E-Physicalism: A Physicalist Theory of Phenomenal Consciousness 9783110325560, 9783110324990

Citation preview

Reinaldo J. Bernal Velásquez E-Physicalism A Physicalist Theory of Phenomenal Consciousness

PHENOMENOLOGY & MIND Herausgegeben von / Edited by Arkadiusz Chrudzimski • Wolfgang Huemer Band 14 / Volume 14

Reinaldo J. Bernal Velásquez

E-Physicalism A Physicalist Theory of Phenomenal Consciousness

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliographie; detailed bibliographic data is available in the Internet at http://dnb.ddb.de

North and South America by Transaction Books Rutgers University Piscataway, NJ 08854-8042 [email protected] United Kingdom, Ire, Iceland, Turkey, Malta, Portugal by Gazelle Books Services Limited White Cross Mills Hightown LANCASTER, LA1 4XS [email protected]

Livraison pour la France et la Belgique: Librairie Philosophique J.Vrin 6, place de la Sorbonne ; F-75005 PARIS Tel. +33 (0)1 43 54 03 47 ; Fax +33 (0)1 43 54 48 18 www.vrin.fr

2012 ontos verlag P.O. Box 15 41, D-63133 Heusenstamm nr. Frankfurt www.ontosverlag.com ISBN 978-3-938793-170-2 2012 No part of this book may be reproduced, stored in retrieval systems or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use of the purchaser of the work Printed on acid-free paper ISO-Norm 970-6 This hardcover binding meets the International Library standard Printed in Germany by CPI buchbücher GmbH

To my parents, Reinaldo and Olga María, my sisters, Mariacatalina and Andrea, and Arianna.

TABLE OF CONTENTS ACKNOWLEDGEMENTS INTRODUCTION.................................................................................. 1 CHAPTER 1: A PHYSICALIST APPROACH TO CONSCIOUSNESS................................................................................ 7 1.0 Introduction................................................................................... 7 1.1 Consciousness and “what it is like”............................................. 7 1.2 Realism about consciousness ..................................................... 13 1.2.1 Eliminativism.......................................................................... 14 1.2.2 The problem of consciousness................................................ 17 1.3 Physicalism about consciousness ............................................... 19 1.3.1 Physical items ......................................................................... 20 1.3.2 Supervenience......................................................................... 21 1.4 The mind/body duality ............................................................... 23 1.5 Minds and bodies ........................................................................ 28 1.5.1 Mind-body causal relations .................................................... 30 1.5.2 Epiphenomenalism ................................................................. 31 1.5.3 Monism................................................................................... 35 1.6 Panpsychism ................................................................................ 36 1.6.1 The combination problem....................................................... 39 1.6.2 Panpsychism and neuroscience .............................................. 42 1.7 Conclusion ................................................................................... 44 CHAPTER 2: STRONG AI AND COMPUTATIONALISM.......... 47 2.0 Introduction................................................................................. 47 2.1 Strong AI ..................................................................................... 48 2.1.1 Conscious behaviour............................................................... 51 2.1.2 Consciousness and behaviour................................................. 53 2.2 Computational functionalism .................................................... 56 2.2.1 Functionalism and consciousness........................................... 59 2.2.2 Functions and information...................................................... 66 2.3 The internal character of consciousness ................................... 70 2.4 Conclusion ................................................................................... 73 CHAPTER 3: SUBJECTIVITY AND THE UNITY OF CONSCIOUSNESS.............................................................................. 75 3.0 Introduction................................................................................. 75

2

3.1 Higher-order representation theories of consciousness........... 77 3.2 The unity of consciousness ......................................................... 84 3.3 The explanatory gap ................................................................... 90 3.3.1 Subjectivity and reality ........................................................... 95 3.3.2 Intersubjective accessibility.................................................... 98 3.4 The homunculus fallacy............................................................ 101 3.5 Conclusion ................................................................................. 105 CHAPTER 4: SUPERVENIENCE, EMERGENCE, AND ONTOLOGICAL NOVELTY .......................................................... 107 4.0 Introduction............................................................................... 107 4.1 Supervenience and ontology..................................................... 108 4.2 Consciousness and supervenience ........................................... 116 4.3 Microphysicalism ...................................................................... 119 4.4 The levels of reality ................................................................... 124 4.5 Emergence.................................................................................. 128 4.6 The case for emergence............................................................. 131 4.6.1 The kinetic theory of gases ................................................... 134 4.6.2 The EPR Paradox and Bell’s theorem .................................. 137 4.6.3 Microphysicalism and nonlocality........................................ 149 4.7 Emergence laws ......................................................................... 152 4.8 Conclusion ................................................................................. 155 CHAPTER 5: E-PHYSICALISM AND CONSCIOUSNESS ........ 157 5.0 Introduction............................................................................... 157 5.1 Experiencers .............................................................................. 157 5.2 The possible worlds strategy .................................................... 159 5.2.1 Physical and metaphysical possibility .................................. 160 5.2.2 The conceivability of possible worlds .................................. 161 5.3 Metaphysical zombies ............................................................... 165 5.4 Kim’s “supervenience argument” ........................................... 171 5.5 Conclusion ................................................................................. 175 CHAPTER 6: E-PHYSICALISM AND PHENOMENAL QUALITIES........................................................................................ 177 6.0 Introduction............................................................................... 177 6.1 Vehicles and content – an analogy........................................... 179 6.2 Phenomenal content and representational content................ 184 6.2.1 One vehicle, different contents ............................................. 186 6.2.2 One content, different vehicles ............................................. 188 6.2.3 Nonrepresentational phenomenal contents ........................... 192

3

6.3 Phenomenal character and physical states............................. 195 6.3.1 Phenomenal character and biological functions ................... 198 6.3.2 Phenomenal space................................................................. 200 6.4 The knowledge argument......................................................... 203 6.4.1 Mary does not acquire new knowledge ................................ 203 6.4.2 The ability hypothesis........................................................... 205 6.4.3 The limits of scientific knowledge ....................................... 207 6.5 Conclusion ................................................................................. 210 CONCLUSION .................................................................................. 213 REFERENCES................................................................................... 217

i

ACKNOWLEDGEMENTS This book is a revised version of a PhD thesis I defended in the Institut Jean-Nicod, in Paris, the 5 of December of 2011. The supervisor was Max Kistler, and the jury was composed by Ned Block, Jérôme Dokic, Pierre Jacob, David Papineau and Jaime Ramos. I would like to thank Max Kistler for his generous and precious support, guidance and feedback. I greatly appreciate the possibility I had of working under his supervision, and enjoyed every discussion we had. I am very grateful to the members of the jury. I was pleased to have the opportunity to discuss with these great philosophers, and to receive detailed and insightful commentaries on my work. My thanks also to the members of the Institut Jean-Nicod from October 2008 to December 2011. In particular: To professors François Recanati—who motivated me to join Jean-Nicod and established the contact with Max Kistler, Joëlle Proust—with whom I worked for three years in the research group in Metacognition, Jérôme Dokic, and Claudine Tiercelin. To the postdocs Conor McHugh, Kirk Michaelian, and Frank Esken. To the PhD students Santiago Echeverri, Pierre Grialou, and Felipe Nogueira de Carvalho. To the master students Santiago Arango and Rafael Quintana. Also contributed to the development of this work, through seminars, informal chats, or with their friendship, Cornelius Mauer, Ewa Stasiak, Cintia Retz Lucci, Marie Guillot, Chiara Chelini, Ophelia Deroy, Francesca Ervas, Markus Kneer, Eduardo García, Hady Ba, Gregory Bochner, Hugo Mercier, Magali Seille, Marion Renaud, Daniela Tagliafico, Giulia Piredda, Hugo Viciana, Neftali Villanueva, Sacha Bourgeois-Gironde, Muriel Cahen, Denis Vinçon, Marie-Christine Nizzi, Margherita Arcangeli, Inga Vendelin, Delphine Blitman, Benoît Conti, Marina Trakas, Joulia Smortchkova, Barry Smith, Dave Wripley, Adrian Briciu, Dan Zeman, Anna Loussouarn, Michael Murez, Jean-Marie Chevalier, Paul Egré, Fabian Bernache, David Landais, and Anne Coubray. I am particularly grateful to Pierre Grialou, for his precious help with a thoroughly revision of the final draft of the thesis. To Arianna Uggè, for all the support, encouragement, patience, and feedback she

ii

gave me, in addition to an outright survey of this revised version. To Conor McHugh, who read a draft of Chapter 1 and provided extremely insightful and detailed comments. And to Santiago Echeverri, for several informal and delightful discussions we had on different philosophical topics; I learned a lot from him and admire his philosophical insight. I am indebted to: The Institut Jean-Nicod, its director Pierre Jacob, Vincent Godefroy and Sophie Bilardello; the Département de philosophie - UPMF (Grenoble II), its director Denis Vernant, and Loredana Truong; the IHPST (Institut d’histoire et philosophie des sciences et des techniques), its director Jean Gayon, and Sandrine Souraya; the École Doctorale de Philosophie - Université Paris 1, its director Chantal Jaquet, and Nicole Saint-Charles. Thanks to them I obtained the financial support to participate in several colloquiums in different places of Europe, where I presented some of the arguments contained in this work. During these events I had the opportunity to discuss, in particular, with Galen Strawson, David Papineau, David Rosenthal, Adrian Cussins, and Ted Honderich. Many thanks to them. Above all, I am immensely thankful to my parents, Reinaldo Bernal González and Olga María Velásquez de Bernal, and to my sisters MaríaCatalina and Andrea, for the extraordinary support they have given me in every respect.

1

INTRODUCTION The so-called “mind/body problem” has occupied a central place throughout the history of philosophy. On one hand, human beings are biological organisms. Their bodies are material entities, thereby subject to the laws of nature. But on the other hand, human beings have a mind: They are “rational”, they have feelings and emotions, and they have subjective perspectives on the world. They have a mental life that seems to evade the rigidity of the physical world. The mind/body problem concerns the relation between minds and bodies. Prima facie, these seem to have different metaphysical natures. But then, how it is possible for the mind to interact with the body? And if the mind is something physical, or the physical is something mental, why they seem to be so different? Most modern philosophers were “dualists”: They considered that the mind and the body belong to different metaphysical categories. Some were “monists”—they claimed that mind and body belong to the same category—but usually they took the body to be some kind of mental entity, and not the other way around—they were “idealists”. Unfortunately, from a contemporary perspective, these philosophers were not able to support their convictions through a compelling solution of the mind/body problem. It was especially during the 20th century that “physicalist” monism (or “materialism”), i.e., the idea that everything that exists has a physical nature, acquired many adherents. Physicalists are confident that the existence of the mind and its activity is, somehow, a natural phenomenon, and explore this possibility to its last consequences. Certainly, interesting and enlightening proposals have been put forward about how to “naturalise” mental phenomena, i.e., about how to account for them in a physicalist framework. Moreover, scientific research has provided useful empirical data and relevant theories in the areas of brain sciences and psychology. However, physicalism has received some compelling criticisms. It still faces the challenge of providing a persuasive solution for the mind/body problem.

2

“The problem of consciousness” is one of the aspects of the mind/body problem. It concerns the question of the nature of subjective experience and its relation with objective phenomena. Human beings happen to be such that there is something it is like to be one of them (Nagel 1974). Experiences like tasting wine, listening to music, looking to a painting, feeling cold, and feeling anxious, have a distinctive “whatit-is-like-ness” or “phenomenal character”. In this sense, a subject that is experiencing is said to be in a “phenomenally conscious” mental state. The problem of consciousness is about this type of states. Until the seventies, most of the work in the analytic philosophy of mind was focused on “the problem of intentionality”: The fact that thoughts and words are about something else (Brentano 1874). Intentionality was taken as the distinctive characteristic of the mental, and thus as the core of the mind/body problem. The main questions were to determine what mental states are, what they represent, how they come about to represent something and, in general, how intentionality is possible. Subjective experience and, in particular, its phenomenal character, was not a primary topic. Certainly, a distinction between “conscious” and “unconscious” (or subpersonal) mental states was in place. Some mental states of a subject—the unconscious ones—were considered to be inaccessible, in one sense or another, to the subject himself. But there was no clear distinction between phenomenal consciousness and other notions of consciousness. And, more important, a comprehensive account of the contrast between conscious and unconscious mental states was not considered to be crucial for an understanding of the nature of the mind. Behaviourism, in particular, dismissed the question of consciousness. Within this view, which until the late fifties provided the main theoretical framework in psychology, it was common to consider a discussion about subjective experience as close to nonsense. The realm of “the subjective” was taken, if not as fiction, as a pseudo-scientific category. Science was only concerned with what is directly “observable”. Everything “mental” had to be reduced in terms of behaviour. During the sixties, with the advent of computationalism, the idea that mental states can be reduced in terms of behaviour was abandoned.

