. Philosophical Studies 118,153226. Chalmers, D.J. (2005). Phenomenal Concepts an d the Knowledge Argument. In P. Ludlow, Y. Nagasawa, & D. Stoljar (Eds.) There's Something About Mary MIT Press. Chalmers, D.J. (2006a). The Foundations of Two-Dimensional Semantics In M. GarciaCarpinte ro & J. Macia (Eds.) Two-Dimensional Semantics: Foundations and Applications. Oxford University Press. Chalmers, D.J. (2006b). Two-Dimensional Sema ntics. In E. Lepore & B. Smith (Eds.) The Oxford Handbook to the Philosophy of Language Oxford University Press. Chalmers, D.J. (2007). Phenomenal Concepts and the Explanatory Gap. In T. Alter & S. Walter (Eds.) Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism Oxford University Press. Chalmers, D. (2009). The Two-Dimensional Argument Against Materialism In B. McLaughlin, (ed.), Oxford Handbook of the Philosophy of Mind Oxford University Press. Crook, S. & Gillett, C. (2 001). Why Physics Alone Cannot Define the Physical': Materialism, Metaphysics, and the Formulation of Physicalism Canadian Journal of Philosophy 31, 333-60. Cunningham, B. (2001). The Reemergence of 'Emergence' Philosophy of Science 68, 63-75. Devitt, M. & Sterelny, K. (1999). Language and Reality: A n Introduction to the Philosophy of Language. Cambridge: MIT Press. Dretske, F. (1981). Knowledge and the Flow of Information MIT Press. Evans, G. (1985). The Causal Theory of Names. in Martinich, A. P. (ed.) The Philosophy of Language Oxford Uni versity Press. Feigl, H. (1958). The `Mental' and the `Physical'. Minnesota Studies in th e Philosophy of Science 2:370-497. Reprinted (with a postscript) as The `Mental' and the `Physical' University of Minnesota Press, 1967. Fodor, J. (1990). A Theory of Content and Other Essays. MIT Press Gendler, T. & Hawthorne J. (Eds.) (2002). Conceivability and Possibility Oxford University Press. Gillett, C. & Loewer, B. (Eds.) (2001). Physicalism and its Discontents. Cambridge University Press. 162

PAGE 163

Gillett, C. & W imer, G. (2001). A 'Physical' Need: Physicalism and the Via Negativa Analysis 61, 302-9. He mpel, C. (1969). Reduction: Ontological a nd Linguistic Facets, in S. Morgenbesser, et al (Eds.) Essays in Honor of Ernest Nagel New York: St Martin's Press. Hempel, C. (1980). Comments on Go odman's Ways of Worldmaking, Synthese, 45, 139. Hendry, R. (2008). Chemistry. In S. Psillos & M. Curd (Eds.), The Routledge companion to philosophy of science New York: Routledge. Hill, C.S. (1997). Imaginability, Conceivab ility, Possibility, and the Mind-Body Problem. Philosophical Studies, 87, 61-85. Hill, C. S. & McLaughlin, B. P. (1998). There are Fewer Things in Reality than are dreamt of in Chalmers' Philosophy. Philosophy and Phenomenological Research, 59(2), 445-54. Jackson, F. (1982). Epiphenomenal Qualia. Philosophical Quarterly, 32, 127-36. Jackson, F. (1998). From Metaphysics to Ethics Oxford University Press. Kim, J. (2006). Physicalism or Something Near Enough. Princeton University Press. Kripke, S.A. (1980). Naming and Necessity Harvard University Press. Levine, J. (1998). Conceivability and the Metaphysics of Mind. Nous, 32, 449-80. Levine, J. (1983). Materialism a nd Qualia: The Explanatory Gap. Pacific Philosophical Quarterly, 64, 354-61. Levine, J. (1993). On leaving out what it's lik e. In M. Davies and G. Humphreys (Eds.) Consciousness: Psychologic al and Philosophical Essays Blackwell. Levine, J. (2001). Purple Haze: The Puzzle of Consciousness Oxford University Press. Levine, J. (2006). Phenomenal Concepts and the Ma terialist Constraint. In T. Alter & S. Walter (Eds.) Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism Oxford University Press. Lewis, D. (1970). How to Define Theoretical Terms. Journal of Philosophy, 76, 427-46. Lewis, D. (1972). Psychophysical and Theoretical Identifications. Australasian Journal of Philosophy, 50, 249-58. Lipton, P. (2004). Inference to the Best Explanation (Second Edition) New York: Routledge. Loar, B. (1997). Phenomenal Stat es (second version). In N. Bloc k, O. Flanagan, & G. Gzeldere (Eds.) The Nature of Consciousness MIT Press. 163