3

In some sense or to some extent, mental states were considered to be internal states. But even though there was, consequently, a place for subjective experience in a theory of the mind, consciousness continued to be a secondary topic. The central question was to determine how mental representations are codified and processed in cognitive systems. Certainly, computationalism had to account for the distinction between conscious and unconscious mental states. But, firstly (and not surprisingly), this distinction was considered to be mainly functional, i.e., a question of access among different mental states or modules. The property of a mental state being phenomenally conscious was not clearly distinguished from its functional properties or role. Secondly, if a mental state had a phenomenal content, this characteristic was considered to be irrelevant for the functional role it could play. But principally during the last three decades the interest in subjective experience and phenomenal consciousness increased. The fact that some mental states are phenomenally conscious is now taken as primordial for the understanding of the mind. The problem of intentionality continues to be central, and there is no general agreement about how to naturalise it. But it seems that the “hard problem” (Chalmers 1996) par excellence is to account for phenomenal consciousness. Indeed, some philosophers claim that consciousness is required for intentionality (e.g., Searle 2002), and others that it plays an essential role for the fixation of the reference of perceptual states (e.g., Campbell 2002). Much philosophical work, with contributions by many of the most prominent philosophers of mind, is been done nowadays on the question of the nature of phenomenal consciousness and the relation between subjective experience and objective reality. This work advances a theory in the metaphysics of phenomenal consciousness that I label “e-physicalism”. It is grounded on the convictions that subjective conscious experience—in the sense of Nagel (1974)—is a real phenomenon, and that some variant of physicalism ought to be true. In Chapter 1, firstly, I elaborate the notion of phenomenal consciousness following Block’s (2007) distinction between access consciousness and phenomenal consciousness. Secondly, I argue for realism about consciousness by contrast with eliminativism. It is not

4

possible to prove that consciousness is a real phenomenon, but neither can eliminativists prove that it is not. For the realist, consciousness is given as a brute fact. Thirdly, I argue that given the mind-body problem, and despite our dualist intuitions, a physicalist monism is the most reasonable metaphysics. Accordingly, I claim that there is a property X, which is a physical property or a supervenient (on the physical) property, such that for an entity S to be conscious is for S to instantiate X. Finally, I criticise panpsychism and conclude that consciousness is a property of some complex physical entities. Chapter 2 concerns Strong AI and computational (or “machine”) functionalism about consciousness. Both take consciousness to be a supervenient property and thus are compatible with physicalism. But I argue, firstly, that the behaviour of an entity S supervenes on a base that includes not only S but also physical systems other than S, and secondly, that a function realised by some hardware H is not an intrinsic property of H. By contrast, consciousness has an “internal character”: It is an intrinsic property of the conscious entity. Therefore, I conclude that consciousness is neither a behavioural nor a functional property and thus I reject both Strong AI and functionalist views. In Chapter 3, firstly, I argue that higher-order representation theories of consciousness (HOR) fall short as accounts of the existence of phenomenal consciousness. The occurrence or possibility of a higherorder mental state M’ representing a mental state M is not sufficient to account for the fact that there is something it is like to be in M. Secondly, I discuss the unity of consciousness (Bayne 2010) and, primarily, “phenomenal unity”. I claim that any theory, and in particular higher-order thought (HOT) theories, must account for this unity; it stands for one of the essential characteristics of subjective experience. Finally, I discuss the “explanatory gap” (Levine 1983). I suggest that the gap appears, at least in part, when we take the subjectivity of consciousness as an ontological condition and not as an epistemological one. The exclusively subjective access there is to phenomenal contents can be explained by the very particular nature of the epistemological relation holding between a subject and his own mental states. Thus, the property of having phenomenal content can be objective despite the subjectivity of phenomenal experience.

5

Chapter 4 is the core of the work. I argue that consciousness does not supervene on physical items, but is a physical property of the conscious entity that emerges from its fundamental constituents. The emergence of properties is conceived as resulting with nomological necessity from the emergence base, and emergent properties are thought as not reducible to fundamental items and endowed with causal powers of their own. This thesis—the “e-physicalism” view—is in conflict with “microphysicalism”, i.e., with the idea that every property of a complex physical system supervenes on fundamental items. Therefore, I argue against microphysicalist metaphysics, and show the plausibility of the emergentist view I advance, through the elaboration of two examples— one in classical physics and one in quantum mechanics. My argument does not show that consciousness is an emergent property, but opens this possibility. The metaphysics of e-physicalism gives a plausible framework for a realist and physicalist view on consciousness that avoids a commitment to panpsychism. In Chapter 5, firstly, I criticise the strategy of using the “conceivability” of a metaphysical world to drive metaphysical conclusions. To determine whether a “world” is metaphysically or physically possible is a nontrivial and uncertain matter. Secondly, I reject—on the base of e-physicalism—Chalmers’ (1996) “zombie argument”. I conclude that an exact physical replica of the actual world cannot be “a zombie world”, and throw doubts about its very metaphysical possibility. Thirdly, I show that Kim’s (2005) “supervenience argument” does not threaten the thesis that consciousness has “original causal powers”, i.e., causal powers that are not reducible to the ones of the fundamental constituents of its emergence base. The e-physicalism view avoids, in particular, the tension between vertical determination and horizontal causation. Chapter 6 concerns phenomenal character and qualia. Its purpose is not to advance a thoroughly elaborated account of phenomenology, but just to make explicit the commitments and consequences of ephysicalism for this difficult question, and to provide the grounds for a further development of the theory. I try to make plausible the idea that qualia, which I define as the ingredients of phenomenal contents, are physical properties. First, I argue that phenomenal content is different

6

from representational content. It can have the function of representing, and in this case the representational content it conveys is nonconceptual. But it can also comply with nonrepresentational functions. Secondly, I suggest that consciousness has biological functions that result from natural selection, and I sketch a model of “phenomenal space”, i.e., of the structure of the phenomenal character of conscious experiences, in order to illustrate in what sense phenomenal properties could be physical properties. Thirdly, I address Jackson’s (1982) “knowledge argument”. I agree that the what-it-is-like-ness of having a given experience can only be known by having the experience, as the argument assumes. However, I argue that this does not prove physicalism to be false. Physicalism is compatible with the idea that not everything that can be known about natural phenomena can be captured in scientific theories. In particular, scientific theories cannot capture phenomenal contents since these are not propositional contents, but nonconceptual ones. The objectives of this work do not include a historical synthesis of the discussion about consciousness, or a recapitulation of the totality of influential arguments that have been given in different directions. I discuss some views, many of them in the most general form, some of them more in detail, as they become relevant as I advance, step by step, in the discussion and elaboration of e-physicalism. I expect some of the arguments I present to be original to some extent and, even though I advance some controversial conclusions, I hope that the view put forward is at least coherent.

7

CHAPTER 1: A PHYSICALIST APPROACH TO CONSCIOUSNESS 1.0 Introduction I will start with an attempt to clarify the concept “phenomenal consciousness”, and provide a definition based on the “what it is like” notion introduced by T. Nagel (1974). Secondly, I will discuss what reasons we have to believe that some entities are phenomenally conscious. I shall discuss the eliminativist approach, and endorse a realist stance on consciousness. Thirdly, I will introduce a form of physicalism, and defend a physicalist conception of the mental and, in particular, of consciousness. I will argue that the acknowledgement of the existence of mind-body causal relations gives good reasons to believe that mental states (including conscious mental states) are physical states or supervene on physical states. I will reject dualist and panpsychist views of consciousness. 1.1 Consciousness and “what it is like” The terms “conscious” and “consciousness” are used in several ways. Someone can be conscious in the sense of being awake, or being aware of something. In the present work I will focus on one particular sense of “consciousness”, which was introduced by Nagel in his seminal paper “what is it like to be a bat?” He wrote: But fundamentally an organism has conscious mental states if and only if there is something it is like to be that organism—something it is like for the organism. We may call this the subjective character of experience. (1974, p. 436. Italics in the original) N. Block labelled this notion of consciousness “phenomenal” (Pconsciousness), by contrast with “access consciousness” (Aconsciousness). He says: The paradigm P-conscious states are sensations, whereas the paradigm A-conscious states are “propositional attitude” states like

8

thoughts, beliefs, and desires, states with representational content expressed by that-clauses. (2007, p. 281. Italics in the original) For Block, there are three main differences between P-consciousness and A-consciousness. Briefly: (1) […] P-conscious content is phenomenal, whereas A-conscious content is representational. (2) […] A-consciousness is a functional notion, and so Aconscious content is system-relative: what makes a state Aconscious is what a representation of its content does in a system. P-consciousness is not a functional notion. (3) […] there is such a thing as a P-conscious type or kind of state. For example, the feel of pain is a P-conscious type—every pain must have that feel. But any particular token thought that is Aconscious at a given time could fail to be accessible at some other time […]. (2007, pp. 280-281. Italics in the original) There are important and controversial questions about the relations between P-consciousness and A-consciousness. But the major point to highlight is that, while A-consciousness is a functional notion, and thus the property of a mental state being A-conscious is relative to some cognitive architecture, P-consciousness refers to a constitutive property of some mental states. Besides the distinction between phenomenal consciousness and access consciousness, D. Rosenthal proposes a further one between “mental state consciousness” and “creature consciousness”. He says: In one use, we speak of mental states as being conscious or not conscious. Mental states, such as thoughts, desires, emotions, and sensations, are conscious if we are aware of them in some intuitively immediate way. But we also apply the term ‘conscious’ to the creatures that are in those mental states. A creature’s being conscious consists, roughly, of its being awake and being mentally responsive. (2005, p. 46)

9

Incorporating Rosenthal’s distinction, I will say that: (Def) An entity (or “creature”) is P-conscious if and only if she is in some P-conscious mental state. A mental state is P-conscious if and only if there is something it is like to be in that mental state for the corresponding entity. I will label “the consciousness property” the property a P-conscious entity or mental state has in virtue of which it is P-conscious. When necessary, I will talk of the “c-consciousness” property to refer to the consciousness property of an entity (or creature), and of the “mconsciousness” property to refer to the consciousness property of a mental state. The conception of phenomenal consciousness advanced by Nagel has been widely accepted in contemporary analytic philosophy. I propose to contract Def, keeping in the background the distinctions between “P-consciousness” and “A-consciousness”, and between “mental state consciousness” and “creature consciousness”, into the following shorter version: (D) An entity is (phenomenally) conscious if and only if there is something it is like to be that entity. Now, let us define “phenomenal content” and “phenomenal character” as follows:1 (PC1) The phenomenal content of a (P-conscious) mental state M of a subject S is the what-it-is-like-ness of being S in virtue of S’s being in M. (PC2) The phenomenal character of the experience of a (Pconscious) subject S is the what-it-is-like-ness of being S. 1

Usually the expressions “phenomenal content” and “phenomenal character” are used interchangeably, and irrespective of whether one refers to (1) the what-it-islike-ness of being in a mental state M or to (2) the what-it-is-like-ness of undergoing a given experience.

10

Thus, phenomenal contents and phenomenal characters are taken to be the members of the class of the possible what-it-is-like-ness of being a P-conscious entity. Note that, given Def, the phenomenal character of the experience of a P-conscious subject S results from the phenomenal content of her P-conscious mental states. By “qualia” I will refer to the constituents of phenomenal contents and characters. Qualia are types, which individuate phenomenal contents and phenomenal characters. Thus, differences in the what-it-is-like-ness of being in two (conscious) mental states M and M’, are due to differences in the qualia that constitute the phenomenal contents of M and M’; analogously for phenomenal characters.2 Intuitively, the idea expressed by D seems clear enough. Statement D says that to be a conscious3 entity is to be an entity that has experiences, e.g., of pain, colour sensations, and fear. But what kind of statement—a priori or a posteriori—is D? I will argue that it is a priori. In fact, I think D is a definition that captures the meaning of the expression “phenomenal consciousness”: The property of a subject being phenomenally conscious is the property of there being something it is like to be that subject. According to S. Kripke (1972)4 if a statement is not informative, i.e., if it has no epistemological import, it is a priori.5 I will claim, thereby, that D is a priori because it is not informative. In fact, I will argue that D gives no empirical test to classify entities between conscious and unconscious ones. To be sure, if I believe there is something it is like to be me, D enables me to judge that I am conscious; 2

I will elaborate on qualia in Chapter 6. Hereafter when using “conscious” or “consciousness” without qualification I shall be referring to the phenomenal notion. 4 Kripke (1972) argues that the dichotomy “a priori/a posteriori” corresponds to epistemological conditions, whereas the dichotomy “necessity/contingency” corresponds to metaphysical ones. The epistemological distinction is orthogonal to the metaphysical. The statement “a triangle has three sides” is a priori and necessary, but the statement “I am here” is contingent despite being also a priori. 5 But the converse is not true, because a statement can be a priori and nevertheless informative. For instance “the standard meter stick is one meter long” is a priori but informative, since it gives an empirical criterion for classifying objects by length (by comparison with the meter stick). 3

11

if I believe there is something it is like to be you, D enables me to judge that you are conscious; and if I believe there is nothing it is like to be a stone, D enables me to judge that a stone is not conscious. Nevertheless, D does not provide any empirical test to establish if an entity is conscious. It just clarifies the meaning of “phenomenal consciousness”. Certainly, it might seem, prima facie, that D entails an experiencebased test to determine if an entity S is conscious. The test would proceed as follows: (T) On the base that there is something it is like to be S, a subject W concludes that S is conscious and, conversely, on the base that there is nothing it is like to be S, a subject W concludes that S is not conscious. However, the following analysis suggests that T does not give any empirical test—not even when a subject wants to establish if he (himself) is conscious. There are four possible situations in which T could be applied: (1) The subject S is conscious and wants to know if S is. (2) The subject S is not conscious and wants to know if S is. (3) The subject W is conscious and wants to know if another subject S is. (4) The subject W is not conscious and wants to know if another subject S is. Under conditions 3, T does not provide any empirical test. Since, trivially, every subject is numerically distinct from any other subject, by means of following T subject W cannot find out if S is conscious. The fact that there is something it is like to be W could at best be used by W to extrapolate, based on some similarity between W and S, that there is something it is like to be S. In order for W to empirically find out if S is conscious, he needs to appeal to some theory that provides third-person

12

point of view criteria for the attribution of consciousness. For instance, a theory that offers behavioural, functional, or biological criteria. Under conditions 2, T cannot be applied. To be able to use T, S needs to have fixed the reference of the property appearing in the antecedent of the conditional, i.e., a property instantiated by him and represented by “something it is like to be S”. But, ex hypothesi, there is not such a property. Therefore, S cannot fix this reference. Note that it is not the case that S, as a consequence of applying T to himself, will establish that there is not something it is like to be him. In condition 2— as was the case in condition 3—T does not give any test for the attribution of consciousness. Subject S needs to appeal to some empirical theory proposing third-person point of view criteria. Only by applying such a theory to himself, S would probably determine that he lacks the property of being conscious.6 Condition 4 combines difficulties present in conditions 2 and 3. Subject W needs to appeal to an empirical theory proposing third-person point of view criteria to eventually establish if S is conscious. The only remaining condition under which T could be applied is 1. Prima facie, it seems that at least in this case T could be exercised and give the correct outcome. However, note that T cannot be applied in case 2 and then, when applied in case 1, it would (if successfully applied) always produce the same outcome, namely, that S is conscious. Because T cannot give at least two different results, it follows that T does not provide any empirical test at all. In sum, D presents a definition of consciousness; it does not entail an empirical test to attribute (or self-attribute) consciousness. Clearly, for a subject W to determine whether another entity S is conscious he needs to appeal to some third-person point of view criteria. Concerning the self-attribution of consciousness, in the next section I will argue that the fact that S is conscious is given as a brute fact to S. Even though D lacks epistemological import, I think it is a useful statement. First, it serves to clarify the concept of phenomenal 6

I say that S would “probably” establish that he is not conscious, because it is highly doubtful that S could ever grasp the meaning of “phenomenal consciousness”, and it is controversial whether it is possible for a subject to believe a proposition he does not understand. See, e.g., Recanati (1997).

13

consciousness and distinguish it from other concepts of consciousness— in particular from functional concepts. Second, and more important, it serves to delineate two metaphysical categories: To the first—the one of conscious creatures—belong entities such that there is something it is like to be one of them (they instantiate what I labelled “the consciousness property”), and to the second—the one of unconscious creatures—belong entities that are not so (they do not instantiate the consciousness property). Therefore, D can be used to assert something about the actual world: That there are (or that there are not) phenomenally conscious entities. If there are (or there are not) conscious entities, this is a fact regardless of the epistemological difficulties we face when we want to establish if it obtains. Despite not being informative, D is conceptually valuable. Eliminativists claim there are no conscious entities—the first category is empty. Panpsychists claim that every entity, however small, is conscious (at least to some degree)—the second category is empty. I believe there are entities in both categories. I will propose a form of realism. 1.2 Realism about consciousness I will defend a form of realism about consciousness. I believe there is something it is like to be me; I believe there is something it is like to be another human being; and I believe there is something it is like to be a member of some animal species. I cannot prove the eliminativist to be wrong but, as many philosophers claim, I think the reality of subjective experience should be acknowledged. It has proven very hard to give a satisfactory answer to the question of what are we talking about when discussing phenomenal consciousness. There are tough, and may be insurmountable, conceptual difficulties. But I think the reality of consciousness can hardly be denied. For instance, take pain. Whatever its metaphysical nature, it is quite “real” in some strong sense. So real, that whenever someone is suffering, say, a chronic headache, the feeling eclipses the existence of the external world. Indeed, pain is so real that it can even lead to suicide when unbearable. To be sure, there are psychoneuronal processes going on whenever someone is in pain, and these processes certainly play a

14

biological function in the organism. But the fact that someone is suffering pain is as real as any fact can be.7 In short, I take consciousness to be a real phenomenon, even though we might end up following Block when he suggests answering the question “what is consciousness?” the way Louis Armstrong answered when asked “what is jazz?”: “If you got to ask, you ain’t never gonna get to know.” (2007, p. 73). Accordingly, I consider “the problem of consciousness” to demand an account of the metaphysical nature of the property “being conscious”, and of how it is related to other properties in the actual world. This is the subject of the present work. D. Chalmers calls this “the hard problem of consciousness” and urges us to take it seriously, since “we are surer of the existence of conscious experience than we are of anything else in the world.” (1996, p. xii). 1.2.1 Eliminativism Certainly, our beliefs about the contents of our own phenomenal experiences, e.g., about what it is like to see a landscape, to hear a tune, to taste chocolate, or to have a headache, can be wrong. We are usually biased by folk psychology and theorising might be misguided by the received and long lasting dualist ontology.8 Now, based on this kind of remarks, the eliminativist constructs arguments to claim that even though it seems to me that I have phenomenal experiences, with content I believe to self-know, this is an illusion.9 For him not only our beliefs concerning the contents of our phenomenology are wrong, but there is indeed no phenomenal content or character; to think otherwise is to be tricked. For instance, J. Searle describes eliminativism, when referring to D. Dennett’s view, by saying: “He thinks there are no such things as […] the feeling of pain. He thinks there are no such things as qualia, subjective experiences, first-person phenomena, or any of the rest of it. Dennett agrees that it seems to us that there are such things as qualia, but this is a matter of a mistaken judgement we are making about what really happens.” (1997, p. 99. Italics in the original). 7 8 9

In Chapter 3 I will argue that the subjectivity of conscious experience is an epistemological condition that does not force us to deny its objective reality. See Papineau (2011). For instance Dennett (1991) and Humphrey (2006) argue this way.

15

The counter-argument against eliminativism is based on a classical idea: Concerning experience, the dichotomy reality/appearance does not apply. In Searle’s words: “[W]here consciousness is concerned the existence of the appearance is the reality.” (1997, p. 112. Italics in the original). Regarding external objects we may distinguish between the noumenon (the thing “in itself”) and the way the object (phenomenally) appears to the subject. There could be noumena in the absence of any mind, and a noumenon can be perceived under different appearances. Indeed, one can even hallucinate there is an “object” being perceived. But concerning phenomenal experiences we cannot distinguish between “the experience in itself” and “the way the experience appears”. First, there are no experiences independently of an experiencing subject. To say that there is something appearing to a subject requires the subject to be having an experience. No “illusion” can threaten the existence of the very experience. It is one thing to say that one can form false beliefs about what the contents of one’s experience are, and another to say that one can be wrong about the fact of being undergoing an experience with some content. Secondly, a given experience cannot “appear” under different forms. No “illusion” can mask or distort the phenomenal character of the experience. The appearance is the phenomenal character of the experience: Experiences are individuated by “appearances”, i.e., by the “what-it-is-like-ness” to undergo a given experience. Nevertheless, the eliminativist might insist that the only fact to be explained is that of believing there is consciousness with such and such phenomenal characters. If we had an explanation for this, she argues, we would be done, since this is the only fact that can be scientifically studied.10 For instance, Dennett proposes to study consciousness following the method of “heterophenomenology”. In short, it consists in taking as data the reports given by the subjects that describe their (illusory) phenomenal experiences, and to look for neuropsychological explanations of the contents of these reports. The most significant characteristic of heterophenomenology is the adoption of a third-person point of view, which excludes from the data how things (are believed to)

10

See Dennett (1991; 2005).

16

look “from the inside”.11 For Dennett, only what is intersubjectively available (verbal reports, body actions, etc.) constitutes data. I agree there is this fact that requires an explanation: The fact that I believe there is something it is like to be me. But I think this is a further fact, related, but different from the fact that there is something it is like to be me (if indeed there is). An explanation of the fact that I am conscious does not amount to an explanation of the fact that I have some belief. My conscious experience might (partly) explain the belief that I am conscious, but not the other way around. Indeed, I think that in order for there to be something it is like to be S, S is not required to have the capacity to form beliefs about her experiences, nor to be able to report them.12 The disagreement between realists and eliminativists can be formulated this way: The realist thinks that the fact that there is consciousness is a given brute fact; by contrast, the eliminativist takes the assertion that there is consciousness to be a theoretical (and false) judgement. Searle—a realist—says: “The subjective feelings are the data that a theory of consciousness has to explain […] Daniel Dennett’s book13 […] denies the existence of the data.” (1997, p. 99. Italics in the original). In short, the realist takes the existence of consciousness as the starting point for the construction of metaphysical and scientific theories of the mind, while the eliminativist considers the only data to be the fact that one might assert there is consciousness, thereby manifesting a belief.14 How can we settle the issue? How can I convince the eliminativist that I am (and she is) conscious? To claim privileged access is begging the question against her. Conversely, how could the eliminativist show me to be wrong? How could she prove there is nothing it is like to be 11 12

13 14

For an interesting critique of the idea of giving an epistemological privilege to the third-person point of view over the first-person one, see Nagel (1986). To believe that one is conscious reveals, according to many theorists, the existence of the self-consciousness faculty, which presupposes consciousness but is not identical to it. The relation between consciousness and self-consciousness is a very interesting (and difficult) question, but a comprehensive discussion is outside the scope of the present work. The referred book is Dennett (1991). See Dennett (1991; 2005).

17

me? For I take the reality of experience as something given, as the brute fact to be explained, and not as a theoretical hypothesis which requires verification. Who has the burden of proof? The realist says the eliminativist does, and the eliminativist claims the opposite. In some respects, the discussion between the realist and the eliminativist about consciousness resembles the classical problem of skepticism about the existence of the external world. The solipsist considers that no argument proves the existence of the external world, while her opponent claims that solipsism, even if tenable, is obviously false. The solipsist is only certain of her existence (whatever she is); her opponent takes the existence of the external world as granted. In sum, it seems there is no conclusive argument that could convince the eliminativist that consciousness is something real. Likewise, the realist can hardly be persuaded to abandon her conviction. This shows that the problem of consciousness is not just an empirical question but also a deep metaphysical one. 1.2.2 The problem of consciousness There is the widespread opinion that consciousness puts forward a scientific problem like any other. It is often claimed that “the mystery of consciousness” will disappear as it happened, e.g., with the former “mystery of life”. For instance: - Dennett says “There are many properties of conscious states that can and should be subject to further investigation right now, and once we get accounts of them in place, we may well find that they satisfy us as an explanation of what consciousness is. After all, this is what happened in the case of the erstwhile mystery of what life is.” (2005, p. 178. Italics in the original). - F. Crick, when asked if consciousness poses a really difficult problem, says: “Well of course that’s what people say. That’s what they said about life. They said there was a vital spirit that you couldn’t explain in terms of physics or chemistry, and because they said it, it became almost a standard point of view. […] It’s an analogy, and the history of the vital spirit, or the élan vital, shows

18

us that you have to be cautious.” (in Blackmore 2005, p. 69. Italics in the original). - P. S. Churchland says: “There are lots of examples where people were convinced that one problem was unsolvable […] and they turned out to be wrong […]. So consider, for example, the perihelion of mercury […]. People think that because we don’t understand how consciousness is produced in brains, this must be telling us something really deep and interesting.” (in Blackmore 2005, pp. 51-52). Certainly, there are in the history of science some problems that were considered to be insoluble but later received satisfying scientific explanations (or at least they do not look so mysterious any more). In particular, the problem of life was considered to be unexplainable by natural science, and the “élan vital” and “entelechy”, unacceptable for a scientific theory, were postulated.15 Later on, with the development of molecular biology, it became conceivable that matter could produce living beings and thus the entelechy and élan vital hypotheses were abandoned. Similarly, a materialist can claim that the mystery of consciousness, despite being very complex, can (in principle) be solved by natural science. Consciousness is, after all, another natural phenomenon. However, I think it is a mistake to consider that consciousness poses an ordinary scientific problem, or that the difference with other issues is just a question of degree (the brain being a very complex system). I believe the analogy between the task of explaining photosynthesis and that of explaining consciousness (Dennett 1991) disregards the most important characteristic of the latter: The subjectivity of experience. Notice that, in domains not directly concerning consciousness, arguments involving subjectivity are not relevant. Even if, metaphysically, consciousness were a natural phenomenon like life (and I believe it is), it is hardly deniable that it poses tough and specific 15

In Carnap (1966, pp. 12-16) there is a classic critique about entelechy. Carnap argues why the entelechy theory is not acceptable as a scientific one.

19

metaphysical and epistemological problems. Witness the “conceivability arguments” (Kripke 1972; Chalmers 1996) and “the knowledge argument” (Jackson 1982). These arguments are not based on the assumption of a dualist ontology; on the contrary, they start from the supposition that physicalism is true—they take consciousness to be a natural phenomenon—and yield uncomfortable consequences. Notice that among scientific research only the study of consciousness crucially involves data that come from an exclusively subjective form of access: The one that each subject has towards his own phenomenal experiences.16 No “neural correlate”17 can be established without taking into account the subjective point of view at some stage of the research (even if reports of conscious experience are analysed from a third-person perspective). At the end, it is the very existence and properties of these subjective experiences what we want to understand. This epistemological peculiarity of the problem of consciousness makes it substantially different from any other scientific problem. 1.3 Physicalism about consciousness I will defend a physicalist view of consciousness, which I label “ephysicalism”, since I believe that consciousness is a natural phenomenon. By “physicalism” I refer to a metaphysical position that says that all the entities inhabiting the actual world, their properties, and all the facts and events involving them, have a physical nature.18 In Kim’s words “The core of contemporary physicalism is the idea that all things that exist in this world are bits of matter and structures aggregated out of bits of matter, all behaving in accordance with the laws of physics 16

17

18

McGinn (1989) claims that phenomenal consciousness has an exclusive firstperson form of epistemological access, while science always operates from a third-person perspective. Then—he concludes—even if consciousness is a natural phenomenon it is out of the explanatory scope of science. Nagel (1974; 1986) gives similar remarks. Naturalists consider that whenever there is a psychological process there is simultaneously a neural one (the discussion concerns the explanation of this relation). The latter is the “neural correlate” of the former. “Physicalism” is used to label different programmes. Historically the term comes from Logical Empiricism, which used “semantic physicalism” for the doctrine saying that the meanings of all words can be reduced (or analysed) in terms of nonintentional vocabulary.