PAGE 164

Loar, B. (1999). David Chalm ers' The Conscious Mind'. Philosophy and Phenomenological Research, 59, 464-71. Loewer, B. (1996). A Guide to Naturalizing Se mnatics. In C. Wright & B. Hale (Eds.) Companion to the Philosophy of Language. Blackwell. Ludlow, P., Nagasawa, Y. & Stoljar, D. (eds) (2004). There's Something about Mary: Essays on Phenomenal Consciousness and Frank Jackson's Knowledge Argument MIT Press. Manson, N. (2002). Consciousness-Dep endence and the Explanatory Gap. Inquiry, 45, 521-40. McLaughlin, B. P. (1992). The Rise and Fall of British Emergentism. In A. Beckermann, H. Flohr, & J. Kim (Eds.) Emergence or Reduction? Prospect s for Nonreductive Physicalism De Gruyter. McLaughlin, B. P. (1997). Emergence. In R. Wilson & F Keil (Eds.) MIT Encyclopedia of Cognitive Sciences, 266-68. MIT Press. Melnyk A. (1997). How to Keep the 'Physical' in Physicalism. Journal of Philosophy, 94 (12): 622-637. Melnyk, A. (2003). Some Evidence for Physicalism In S. Walter (Ed.) P hysicalism and Mental Causation: The Metaphys ics of Mind and Action. Thorverton UK: Imprint Academic. Metzinger, T. (ed.) (1995). Conscious Experience Ferdinand Schoningh Press. Montero, B. (1999). The Body Problem, Nos, 33, 183-20. Montero, B. (2001). Post-Physicalism. Journal of Consciousness Studies, 8, 61-80. Montero, B. & Papineau, D. (2005). A Defence of the Via Negativa Argument for Physicalism Analysis 65, 233-37. Nagel, T. (1979). Mortal Questions. Ca mbridge University Press. Nagel, T. (1986). The View from Nowhere. Oxford University Press. Papineau, D. (1998). Mind the Gap. Philosophical Perspectives, 12, 373-89. Papineau, D. (1993). Philosophical Naturalism Oxf ord: Blackwell. Papineau, D. (2002). Thinking about Consciousness. Oxford University Press. Papineau, D. & Spurrett, D. (1999). A Note on the Completeness of 'Physics' Analysis 59, 2529. Perry, J. (2001). Knowledge, Possibility, and Consciousness MIT Press. Poland, J. (1994). Physicalism: the Philosophical Foundations. Oxford: Clarendon Press. 164

PAGE 165

Putnam H. (1975). The Meaning of "Meaning". In K. Gunderson (Ed.) Language, mind and knowledge, 131--93. Minneapolis: Univers ity of Minnesota Press. Scerri, E. (2007). Reduction and Emergence in Chemistry: Two Recent Approaches. Philosophy of Science, 74. 920-31. Searle, J. (1982). Proper Names and Intentionality. Pacific Philosophical Quarterly 63, 205-25. Searle, J. (1997). Proper Names. In P. Ludlow (Ed.) Readings in the Philosophy of Language. Cambridge, Mass.: The MIT Press. Searle, J. (1999). Intentionality. An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press. Shear, J. (Ed.) (1997). Explaining Consciousness: The Hard Problem MIT Press. Shoemaker, S. (1999). On David Chalmers' The Conscious Mind Philosophy and Phenomenological Research, 59, 539-44. Soames, S. (2004). Reference and Description: The Case Against Two-Dimensionalism Princeton University Press. Sterelny, K. (1983). Natural Kind Terms. Pacific Philosophical Quarterly 64, 110-25. Sturgeon, S. (1994). The Epistemic View of Subjectivity. Journal of Philosophy, 91, 221-35. Tye, M. (2000). Consciousness, Color, and Content MIT Press. White, S. (1986). Curse of the Qualia. Synthese, 68, 333-68. W ilson, J. (2006). On Characterizing the Physical. Philosophical Studies, 131, 61-99. Yablo, S. (1999). Concepts and Consciousness. Philosophy and Phenomenological Research, 59(2), 455-63. Yablo, S. (2002). Coulda, Woulda, Shoulda. In T. Gendler & J. Hawthorne (Eds.) Conceivability and Possibility Oxford University Press. Yablo, S. (2000). Textbook Kripkeanism and the Open Texture of Language. Philosophical Quarterly, 81(1), 98-122. 165

PAGE 166

BIOGRAPHICAL SKETCH Ana-Maria Andrei completed he r undergraduate studies in Rom ania, where in 1998 she received a BA in Philosophy from Bucharest Un iversity. In 2002, she was awarded an MA in philosophy, and in 2009 an MA in German Studies both by the University of Florida. In 2009, she started pursuing a Ph.D. in German Studi es at Cornell University. In August 2010, she completed a Ph.D. in philosophy at th e University of Florida.