20

[…].” (2005, pp. 149-150). Something has “a physical nature” in the following sense: (PN) An item T has a physical nature if it is a physical item, or it supervenes on a set of physical items. What I call “an item” is any kind of entity, property, fact, event, or law ruling phenomena. Any element of any ontology is subsumed under the category of “items”. A “physical item”, in particular, is any constituent of a physical system, any physical property, any physical fact, any physical event, or any law of physics. What is for an item to be physical, and what is for an item to supervene on physical items, are crucial issues I will briefly address next and develop subsequently.19 1.3.1 Physical items Intuitively, we may consider the physical items to be the subject matter of physics. It is easy to find examples of physical items: A stone, the moon, and an electron are physical entities; electric charge and mass are physical properties; that the hydrogen atom has one electron and that a year takes approximately 365 days are physical facts; the decay of a nucleus and a solar eclipse are physical events; that energy is conserved and that the greatest speed is the speed of light are laws of physics. By contrast, entities like persons and cheques; properties like being modest or being expensive; the fact that a couple is married; the fact that 1 Euro is more expensive than 1 Dollar; an event like a marriage; an event like the occurrence of an economic crisis; a law of the market; are not physical items. However, this intuitive picture of what it is to be physical falls short of being satisfactory mainly for two reasons, which put forward “Hempel’s dilemma”.20 First, it may be somewhat circular: I said that physical entities are the object of study of physics, but certainly physics is often defined as the science that studies physical entities. Secondly, since physics is an empirical enterprise its ontology is open to 19 20

See Chapter 4. See Hempel (1969).

21

substantial revisions. Physical entities are the object of study of physics, but they cannot be defined as the entities that have the properties and obey the relations acknowledged by contemporary physics. Moreover, current physics poses puzzling issues concerning the metaphysics of matter, for example, with the wave-particle duality of quantum mechanics. I shall not attempt to provide a positive characterisation of what it is for an entity or property to be physical. This task would demand a long digression, with little chances of success for “after all, the only concept Plato succeeded in defining was mud (dirt and water).” (Davidson 2001, p. 156). I hope our intuitive conception of “the physical” will do for present purposes. Now, the strategy to support physicalism will be to show that nonphysicalist alternatives are incoherent or unappealing. Thereby, physicalism should be adopted for being the most compelling position. 1.3.2 Supervenience Take three nonparallel lines in Euclidean space, and consider the three segments (each one belonging to one line) that form a triangle. Note, firstly, that the “existence” of the triangle and its geometrical properties (like having angles which add up to 180º) is logically entailed by the “existence” of the three lines and their spatial relations. And note, secondly, that any change in the geometrical properties of the triangle requires a modification of its constitutive segments and is logically entailed by it. Accordingly, I would say that the triangle and its properties “supervene” on the three lines in Euclidean space.21 There are several definitions of “supervenience”. The one I will advance classifies, in J. Kim’s (1993) terms, as one of “global supervenience”. It is similar to Chalmers’ definition, which says: “Bproperties supervene on A-properties if no two possible situations are identical with respect to their A-properties while differing in their Bproperties.” (1996, p. 33).22 It is as follows: 21 22

In Chapter 4 I will argue that logical necessity is a kind of metaphysical necessity. I do not adopt Chalmers' definition because it crucially relies on the (problematic) notion of “possible worlds”. See Chapter 5.

22

(Sup) An item U supervenes on a set of items Pi if U is entailed by {Pi} with metaphysical necessity, and U ∉ {Pi}. The set of items Pi constitutes the “supervenience base”, and recall that “an item” stands for any kind of entity, property, fact, event, or law that rules the behaviour of entities. As an example of supervenience on physical items, consider some molecules properly related (in a dynamic way) and placed in an appropriate environment, so as to constitute a living cell. Note that the existence of a cell and the property of the cell being alive, are metaphysically entailed by the existence of these organised molecules and by their relations with the environment. And note that any event involving the cell, for instance, its death, is metaphysically entailed by some physical items; for example, by the event of the disintegration of the membrane. For a cell to exist, to live, to die, and so on, nothing more is required apart from some physical items existing or obtaining. Therefore, the cell, its properties, the facts about it, and the events involving it, supervene on physical items. Certainly, the cell type cannot be identified with a physical type, and the dying event type cannot be identified with a physical event type. Different arrangements of molecules can constitute cells, and many alterations of these structures correspond to an event of dying. But no cell can exist without some molecules being arranged in an appropriate way, and no cell can die without some physical events occurring. Cells have a physical nature. Other types of items, like cheques, the fact that I have a cheque in my wallet, and the event of paying for a book, also supervene on physical items. Certainly, the relation between the social and the physical is very intricate. Firstly, nothing is a cheque in virtue of having any particular physical properties, and no action is a paying-action in virtue of involving any particular physical event. Secondly, the existence of cheques and of paying-actions is relative to the existence of social institutions, social facts and social events. However, there can be no cheques without some physical vehicles, and there can be no paying without some physical events occurring. In short, if physicalism is true all the items studied by the social sciences, somehow, supervene on

23

items studied by physics. If we “fix” all the physical items, we thereby “fix” all the social items (including the psychological ones) with metaphysical necessity. Importantly, physicalism—as a metaphysical view—is not committed to the possibility of an epistemological reduction of the social sciences in terms of the natural sciences, and of the natural sciences in terms of fundamental physics.23 Physicalism is committed to a metaphysical reductionism: Every psychological or social item has a physical nature. There is an important point to highlight about supervenience. Supervenient items obtain with metaphysical necessity, by contrast with nomological necessity. If the existence of an item U were entailed by a set of physical items in virtue of some laws of nature, the entailment relation would not be one of supervenience. For instance, one litre of water and its property of being liquid do not supervene on the existence and properties of the mereological sum of approximately 335.1025 H2O molecules. The properties of H2O molecules alone do not metaphysically entail the water and its liquidity. In fact, these H2O molecules can also form ice or vapour. There are laws governing the interaction between the H2O molecules, which determine the liquidity of water as a function of the physical conditions. Water and its liquidity supervene on a larger supervenience base than the set of H2O molecules. The supervenience base also includes the pertinent laws of nature and the boundary conditions.24 Intuitively, the idea of “supervenience” might be easy to grasp. However, it involves a concept of “metaphysical necessity” that is not easy to capture in simple terms. I will elaborate the notion of supervenience and the differences between nomological and metaphysical entailments in forthcoming chapters.25 1.4 The mind/body duality By “dualism” I will refer to the traditional metaphysical thesis that says that the entities populating the world—and constitutive of human 23 24 25

See the influential papers “Special sciences” by Fodor (1974) and “Mental events” by Davidson (1980). There is a sense in which laws of nature can be said to be metaphysically necessary (Kistler 2005). I will discuss this point in Chapter 5. See Chapters 4 and 5.

24

beings—can be classified into two exclusive metaphysical categories: The physical and the mental. R. Descartes (1641) called the former “res extensa” and the latter “res cogitans”. For him, roughly, the mental phenomena (thoughts and phenomenal experiences) resulted from the properties and activity of the res cogitans; the remaining phenomena (digestion, locomotion, etc.) were attributed to the res extensa. Now, Descartes acknowledged there are intimate relations between the mental and the physical, and thus faced the problem of explaining this. Despite his efforts, it was never clear how it is possible for these two kinds of substances to interact, given the radical differences between their metaphysical natures. I consider the main problem of dualism to be that its proponents were never able to flesh it out in a coherent way. Intuitively, no doubt minds do not seem to have a physical nature, particularly from the firstperson perspective.26 Feelings, thoughts, sensations, appear to have a mode of existence substantially different from the mode of existence of everything else. If pain is real, it does not seem to be “real” in the same sense that a stone is. But notwithstanding intuitions, dualists were never able to elaborate a compelling alternative to physicalism. Our dualist intuitions are recalcitrant, we experience some uneasiness with physicalism—at least many philosophers do, but I think there is no argument for dualism both persuasive and positive. Typically, dualism gets support from arguments against physicalism, and not from its own explanatory virtues. Let us explore what the concept of an “immaterial substance” could amount to. I will try to show that dualist alternatives are even more problematic than physicalism. Every physical entity can be individuated by its physical properties.27 Moreover, every physical entity can be classified as 26

27

The “first-person perspective” on a mind is the exclusive point of view a subject has regarding his own mind. It may require introspection abilities. By contrast, the “third-person perspective” on a mind is the one a subject has regarding somebody else's mind. Certainly, in fundamental physics there are systems where two numerically distinct particles (bosons) are claimed to be empirically indistinguishable. But notice that the boson types can be individuated in virtue of their properties, despite the possible indistinguishability of some boson tokens.

25

belonging to some physical kind, where physical kinds are individuated by sets of physical properties. For instance, it is constitutive of electrons to have a specific (rest) mass, a specific spin, and a specific electric charge. Now, can a physical entity instantiate some property of a nonphysical nature?28 For example, can an electron besides its mass, charge, and so on, have a property that is nonphysical? Furthermore, can there be substances without any property of a physical nature? Can there be purely immaterial substances? To answer positively to the first question is to embrace a form of “property dualism”, and to answer positively to the second one is to embrace a form of “substance dualism”. The former is a weaker nonphysicalism than the latter, but I consider the distance between physicalism and property dualism to be greater than the distance between property dualism and substance dualism. If one accepts the possibility of properties with a nonphysical nature, the possibility of entities with a nonphysical nature gains plausibility. I will discuss both forms of dualism together. Consider P to be a property for which it is undetermined whether it is material or immaterial. There are two options: Either P can be (when instantiated) causally related to physical properties, or it cannot.29 I will argue that the first case, i.e., P being able to be causally related to a physical property, gives not only a necessary but also a sufficient condition for P being physical. Accordingly, immaterial properties (if they exist) are properties that cannot be causally related to any physical property. Let us proceed considering both scenarios: (1) P is not causally related to any physical property; (2) P is causally related to a physical property. (1) If a property P is not causally related to any physical property, it cannot be studied by natural science. P is out of reach of empirical research, since it does not (causally) interact with any experimental device (including our perception senses).30 Thus, since P is not part of the subject matter of physics, P does not classify as a physical property. 28 29 30

Recall that the category of items with a nonphysical nature excludes entities and properties that supervene on physical ones. I will advance a definition of causality in Chapter 4. Even if one considers that mental processes are not physical processes, no doubt perception also requires physical processes.

26

Accordingly, for a property P to be able to causally interact with physical properties is a necessary condition for P being physical. (2) If a property P is causally related to a physical property, like mass or electric charge, P can (in principle) be studied by natural science. In fact, since physical properties can be studied property P can also be studied in virtue of its relations with them. Now, if physical entities and properties are—as I claimed—the subject matter of physics, any property related to physical properties is ipso facto a physical property. Consequently, for a property P to be related to a physical property is a sufficient condition for P being physical. Thereby, properties of a nonphysical nature do not causally interact with any physical property, and immaterial entities (entities with exclusively nonphysical properties) do not causally interact with any physical entity. And if there were immaterial properties instantiated in entities with additional physical properties, only the latter could be causally related to other physical properties. In short, immaterial entities or properties proposed by dualism cannot be, given their nonphysical nature, granted with any causal powers with respect to the physical. There is no contradiction in the supposition that there are properties with a nonphysical nature. There is no contradiction in the idea that there can be nonphysical substances, or in the idea that an entity can instantiate both properties with a physical nature and properties with a nonphysical nature. These are metaphysical possibilities. However, they are not very interesting or compelling. In what sense would nonphysical properties be “real” despite lacking causal powers over the physical?31 How can we prove their reality if no empirical knowledge can be obtained about them? In any case, immaterial properties are of no use to explain any phenomena involving physical entities or properties causally interacting; in particular, they can shed no light on an understanding of mind-body causal interactions. Traditionally, immaterial substances and properties were defined negatively, i.e., by subtracting from the definiendum paradigmatic physical properties. These attempts confront the following problem: If 31

For a defence of the idea of causal powers as a criterion for reality see Kistler (2002). I will discuss this topic in Chapter 4.

27

we subtract from a substance every physical property, it is not clear what would remain. Just to say there remain “properties with a nonphysical nature” or “mental properties” begs the question. We would like to have criteria to identify these properties. And if we subtract from a substance only some physical properties, the remaining substance would keep some causal powers and thus classify—as I argued—as a material substance.32 Contemporary physicalist philosophers are sometimes criticised for being dogmatic about monism. They are accused of holding scientism, i.e., of blindly endorsing the scientific view of reality (besides lacking imagination capacities). There could be, it is argued, other components of reality apart from the “physical stuff” (and what supervenes on it). No doubt, science should not be uncritically trusted, physicalism is something we need good reasons to believe in, and our idea of “the physical” might require important revisions. However, critics of physicalism have not being able to propose a coherent alternative. Even in sophisticated contemporary philosophical discussions, it can be noticed that the anti-physicalism arguments do not offer new alternatives to be considered. The influential “conceivability arguments”, by Kripke and Chalmers, and F. Jackson’s “knowledge argument”, certainly present an important challenge for physicalism.33 But they throw no light on the elaboration of a plausible alternative. If it is true that our contemporary scientific picture of reality leaves out important elements of it—and I think it does—we should better conceive a form of physicalism that enlarges its scope, instead of denying it. As Nagel says “[n]othing is proved by the inadequacy of physicalist hypotheses that assume a faulty objective analysis of mind. It would be truer to say physicalism is a position we cannot understand because we do not at the present have any conception of how it might be true.” (1974, p, 328. Bold in the original). In sum, if we believe there are causal relations between minds and bodies we have no other option than to accept that, despite appearances, minds have a physical nature. Now, I will discuss what reasons we have 32 33

Note that just to attribute a spatiotemporal location to a supposedly immaterial substance is to grant it with paradigmatic physical properties. These arguments will be discussed in Chapters 3, 5 and 6.

28

to acknowledge that indeed there are mind-body causal relations. 1.5 Minds and bodies It is a difficult question how to individuate mental states. We might endorse an atomistic picture, where the mind is seen as a collection of mental states that can be sharply distinguished and characterised, for example, in virtue of their contents. Or we might endorse a holistic picture, claiming that it only makes sense to describe our cognitive system as being, at every moment, in some single (though very complex) mental state. We might also prefer a dynamic view of the mind, where there are solely “mental processes” instead of the static “mental states”. Furthermore, what is the nature of mental states is a very controversial question. Proposals vary according to the general picture of the mind held. For dualists, just to talk of mental “states” might lack sense, since this is materialist jargon not appropriate to characterise immaterial substances. For behaviourists, mental states are no more than dispositional states, reducible in terms of stimulus-response schemes. For functionalists, mental states are functional states realised over some hardware (brain states in the case of humans).34 For connectionists, a mental state corresponds to some activation pattern of a neural network. For eliminativists, a mental state is nothing more than a brain state. Is there something in common between all these conceptions of “mental states”? Probably just the fact that mental states are whatever psychology is about. I do not want to adhere to some quite specific view of mental states or of the structure of the mind. I will not argue for functionalism, or connectionism, or massive modularity, etc. The reason is that I want to preserve throughout this discussion as much generality as possible. The fact that there are material entities such that there is something it is like to be one of them, does not force the endorsement (at least from the outset) of some particular picture of the mind. I will occasionally talk of conscious and unconscious “mental states”, and of “minds” as bundles of several mental states. It happens that this is the most common way of talking in the contemporary 34

I will discuss behaviourism and (computational) functionalism about consciousness in Chapter 2.

29

literature. However, I am not endorsing a static atomistic view of the mind. Indeed, the holistic and dynamic picture seems more appropriate to account for the phenomenon of consciousness. First, note that “the unity of consciousness” favours a holistic view where, so to speak, all the conscious mental states merge to give rise to some unity.35 Second, note that an essential characteristic of experience is its temporality; consciousness is a flowing “stream” (James 1890). Whether one adopts the atomistic or the holistic picture of the mind has implications for the relation between “phenomenal content” and “phenomenal character”. I will say that: (PC) If a subject S is taken to be at time t in a single mental state M, the phenomenal character of the experience of S at time t is identical to the phenomenal content of M. If a subject S is taken to be at time t in several mental states Mi, the phenomenal character of the experience of S at time t is the mereological sum of the phenomenal contents of the Mi into a conjoint experiential character.36 There is also the question of whether we should consider consciousness as constitutive of mental states. It is usually held that for a cognitive state to be mental it has to be intentional,37 but some philosophers go further and suggest that consciousness is required for intentionality.38 As I said, I will talk of “conscious” and “unconscious” mental states. However, this is not to be taken as a commitment to the idea that consciousness is not essential for some state to be mental. In fact, this question largely depends on the general picture of the mind one holds. Now, there are plenty of reasons to believe there are relations between physical states and psychological states. The question of the 35

See Bayne (2010). I will discuss the “unity of consciousness” in Chapter 3. The concept of “conjoint experiential character” is introduced by Bayne (2010) to refer to the phenomenal unity of consciousness. See Chapter 3. 37 Brentano (1874) established intentionality as the mark of the mental. Roughly, for a state to be intentional is for it to be about something else. However, it has been argued that states like generalised anxiety, depression or euphoria, are mental but not intentional. 38 For instance Searle (2002) argues in this direction. 36

30

nature of these relations is the general “mind-body problem”.39 There are three possibilities worth considering: (1) There are both mind-to-body and body-to-mind causal relations; (2) there are only body-to-mind causal relations, not mind-to-body ones; (3) despite appearances, there are no relations between minds and bodies. Let us briefly address them. 1.5.1 Mind-body causal relations Psychology, including folk psychology, is based on the following ideas: (1) Our mental states are a function of the stimuli we receive, especially through perception mechanisms (besides also being a function of other cognitive and bodily states); and (2) our (intentional) behaviour is a function of our mental states, for we act as we do because we have beliefs and desires with a certain content. The best justification for 1 and 2 is taken to be that, somehow, the physical has causal powers over the mental and, somehow, the mental causally affects behaviour. It is controversial whether psychology is a natural science; it uses some intentional categories (beliefs, desires, etc.), by contrast with the extensional and nonmentalistic terminology of the hard sciences.40 However, it can hardly be denied that psychology has some predictive and explanatory success. Thus, if we accept 1 and 2, and the best explanation for their validity is that there are mind-body causal relations, psychology would be giving evidential support to the idea that there is indeed this kind of relations. It does not seem very controversial that the external physical world causally influences minds. Perception processes relate the physical entities that we see, hear, touch, and so on, with mental phenomena and contents, in a systematic way. We have experiences while we perceive,41 we remember what we saw, and we form beliefs about events we 39 40

41

For a discussion about the relations between the mind-body problem and the more specific problem of consciousness see Kim (2005). A vocabulary has an extensional taxonomy if coreferential terms can be exchanged salva veritate. Otherwise, its taxonomy is intensional. Now, it is commonly agreed that a vocabulary that includes intentional terms has an intensional taxonomy. But there is strong empirical evidence for the claim that there are also phenomenally unconscious (or “sub-personal”) perceptions. See Baars et al. (eds.) (2003).

31

witness. And if we drink alcohol, take a psychedelic drug, swallow an aspirin when suffering a headache, or receive a strike that injures our central nervous system, the psychological effects are quite notorious and systematic. Clearly, external objects, chemicals, etc. seem to causally produce effects on minds. Conversely, it also seems that minds causally influence the body and the surrounding material world. Even though it is very controversial whether mental states really have causal powers over the body, prima facie this seems to be the case.42 If I want to raise my arm and I perform the corresponding mental act, some physiological processes—which are physical events—are initiated and my arm rises. My desire to raise my arm seems to be causally related with its rising. If I feel hungry, I might go to the fridge, take some food, prepare a dish and eat it. My chain of activity seems to be causally related to my appetite. If I believe I should have a higher education degree, I may register in some university and comply with all the requirements to get it. The accomplishment of my project seems to be causally related to my belief. Moreover, we interpret the (intentional) behaviour of other creatures by attributing mental states to them. We take the agent’s behaviour as causally resulting from the beliefs and desires we attribute to him. It is certainly difficult to understand how intentional states could be causally related to the production or guidance of physical phenomena. Beliefs and desires belong to “the realm of reasons”, and it is not easy to reduce or relate them to something belonging to “the realm of causality”. However, that there are causal relations between intentional states and behaviour certainly seems intuitively true; for instance, J. Fodor says: “If it isn’t literally true that my wanting is causally responsible for my reaching, and my itching is causally responsible for my scratching, and my believing is causally responsible for saying […], if none of that is literally true, then practically everything I believe about anything is false and it’s the end of the world.” (1990, p. 156). 1.5.2 Epiphenomenalism It can be argued that the empirical evidence does not show that there are 42

For a discussion about whether mental states have causal powers, and a defence of the thesis that they do, see Kistler (2006b; 2007).

32

mind-body causal relations, but just that there are correlations between mind and body phenomena. What is observed is that whenever there is a psychological phenomenon there is simultaneously some brain activity. Moreover, some brain patterns are characteristically correlated to some mental phenomena. For instance, by observing brain activity it is possible to determine if someone is awake or dreaming while sleeping. Indeed, some scientists like Crick (1994) consider that the scientific study of consciousness consists in determining the “neural correlate” of conscious phenomena; to each type of conscious experience they want to associate some characteristic brain activity. The question is how to explain these correlations. One possibility, as we just saw, is to claim that there are causal relations in both directions: The physical has causal powers over the mental, and the mental has causal powers over the physical.43 But there are other possibilities: (1) The correlations are due to causal influences that only go in one direction, in particular, from the physical to the mental but never from the mental to the physical. (2) The correlations derive from the fact that brain activity and psychological activity have a common cause. (3) The correlations do not result from the existence of any causal relations between the mental and the physical, but from something else that is not causal. Option 3 is not very appealing. If there is no causal connexion at all (not even indirectly) between physical and mental phenomena, it is hard to explain why there is a correlation. It seems necessary to appeal to some pre-established harmony in nature—as Spinoza did—that provides noncausal laws not involving or resulting from any causal interaction. This is a metaphysical possibility but, in the contemporary context, an extravagant one. Option 2, i.e., that mental and physical phenomena have a common cause, does not provide a real alternative. What is the nature of the common cause? If the common cause is a physical event, then there are physical phenomena causing mental phenomena. Analogously, if the common cause is a mental event, then there are mental phenomena 43

If the mental is taken to be irreducible to the physical, this possibility faces the problem of “overdetermination”. See Kim (2005).

33

causing physical phenomena. And if the common cause is neither a physical nor a mental event, then a third fundamental metaphysical category is required. If dualism already poses important problems, trialism surely does not fare much better and faces analogous difficulties. In fact, the acknowledgement of correlations between physical and mental events either involves a commitment to the existence of some causal relations somehow involving the physical and the mental, or requires taking these correlations as completely mysterious. Option 1, which corresponds to the so-called “epiphenomenalism”, acknowledges the existence of causal interactions between bodies and minds, but only in one direction: Mental phenomena are caused by physical phenomena, but the mental does not have any causal powers.44 It is a quite influential position; I will continue by briefly discussing it. Epiphenomenalists explain mind-body correlations by claiming that physical phenomena cause mental phenomena, but deny the existence of causal powers of the mental over the physical. In other words, epiphenomenalism claims that physical events can have two types of causal effects: On the one hand they cause—as usual—other physical events, but on the other hand they may also cause, as a kind of by-product, mental phenomena. Epiphenomenalism is an attempt to reconcile a metaphysical thesis with a metaphysical principle: The thesis is that mental (and in particular conscious) phenomena exist but are not eliminatively reducible in terms of physical phenomena. The principle is the “closure” of the physical realm: Every physical event is caused by another physical event.45 Epiphenomenalism is not exempt from serious problems. If it embraces dualism, it has to explain how a physical event can cause an event of a completely different nature. And if it embraces (as usually) physicalism, then it is not clear why mental states (being physical or supervenient on the physical) do not have any causal powers. Moreover, even though there is no problem with the idea of a physical event having 44 45

Epiphenomenalism specifically about consciousness denies that the phenomenal properties of conscious mental states have causal powers. See Kim (2005). See Kim (2005).

34

two effects,46 it is not clear why a causal chain would stop at a given mental event, especially if a mental event is such that it can enter into a causal chain. Causal chains do not have an end.47 Concerning in particular conscious phenomena, there are tricky experimental results in cognitive science that throw doubt on conscious mental acts causing bodily acts. More precisely, there are doubts about the conscious mental act of deciding to perform a bodily act being the cause of the performance of the act. B. Libet (2004) argues, on the base of considerable experimental work, that the psychomotor processes corresponding to the performance of the act start before the subject is consciously aware of his decision to act.48 It is known that bodily acts initiate with electrochemical activity in the motor cortex. An “action potential” is produced, and transmitted through the nerves to the corresponding muscles. As a result of the stimulus that the action potential provides to the muscles, these contract. In his experiments, Libet asked subjects to report the exact moment when they decide to perform an act, measured the exact moment when the action potential was generated in the motor cortex, and compared these data. He concluded that the action potentials were produced before the subjects were consciously aware of their decision to perform the act. The results of Libet’s experiments are intriguing, but must be carefully interpreted. They do not prove the general claim that mental states do not have causal powers. First, they concern conscious mental acts, and not mental acts in general. They do not show that unconscious mental acts (if there are such acts) did not cause the bodily act. Second, they concern simple and short acts (like the pressing of a button). They do not concern behaviour that is articulated in a planning and an execution extended over time, like feeling hungry and getting some food, or wanting a university degree and registering in some institution. Third, and more important, even if a subject happens to be aware of his decision to perform an act when—it is argued—the corresponding 46 47 48

In general, a single event can have many causes and produce many effects. I would say that a causal chain would have an end only if it were the end of the whole universe. In a causal relation, the cause must temporarily precede the effect (and this must hold in every frame of reference).

35

physiological process already began, it is still possible that his decision is causally responsible for his act. In fact, in Libet’s experiments (for practical reasons) it is not really the moment when the conscious mental act occurs that is taken into account. What is registered is the moment when the subject is aware of starting the performance of the mental act. Now, it’s not clear why we should accept that the subject’s judgement of the timing of the conscious decision, as expressed through Libet’s methodology, should correspond to her actual awareness of that conscious event. In addition, there is a difference between having a conscious experience and becoming aware of having the conscious experience, and it might take some time between the occurrences of the two phenomena. To be able to report conscious experiences, it seems necessary to “monitor” them and produce a higher-order mental state. Then, it might be the case that the lower-order mental state is a conscious state which causes the start of the psychomotor process, and that the higher-order mental state, which self-attributes the lower-order one to the subject and enables the reporting, is taking some additional time to be produced. This additional time might explain the surprising results.49 1.5.3 Monism It can be maintained that there are no relations at all between the mental and the physical, not even correlations. If one embraces this view in the frame of dualism, one has such an extravagant position that it can hardly be conceived. Reality would be split in two completely separate realms; minds would belong to one realm, and bodies to the other. For one thing, it would not be possible to explain how each subject associates one body (the one he takes to be his body) with his mind. In a monist frame, one option is to deny the existence of the physical. For instance, idealists considered every substance to have a mental nature; the physical was considered to be a kind of mental entity. This position faced many difficulties. For example, G. Berkeley (1710) had to appeal to God’s mind to grant the existence of an objective physical world we could intersubjectively share. In any case, notice that 49

For an interesting critique of the interpretation of Libet's experiments—in particular of the idea that they show there is no free will—see Mele (2009).

36

idealism does not rule out mind-body causation; the mind-body relations are taken to be a species of mind-to-mind relations. The other option is to deny the existence of the mental. For example, some behaviourist psychologists of the 20th century claimed that the mental is eliminatively reducible in terms of behaviour, which in turn can be exhaustively described in nonintentional terms. For them, mental states were no more than dispositions to behave some way. It is controversial to what extent dispositional properties do have causal powers.50 But in any case, the stimulus-reaction schemes proposed by behaviourism to replace mental concepts were considered to be realised by causal processes. Indeed, behaviourism gained empirical support mainly when it was showed that existing schemes are modifiable (“classical” conditioning), and that new schemes can be produced (“operant” conditioning), as an effect of conditioning. Therefore, eliminativism about the mental does not really deny that there are causal relations between physical and mental phenomena. These are taken to be physical-to-physical relations. In sum, if one wants to deny that there are causal relations between the mental and the physical, the only plausible alternative is to claim that one category can be eliminatively reduced in terms of the other. When it is the category of the physical that is preserved, the possibility of causal relations between physical events and (reduced) mental events is obviously preserved as well. 1.6 Panpsychism Dualism must explain how it is possible for the mind to interact with the body, if the mental and the physical are irreconcilably different. Monism has the advantage that it can explain the existence of mind-body interactions in a straightforward manner, but the price is to take the mental to be physical, or the physical to be mental. Panpsychism is proposed as a third alternative to account for physical processes involving mental processes, without reducing the mental to the physical or vice-versa. The idea underlying panpsychism is that “the basic physical constituents of the universe have mental (and in particular experiential) 50

See the introduction in Gnassounou & Kistler (eds.) (2007).

37

properties.” (Nagel 1979, p. 81). For panpsychists, consciousness is an essential property of every physical entity and, in particular, of the fundamental ones. This is the idea I will discuss. Note that panpsychism resembles dualism of properties, for besides ordinary physical properties there are the mental ones. However, dualism of properties, unlike panpsychism, is not committed to the idea that every entity, and in particular fundamental entities, has mental properties. No doubt panpsychism is a counterintuitive view and, e.g., Searle considers it to be just “absurd” (1997, p. 156). Nevertheless, others like Chalmers claim that “we ought to take the possibility of some sort of panpsychism seriously: there seem to be no knockout arguments against the view.” (1996, p. 299). Moreover, G. Strawson argued for a form of panpsychism he first called “realistic monism” and latter “FundamentalDuality Monism”, which says that “[t]here is only one fundamental kind of stuff. It (all of it) has both experiential reality and nonexperiential reality. There is no other kind of reality.” (2006, p.236). I will examine the motivations for panpsychism, its commitments, and the problems it faces. Suppose (i) that “ultimates”,51 i.e., simple entities, do not have the consciousness property, and suppose (ii) that every property of a complex entity supervenes on the properties of ultimates. According to panpsychists, from i and ii it follows that no entity could have the consciousness property. Therefore, from realism about consciousness panpsychists conclude that i must be false. Ultimates must have the property that there is something it is like to be one of them. It is important to note that ii is not equivalent to the claim that physicalism is true. I defined physicalism so that every entity is a physical entity or supervenes on physical items, where “physical entities” encompass both simple and complex ones. The underlying intuition in the previous argument for panpsychism is that consciousness cannot result from physical (as opposed to experiential) properties, given that “physical stuff is, in itself, in its fundamental nature, something wholly and utterly nonexperiential” 51

This concept of “ultimates” is introduced by Strawson (2006) to name entities that are simple and thus indivisible. It has the advantage that it remains neutral concerning their metaphysical nature.

38

(Strawson 2006, p. 11) and “[...] you can’t get experiential phenomena from P [physical] phenomena.” (Strawson 2006, p. 24). For panpsychists there is an insurmountable metaphysical gap between the property of being conscious and any physical property, of a simple or complex entity. The fact of nonultimate entities being conscious can only be explained, they claim, by them being composed of conscious ultimates. We may articulate the argument for panpsychism as follows: (i) All the properties that ultimates instantiate have a physical metaphysical nature. [supposition] (ii) All the properties of complex entities supervene (are entailed with metaphysical necessity) on the properties of ultimates. [premise] (iii) The supervenience (metaphysical entailment) relation between properties is such that the metaphysical nature of the properties in the consequent is the same as in the antecedent. [premise] (iv) All the properties of complex entities have a physical metaphysical nature. [from i to iii] (v) There are complex entities which instantiate the consciousness property. [premise] (vi) The consciousness property is of a different metaphysical nature than the physical properties. [premise] (vii) There are complex entities which instantiate a property that does not have a physical nature. [from v and vi] Clearly, there is a contradiction between iv and vii; the supposition i must be false. Accordingly: (c1) Ultimates instantiate properties that are not of a physical metaphysical nature. But then, when replacing i by c1, it follows that: (c2) Ultimates (or at least some of them) instantiate the consciousness property. [from v, vi, iii and ii]

39

Premise ii plays the pillar role in this argument, and indeed in other influential ones like Chalmers’ “zombie argument”.52 It corresponds to a commonly accepted metaphysical picture of reality, that I label “microphysicalism”. Note that panpsychism would be unmotivated if microphysicalism were rejected. If the properties of complex systems are not required to supervene on the properties of the composing ultimates, the possibility is open for the consciousness property to be instantiated in the former but not in the latter. Premise iii does not seem very controversial. In metaphysical entailment relations the consequent is supposed to be “contained” in the antecedent. In forthcoming chapters I will elaborate the notion of “metaphysical necessity”.53 Premise v just states realism about consciousness, and premise vi is the main idea shared by panpsychists and dualists. They lie at the core of the mind-body problem, and capture the intuition that raises the problem of consciousness. The general purpose of the present work is to make the denial of vi plausible. But, for the sake of the argument, let us accept vi and discuss further problems of panpsychism. I will focus on the plausibility of conclusion c2, which says that there is something it is like to be an ultimate. Firstly, I will discuss the “combination problem” (Strawson 2006, p. 248). Secondly, I will confront panpsychism with some scientific theories and data about “the neural correlate” of consciousness. 1.6.1 The combination problem According to panpsychism, each individual ultimate instantiates the consciousness property, i.e., there is something it is like to be an ultimate.54 Now, take a body B composed of n ultimates. There are two interesting possibilities concerning the instantiation of the consciousness property in B: (1) B instantiates the consciousness property n times, one time in each constitutive ultimate; (2) B instantiates the consciousness property only one time, as a whole. Now, recall that the starting point for realism about consciousness is that there is something it is like to be me: 52 53 54

I will discuss this argument in Chapter 5. See Chapters 4 and 5. But note that, strictly speaking, the previous argument for panpsychism would show that at least one of the ultimates constituting a complex conscious entity must instantiate the consciousness property.

40

I instantiate the consciousness property. According to possibility 1, the “I” which instantiates the consciousness property is an ultimate; according to 2, it is a complex system. I will argue that 1 is unreasonable, and that 2 is not open for the panpsychist. Other alternatives for how many times B instantiates the consciousness property can be constructed from combinations of 1 and 2. Therefore, they are also rejected via the rejection of 1 and 2. Suppose that the “I” which instantiates the consciousness property (there is something it is like to be me) is a particular ultimate amongst a plurality of (conscious) ultimates composing my body. Which ultimate is it? A human being, a nervous system, a brain, a cortico-thalamic network, and indeed any biological entity is a highly complex system composed of huge amounts of ultimates. It then seems arbitrary, to say the least, to attribute the consciousness property that “I” instantiate to one particular ultimate. Indeed, if we take out this single particle from my biological organism, would my consciousness be disembodied from the organism? Being possibility 1 unreasonable, no doubt that panpsychists propose something corresponding rather to option 2. They primarily want to explain how it is possible for complex systems like human beings to be conscious, and they believe that this can only be done if we take ultimates to have experiential properties. According to option 2, the “I” which instantiates the consciousness property is a complex system that could be, for instance, my brain or some part of it. Somehow, some ultimates merge their individual tokens of the consciousness property in such a way as to give rise to the instantiation of a single token of this property in the system they compose. This is an interesting possibility. However, option 2 is not really open for the panpsychist because it is incompatible with premise ii (microphysicalism). On the one hand, to comply with option 2 panpsychism needs to appeal to some process by virtue of which the ultimates merge their consciousness properties. Somehow, the ultimates must interact in a lawful manner. There must be laws that define which ultimates merge

41

their consciousness properties and under what conditions they do so.55 But on the other hand, premise ii states that the properties of a complex entity are metaphysically entailed by its constitutive ultimates, and this certainly excludes the above-mentioned process. Some relation between ultimates and a complex system, which instantiates the consciousness property in the latter by merging the consciousness properties of the former, does not classify as a “metaphysical entailment”.56 In other words, to embrace option 2 the panpsychist has to accept the existence of mechanisms that produce a high-level property, i.e., a property which (i) is instantiated in complex systems and (ii) is not just the aggregate of properties of the composing ultimates. But then, the case for panpsychism is seriously weakened. Not only is microphysicalism—which plays the key role in the argument for panpsychism—sacrificed. In addition, observe that the idea that there are laws that can operate on the experiential properties of ultimates to create a new property—a unified consciousness property, is not that far from the idea that there are laws of nature that can operate on physical (nonexperiential) properties to create a new property—the consciousness property.57 In any case, as long as panpsychism does not propose a compelling solution for the composition problem it falls short of providing a good alternative for the problem of consciousness.

55

56 57

The laws must determine the frontiers between a composed system that instantiates one token of the consciousness property and other (simple or complex) entities that instantiate other tokens. I will develop this point in Chapter 4. Panpsychists usually consider that the only alternative to microphysicalism is a sort or “brute emergence”. Since brute emergence is, roughly, the idea that some complex entities instantiate causally efficacious high-level properties that are not related to the properties of its microconstituents, panpsychists (rightly) reject this view as implausible. But the composition problem seems to force the panpsychist to accept something similar to brute emergence. As Goff says: “Unfortunately, panpsychism is also committed to a kind of brute emergence which is arguably just as unintelligible as the emergence of the experiential from the nonexperiential: the emergence of novel ‘macroexperiential phenomena’ from ‘microexperiential phenomena’.” (2006, p. 53).

42

1.6.2 Panpsychism and neuroscience Panpsychism seems to be a metaphysical possibility; there is no definitive metaphysical consideration that prevents ultimates from being conscious. However, the available empirical data strongly suggest that consciousness is a property of highly complex and, moreover, dynamic systems. First, it is remarkable that neuroscientists, despite the differences in the theories of consciousness they propose, agree on the idea that consciousness (in humans) is globally58 instantiated in highly complex neural networks undergoing electrochemical activity. For instance: - F. Crick says: “Consciousness depends crucially on thalamic connections with the cortex. It exists only if certain cortical areas have reverberatory circuits (involving cortical layers 4 and 6) that project strongly enough to produce significant reverberations.” (1994, p. 252). - G. Edelman & G. Tononi say: “[The dynamic core hypothesis] states that the activity of a group of neurons can contribute directly to conscious experience if it is part of a functional cluster, characterized by strong mutual interactions among a set of neuronal groups over a period of milliseconds. To sustain conscious experience, it is essential that this functional cluster be highly differentiated, as indicated by high values of complexity. Such a cluster, which we call “the dynamic core” because of its ever-changing composition yet ongoing integration, is generated largely, although not exclusively, within the thalamocortical system.” (2000, p.139). - Libet says: “[...] we may view conscious subjective experience as if it were a field, produced by appropriate though multifarious neuronal activities of the brain […]. Such a field would provide communication within the cerebral cortex without the neural connections and pathways in the cortex. 58

The consciousness property is “globally instantiated”, roughly, if there is not a single precise closed region of the brain (or of the body) where it is instantiated.

43

A conscious mental field (CMF) would provide the mediator between the physical activities of nerve cells and the emergence of subjective experience […]. A chief quality or attribute of CMF would be that of a unified or unitary subjective experience.” (2004, p. 168. Italics in the original). In fact, the complexity of the nervous system is astonishing. The number of neurons is of the order of 100 billion, each one with an average of 1000 synaptic connexions. Whenever a subject has, for instance, a visual experience, billions of neurons show some activity. This suggests that in order for consciousness to be instantiated a very complex system undergoing electrochemical processes must take place. This is clearly at odds with panpsychism. To be sure, the complexity of the brain and its activity does not prove panpsychism to be wrong; but it is usually the case that the complexity of a system that performs a given function is required for it to have this ability. Secondly, it is a remarkable characteristic of conscious experience that, as Searle points out, it seems to switch on and off: “[C]onsciousness” refers to those states of sentience and awareness that typically begin when we awake from a dreamless sleep and continue until we go to sleep again, or fall into a coma or die or otherwise become “unconscious.” (1997, p. 5) It seems it is not the case that all the time there is something it is like to be a human being. It seems that consciousness disappears while the organism is sleeping without dreaming, under the effect of some substances, or damaged in certain ways. Now, if panpsychism were true we would expect there to be always something it is like to be a human organism. In fact, a human organism sleeping (without dreaming), or with brain damage, never ceased to be composed by ultimates with putative phenomenal properties. The panpsychist can reply in two ways. One, is to claim that there is always something it is like to be a person, but one does not remember that this was the case while one was sleeping or in coma. The other way,

44

is to argue that sometimes the (putative) process which merges the consciousness property of ultimates to produce the (unified) consciousness property of the person is reversed. However, both replies seem ad hoc; the best explanation for the appearance of consciousness switching on and off is that indeed it switches on and off. If one claims the contrary, one has the burden of proof. In sum, even though the empirical data and the available scientific theories about consciousness do not rule out panpsychism, they are in tension with it. We have good reasons to believe that for consciousness to be instantiated very complex physical systems undergoing complex dynamic processes are required. Finally, note that our conscious experience is something very rich that flows and changes over time. There are countless possibilities for the what-it-is-like-ness of being someone at any moment. But an ultimate, given that it lacks internal structure, seems only able to occupy a single, simple, and static experiential state. In fact, a composed system can be in different states since the relations between its constituents can vary. But a fundamental particle, an electron for instance, just has a given mass, a given charge, and a spin, which do not take different possible magnitudes. Therefore, its “experience” would be so different from ours that it is really difficult to conceive in what sense there would be something it is like to be an ultimate. 1.7 Conclusion Phenomenal consciousness is as a property some entities have, and an entity is conscious if and only if there is something it is like to be that entity. From this definition no empirical test can be provided to establish if and entity S is conscious or not. S can believe that she is conscious because indeed she is, but for another subject W to attribute consciousness to S a theory is required. This theory must provide intersubjective criteria, based on the observation of behaviour, physical properties, or something of the sort. In short: (C2) In order for a subject S to determine whether another entity W is conscious, she needs to appeal to an empirical theory proposing third-person point of view criteria for the attribution of

45

consciousness. It is not possible to prove that consciousness is a real phenomenon, but neither can eliminativists prove that it is not. For the realist, consciousness is given as a brute fact, whereas for the latter the existence of consciousness is a theoretical claim. I adhere to a realist view of consciousness, supported by the fact that it certainly seems to me that I have phenomenal experiences, and by the idea that the reality/appearance dichotomy does not apply to phenomenal characters. There clearly are correlations between physical and mental phenomena, which give good reasons to believe that there are causal relations between the physical world (including the body) and (conscious) minds. Now, dualism of substances, dualism of properties, epiphenomenalism, and panpsychism, fall short of providing a compelling explanation for these correlations. Some form of physicalism appears to be the most reasonable alternative, despite the fact that minds do not seem, from the subjective point of view, to have a physical nature. In all probability, the recalcitrant intuition that natural phenomena cannot account for phenomenal experiences is partly due, as D. Papineau argues, to the fact that dualism is almost “hard wired” in our conceptual schemes. He says about the “explanatory gap”,59 which is claimed to hold between any phenomenal experience and any natural phenomenon that could be identified with the experience, that: “The feeling of an ‘explanatory gap’ arises only because we cannot stop ourselves thinking about the mind-brain relation in a dualist way.” (2011, p. 5).

59

I will discuss the explanatory gap in Chapter 3.

47

CHAPTER 2: STRONG AI AND COMPUTATIONALISM 2.0 Introduction In the previous chapter I defined phenomenal consciousness as follows: (Def) An entity (or “creature”) is conscious if and only if she is in some conscious mental state. A mental state is conscious if and only if there is something it is like to be in that mental state for the corresponding entity. I argued for a realist stance towards consciousness, and for some form of physicalism: (1) There is a property some entities, in particular human beings when they are awake and in normal conditions, have: They are (phenomenally) conscious, i.e., there is something it is like to be one of them. (2) Physicalism is the most appealing metaphysics given the existence of mind-body relations. According to physicalism, all the entities inhabiting the actual world, their properties, and all the facts and events involving them, have a physical nature. Now, something is said to have “a physical nature” if it is a physical item or it supervenes on a set of physical items.60 Statements 1 and 2 entail the following claim: (C1) There is a property X, which is a physical property or a supervenient property,61 such that for an entity S to be conscious is for S to instantiate X. 60 61

Recall that “an item” is any kind of entity, property, fact, event, or law ruling phenomena. Hereafter, when saying of an item that it is “supervenient” without specifying the nature of the supervenience base, I will mean supervenient on physical items.

48

I also argued that in order for a subject S to determine whether another entity W is conscious, she needs to appeal to an empirical theory proposing third-person point of view criteria for the attribution of consciousness. This theory connects the metaphysical claim of the existence of consciousness with the epistemological question of giving sufficient and necessary conditions for its attribution. In this chapter I will consider two influential families of approaches to the question of the nature of consciousness. They do not take it to be a physical property, but a supervenient one. Namely, I will discuss Strong Artificial Intelligence (Strong AI) and computational functionalism. Recall that I consider consciousness to be a natural phenomenon, and (some form of) physicalism to be true. Hence, I require for each view to account for consciousness inside this metaphysical framework. If one is a dualist or an eliminativist, one has different requirements for a theory of consciousness and, consequently, one would accept or reject the arguments I will advance on different grounds. 2.1 Strong AI The behaviourist psychological theory lost its popularity fifty years ago. However, a behaviourist approach to consciousness, labelled by J. Searle “Strong AI”, is still quite influential. Roughly, the idea is to attribute consciousness exclusively on the base of behavioural criteria. As a view of the mind in general, and not of consciousness in particular, Strong AI can be characterised by two theses: (1) Artificial minds (not given by artificial replicas of biological systems) are physically possible. (2) To have a mind is to exhibit intentional behaviour. Point 1 states the belief that, in order to have a mind, a system does not need to have a biological brain, made out of neurons and so on. It opens the possibility for a system made out of, for example, silicon chips, to have a mind. The validity of 1 depends, obviously, in the underlying

49

conception of “a mind”. This conception is given in point 2: To have a mind is to behave some way. This is a metaphysical claim, with an epistemological implication: (3) To exhibit intentional behaviour is necessary and sufficient for the attribution of a mind. No doubt, it is reasonable to consider that the criteria to attribute to someone a mental quality, e.g., cleverness, are behavioural. If a person excels at solving problems, ratiocinating, and so on, she meets the conditions to be considered clever. Similarly, it is reasonable to hold that the criteria to attribute, in general, intentional states, are behavioural. Mental states, even if they were internal states, are not (under normal conditions) private states. Their content is manifest in the behaviour of the corresponding entities. Indeed, not only for humans but also for animals it seems reasonable to appeal to behavioural criteria for the attribution of mental states. For instance, D. Dennett (1987) argues that the attribution of different orders of intentionality to an animal should be a function of the complexity and flexibility of its behaviour.62 Following this line of reasoning, Strong AI claims that we should likewise use behavioural criteria for the attribution of intentionality to a machine. If a machine shows belief-like behaviour, desire-like behaviour, and so on, she meets the requirements for the attribution of the corresponding intentional states and, more generally, for the attribution of a mind. In fact, for Strong AI to exhibit intentional behaviour is to have a mind. The pragmatic approach of Strong AI is certainly attractive. However, there are two problematic questions worth mentioning. First, note that it is one thing to attribute a particular intentional state to some entity on the basis that this creature has a mind, and another thing to 62

Dennett (1987) claims that we should always attribute to an animal the lowest level of intentionality that permits to successfully interpret its behaviour. However, Dennett is an instrumentalist about intentionality: strictly speaking, mental states do not have any property corresponding to “intentionality”. We adopt an “intensional stance” to describe and predict the behaviour of persons and (some) animals. Attributions of intentionality have a pragmatic value.

50

attribute a mind. It might be true that we actually use behavioural criteria to attribute mental states, but it is not clear whether the attribution of a mind as such also results from behavioural considerations. Second, the criteria to be used for the attribution of intentional states to a machine, or may be to an alien, must be specified and justified. Now, it is not trivial to determine what counts, in general, as “intentional behaviour”. We only know of human and animal behaviour, and these are similar given that we share our genetic history to an important extent. In fact, it is an open question whether human beings confronted with alien creatures equipped with “minds” would be able to make sense of their behaviour.63 In any case, the essential point for the present purposes is that, according to Strong AI, the attribution of a mind to an entity S should not result from the instantiation of some physical property in S, but from S behaving some way. Now, if we attribute a mind on the basis of behaviour, it seems reasonable to attribute consciousness on the basis of behaviour as well; after all, to be conscious is to have a conscious mind. This is what proponents of Strong AI about consciousness claim. The idea can be stated as follows: (SAI) If an entity S exhibits conscious-like behaviour, S instantiates the consciousness property. Certainly, this view has some advantages. First, we desire third-person criteria for the attribution of consciousness and behaviour is intersubjectively available. Second, we do not have scientific reasons to believe that only human organisms and their close cousins can be conscious; Strong AI (as well as Weak AI64) leaves open the possibility for artificial systems to be so. Third, the use of behavioural criteria seems to be in agreement with common sense and folk psychology. However, there are two difficult questions Strong AI must face. First, which are the behavioural patterns that (are claimed to) justify the 63 64

See, e.g., Wittgenstein (1953). Weak AI, like Strong AI, considers that artificial systems can instantiate the consciousness property. However, Weak AI denies that human-like behaviour is a sufficient condition for the attribution of consciousness. See Searle (1997).

51

attribution of consciousness? Secondly, can behavioural characteristics indeed justify the attribution of consciousness? The first question mainly concerns the epistemological applicability of SAI, while the second points towards the metaphysical picture it proposes. I will address them in turn. 2.1.1 Conscious behaviour There are two possibilities concerning what to take as “conscious-like behaviour”: We might (1) use human conscious behaviour as the norm of conscious behaviour; or we might (2) advance a general characterisation of “conscious behaviour”, which is going to encompass human conscious behaviour. Possibility 1 seems tractable: There is an important amount of knowledge about human conscious behaviour. However, it is clear that to take human-like behaviour as the norm for the attribution of consciousness to any possible creature is too restrictive (and too anthropocentric). Human behaviour resulted from the particular natural history of life on Earth. Had the history been different, creatures with quite different characteristics could inhabit the planet and show behaviours very distant from the human. Now, as far as we know, the possibility is open for some of these hypothetical creatures to be conscious. Human-like behaviour could give, at best, a sufficient condition for the attribution of conscious-like behaviour and, thereby, of consciousness. Possibility 2 is more reasonable that 1, but is less tractable. We only know of human behaviour as a form of conscious behaviour.65 There is no empirical basis (at least so far) to ground a general characterisation of conscious behaviour. Moreover, a threat of circularity appears here: On one hand, it is clear that in order to develop a general description of conscious behaviour we need to be certain that the nonhuman creatures we will get the empirical information from are indeed conscious. But on the other hand, the general characterisation of 65

If some animals are conscious they also exhibit conscious-behaviour. But note that animal behaviour—not surprisingly—is similar in various respects to human behaviour. In fact, the biggest the similarity, the more we are inclined to attribute consciousness to them.

52

conscious behaviour is precisely supposed to serve as the basis for the attribution of consciousness to nonhuman creatures. Even though option 1 is too narrow, it can be claimed to be acceptable after all. Since human beings are (our) paradigmatic conscious creatures, they can provide the referent for conscious behaviour. Indeed, the main target of AI research is, clearly, to develop artificial minds where the referent of “a mind” is the human mind. Thus, a variation of SAI centred on human behaviour deserves consideration: (SAI’) If an entity S exhibits human-like behaviour, S instantiates the consciousness property. Suppose we accept this criterion and want to apply it. Should an artificial or alien creature behave in every respect as a human being in order to be considered conscious? Certainly not. Only human organisms can behave as human organisms in every respect. So, what behavioural patterns are the relevant ones? One candidate is the patterns associated with natural language skills. Indeed, Strong AI followers claim that if we had a machine that should pass the Turing test we would have produced artificial intelligence.66 Unfortunately, natural language skills do not seem essential for human consciousness. To require language skills for the attribution of consciousness would rule out preverbal infants and aphasic adults (and all the animal species).67 And even if we accepted this requirement, what is the demanded complexity for these capacities? The level of a three-years-old child? The level of an adult? Another candidate is the behavioural patterns associated with emotional responses, i.e., pain-behaviour, fear-behaviour, etc. Unluckily, 66

67

Turing (1950) presented a very influential behaviourist approach to artificial intelligence and proposed the Turing test as a criterion to attribute intelligence to machines. Roughly, a machine passes the Turing test if, statistically, an ordinary human being chatting (under appropriate circumstances) with it could not tell it is a machine rather than another human being. We do not know at what stage of development a human creature becomes conscious. But just as it is not reasonable to attribute consciousness to a fertilised ovule, it is not very reasonable to take a one-year-old baby as lacking consciousness.

53

this does not seem to entail the attribution of consciousness. Animals show emotional behaviour (different species show different degrees of similarity to humans), and yet there is clearly no agreement about whether there are conscious animals and, if there are, which ones are conscious.68 In fact, given the complexity and diversity of human behaviour it is not clear what characteristics of it are to be taken as the hallmark of consciousness. A given set of human behavioural patterns (e.g., the ones associated with language skills) can at best give sufficient conditions for the attribution of conscious-behaviour (and thereby of consciousness) to an artificial or alien entity. It does not seem necessary for a conscious entity to behave, in general, as a human being. Furthermore, it does not even seem necessary for a conscious entity to exhibit any pattern of human behaviour in particular. But are there any behavioural considerations that could suitably justify the attribution of consciousness? 2.1.2 Consciousness and behaviour The thesis that (some) human-like behaviour gives, if not necessary conditions, at least sufficient ones, for the attribution of consciousness, is challenged by the traditional “zombie argument”. It is claimed that there could be entities that behave like human beings and yet do not instantiate the consciousness property (they are “zombies”). In particular, it seems that a machine could be programmed to emulate human behaviour (in the relevant respects), despite there not being anything it is like to be that machine. Indeed, for Searle (1997) Strong AI does not make the difference between having a mind and simulating a mind, and it happens that a computer simulation of a conscious mind is no more a conscious mind than a computer simulation of an earthquake is an earthquake.69 In order to evaluate the zombie argument against SAI’ we should 68 69

For instance Carruthers (2005), despite being a realist about human consciousness, claims that animals do not have conscious mental states. Searle (1997) extrapolates his “Chinese room argument” from the case of computational functionalism for meaning to the case of Strong AI. Just as syntax does not entail meaning, human-like behaviour, he argues, does not entail consciousness.

54

distinguish between two levels at which it can be formulated. The claim can be that there are metaphysically possible worlds inhabited by zombies. In this case, the argument is that the idea of nonconscious entities behaving exactly like human beings shows no contradiction. But the claim can also be that there are physically possible worlds inhabited by zombies. This is an empirical claim, which requires for its justification to take into account natural science. Metaphysical possibility does not entail physical possibility, but physical possibility obviously requires metaphysical possibility.70 Usually, the arguments based on metaphysical possibilities are taken as definitive to test a thesis about the nature of consciousness. Even if zombies were not physically possible, the metaphysical possibility of zombies is taken to be a sufficient reason to rule out, for instance, the Strong AI view. I am reluctant to accept this kind of arguments; in a subsequent chapter I will criticise what I will call “the possible worlds strategy”.71 Here, I will briefly discuss the physical possibility of “behavioural zombies”. Are behavioural zombies physically possible? If the zombie is required to behave in every respect as a human being, they are not possible. As I said, only human beings can behave as human beings in every respect. We are embodied minds; our behaviour obviously depends on our biological characteristics. Were our bodies different, our behaviour would also be different.72 But suppose the zombie is required to exhibit only some of the characteristically human behavioural patterns. For instance, it is required to have the language skills of a competent adult. Are these zombies physically possible? The short answer is that we do not know. On the one hand, human behaviour, even when restricted to linguistic behaviour, is clearly very complex. Natural language has not only 70 71 72

In Chapters 4 and 5 I will elaborate the distinction between metaphysical and physical possibility. See Chapter 5. Indeed, I doubt that entities differing substantially in their physical characteristics could share all of their mental contents. In the case of humans, many of these contents are clearly determined by the specificities of, e.g., the mechanisms of perception, proprioception, mobility, and the spatial characteristics of the body. Minds are “embodied”.

55

syntactic rules, but also complex semantic and pragmatic ingredients. On the other hand, physical possibilities are constrained by the laws of nature. For example, it is not possible to have superconductivity at high temperatures; conductivity depends on molecular structure and temperature, and molecular structure in turn depends on the chemical possibilities of the different kinds of atoms. The upshot is that the actual world could be such that (i) the only entities which can show the linguistic behaviour of human beings are entities with a neuronal system structurally equal (in the relevant respects) to the human nervous system, or that (ii) any entity which can show the linguistic behaviour of human beings is such that there is something it is like to be that entity. If i were the case, Strong AI (as well as Weak AI) would be false. If ii were the case, behavioural zombies would simply not be physically possible. In any case, I will now argue on grounds other than the zombie argument against the idea that consciousness can be reduced in terms of behavioural patterns. The argument is based on a simple remark that I will develop afterwards: By definition, an entity S is phenomenally conscious if and only if there is something it is like to be that entity; something for the entity. Accordingly, if S is conscious this is an intrinsic (or “internal”) property of S. SAI’ does give a criterion for the attribution of consciousness. But for the realist about consciousness there is also the question of what is the ontological nature of consciousness. Indeed, to be properly justified, SAI’ should follow from the fact that consciousness has a given nature. So, what is consciousness supposed to be under the SAI’ view? SAI’ does not provide an answer for this question unless the consciousness property is taken to be a property of behaviour (which is instantiated in human behaviour). The idea would be that to be conscious just amounts to behave in accordance with some particular patterns, and nothing more. Certainly, this proposal entails SAI’, and is consonant with the spirit of Strong AI.73 But, unfortunately, it is obscure (especially for a realist about consciousness) how could a property like consciousness be identical to some characteristic of behaviour. And even if this option made sense, there is a further important problem. 73

Recall that, for Strong AI, to have a mind is to exhibit intentional behaviour.

56

Despite the fact that behaviour supervenes on physical items, the properties of the behaviour of an entity S are not intrinsic properties of S. Behaviour concerns relations between a creature and its environment. For instance, in psychological behaviourism the behavioural patterns are given by the characteristics of the relations, mediated by the creature, between stimuli and responses. By contrast, whenever an entity S is conscious this is an intrinsic property of S, and not a relational one. So, the consciousness property cannot be a behavioural property. Moreover, note that behaviour seems not only to involve a relation between a creature and the environment, but also to include an interpreter of this relation. The determination of how an entity behaves is only possible from an “intentional stance” (Dennett 1987). Thus, the supervenience base of the behaviour of an entity surely includes not only its relations with the environment, but also the phenomena corresponding to the interpretation of some chains of events as a given behaviour. It is not absurd to believe that the instantiation of the consciousness property in S is a necessary condition for S to show (the relevant) human-like behaviour. In other words, it is not incongruous to deny the physical possibility of behavioural zombies, and thus to accept SAI’ as a sufficient condition for the attribution of consciousness. But at the same time, one can believe that the consciousness property is not a property of behaviour. Then, there is still the question of what is the ontological nature of the consciousness property, and of how it determines (if it does) the behaviour of conscious entities. 2.2 Computational functionalism The distinctive thesis of functionalism is that mental states are functional states. Thus, mental states are to be individuated in virtue of the function they accomplish in the cognitive economy of an organism. Now, every mental state is realised by a brain state, but mental state types cannot be bijectively related with brain state types. A given mental state can be realised in different ways by different physical systems. In this sense, mental states are “higher-order” states. Classical functionalism is “computationalist” in that it considers the mind to be a Turing machine, and a mental state to be a Turing

57

machine state.74 A mental process is conceived as an informationprocessing process. Computationalism was originally developed by H. Putnam and J. Fodor. In Putnam’s words: According to this model, psychological states (“believing that p,” “desiring that p,” “considering whether p,” etc.) are simply “computational states” of the brain. The proper way to think of the brain is as a digital computer. Our psychology is to be described as the software of this computer—its “functional organization.” (1988, p. 73) Functionalism was popular during the seventies and eighties. Indeed, it provided the theoretical framework of cognitive science. But Putnam and Fodor later rejected this theory. The former mainly criticised its internalist character and the notion of “narrow content”; the latter acknowledged that it does not provide a general account of how the mind works (given “the frame problem”).75 Nevertheless, some functionalist theories of consciousness are still influential. But it is to be noted, firstly, that neither Fodor nor Putnam took functionalism to account for phenomenal consciousness. Indeed R. Jackendoff, who developed a cognitivist (computationalist) theory of consciousness, claims that: [W]hen we turn to the relation of the computational mind to the phenomenological mind [...] we are in the same situation as we were with the phenomenological mind-body problem. Although the computational theory of mind may help in elucidating the units and distinctions that are present to experience, and although the organization of the phenomenological mind may be more closely paralleled by the computational mind than by raw 74

75

There are variants of functionalism besides computationalism. In particular, there is “causal role functionalism”, originally developed by D. Lewis. See Lewis (1983). See, e.g., Putnam (1988) and Fodor (2008). Fodor (2008) maintains that his “language of thought” version of computationalism is the best available theory about how the mind works, even though this theory cannot be true.

58

neurophysiology, we still have gotten nowhere with the essential nature of consciousness itself. (1987, pp. 21-22) Secondly, it is also to be noted that even though the theory of Dennett (1991) is computationalist and Chalmers (1996) speculates that a functionalist theory could be right, the former is an eliminativist, and the latter defends a form of dualism. A functionalist view of (phenomenal) consciousness is a thesis of the following form: (CF) If a system T has a functional architecture with property F, then T instantiates the consciousness property. Obviously, a functionalist theory of the mind must provide its functional architecture and, if it pretends to account for consciousness, it must specify the property F. But here the important point is that, in a functionalist view, it is in virtue of the instantiation of a functional architecture with some property F that a system T is conscious. In other words, what accounts for the instantiation of consciousness and determines its characteristics are not the physical properties of the system, but its informational states and processes. The consciousness property is not taken to be a physical property. Some appropriate “hardware”, certainly, is required for the instantiation of a functional architecture. But the consciousness property cannot be reduced in terms of physical properties because consciousness is a functional feature. As N. Block says: “The difference [between physicalism76 and functionalism] is that the functionalist says that consciousness is a role, whereas the physicalist says that consciousness is a physical or biological state that implements that role.” (2007, p. 118). Functionalism has some appealing ideas and consequences. First, it is widely accepted that (at least many types of) mental states have some representational content; that their cognitive function is indeed to 76

In this quote, “physicalism” does not stand for the general metaphysical thesis I labelled this way. It stands for the claim that the consciousness property is a physical property and not a supervenient one.

59

represent;77 and that “to think” corresponds to an information-processing activity. Now, some processes involving paradigmatic conscious states, like sensory processes, are commonly considered to be, essentially, information processes, in accordance with the functionalist view of the mental. Secondly, for functionalism (conscious) minds are multirealisable: A system is not required to be made out of neurons and biological tissues to be conscious, in accordance with AI views. Now, if some mental states instantiate the consciousness property, and mental states are exhaustively captured by their cognitive function, it follows that the consciousness property is a functional property. 2.2.1 Functionalism and consciousness The possibility for a functionalist theory of consciousness to be right has received serious criticism. Several arguments are taken to show that functionalism cannot account for the reality of phenomenal experiences or to show that, even worse, functionalism is refuted by the existence of consciousness. I will briefly discuss the “Chinese nation” thoughtexperiment, the “inverted spectrum argument”, and the antifunctionalism version of the zombie argument. Block (2007) imagined the following scenario: Suppose that the mind is—as functionalism claims—a functional structure, and that we have a complete description of it. Given multirealisability, this structure can be realised by different systems. For instance, we might take a large enough group of people, e.g., the population of China, and organise them accordingly. For example, one individual could perform the functional work of one neuron. Thereby, we would have a realisation of a mind. Indeed, for Block “[i]t is not at all obvious that the China-body system is physically impossible. It could be functionally equivalent to you for a short time, say an hour.” (2005, p. 71). Now, is it reasonable to expect that there would be something it is like to be the whole “China-body”? For Block, it is not. This “homunculi-headed system” does not plausibly realise a subject of experience. Block’s argument does not include a proof of why the Chinese 77

But there is wide disagreement about where the representational function of mental states derives from and about how to determine their representational content.

60

nation does not realise a conscious mind. But it shows a consequence of CF that looks hard to accept. Why would a mind, and in particular a conscious mind, result from the functional organisation of this “body”? For Block, something is missing; functional organisation does not seem to be a sufficient condition for consciousness. He says: What makes the homunculi-headed system […] just described a prima facie counterexample to (machine) functionalism is that there is prima facie doubt whether it has any mental states at all— especially whether it has what philosophers have variously called “qualitative states”, “raw feels,” or “immediate phenomenological qualities.” (2007, p. 73) I take it to be true that functionalism “has an absurd conclusion that there is no independent reason to believe.” (Block 2007, p.77). Furthermore, the China-body thought-experiment can be used to show that functionalism falls short of giving an account of “the unity of consciousness”.78 Why are we prone (if we are) to reject the possibility of this system realising a conscious mind? After all, the scenario is such that the components of the system are already conscious entities. I believe one reason is that we do not see how solely in virtue of functional organisation a unified consciousness would appear.79 Even though it is the case that each homunculus of this system has the consciousness property, no functional organisation seems to be able to explain how a new single conscious mind could appear by virtue of this organisation, so that we would have a subject of experience. Now, the functionalist can certainly “bite the bullet” and claim that, despite 78

79

The “unity” of consciousness is a prominent characteristic of it. Bayne (2010) distinguishes between “subject unity”, “representational unity” and “phenomenal unity”, and claims that human consciousness complies with the three conditions. The “unity of consciousness” concept embraces all these unities. I will discuss this topic in Chapter 3. Regarding in particular perception, the question of how to account for its phenomenal (and representational) unity is called “the binding problem”. For instance, empirical research has shown that there are in the visual system different modules with specific processing tasks. The question is how are the different outputs “bound” to give rise to a unified conscious percept.

61

contrary intuitions, the China-body would indeed realise a mind because there is nothing else to minds and consciousness than a functional structure. In particular, she can claim that functional organisation could explain “qualitative states” and the unity of consciousness; the problem—the functionalist might claim—is that we do not yet have enough knowledge of the structure of the mind. There is another influential argument against functionalism about consciousness. It says it cannot account for the possibility of “qualia inversion”.80 Recall that “qualia” are the constituents of phenomenal contents and characters, which individuate them. Differences in the what-it-is-like-ness of being in two (conscious) mental states M and M’, are due to differences in the qualia that constitute the phenomenal contents of M and M’. For instance, (under normal conditions) the whatit-is-like-ness of seeing something red is different from the what-it-islike-ness of seeing something blue; thus, the phenomenal content of the perceptual state corresponding to seeing red differs in its qualia from the phenomenal content of the perceptual state corresponding to seeing blue. Let us call “p-red” and “p-blue” the respective qualia associated to seeing red and seeing blue for a subject S. Whenever S looks at a clear sky he has a p-blue experience; whenever he looks at a ripe tomato he has a p-red experience. Now, imagine there is a subject W who has his phenomenological spectrum “inverted” relative to (the reference subject) S: Whenever W looks at a clear sky he has a p-red experience; whenever he looks at a ripe tomato he has a p-blue experience. The “inverted spectrum argument” hinges on the idea that functionalism is incompatible with this possibility of a subject experiencing an inverted spectrum. On one hand, it is claimed that the state with phenomenal content p-red can play—in the cognition of S—exactly the same functional role that the state with phenomenal content p-blue—in W’s cognition. On the other hand, it is noted that functionalism individuates mental states solely in virtue of the functional role they play. Therefore, it is concluded that functionalism cannot account for the phenomenal 80

This argument, called the “inverted spectrum” argument, was presented by Shoemaker (1982). The question of the possibility of inverted spectrum scenarios can be traced back to Locke (1689).

62

differences in inverted spectrum cases. Something is missing in the functionalist picture, namely, the phenomenal contents and characters, i.e., the “mental paint” (Block 2007). This argument can be formalised as follows: (1) Functionalism is true: Mental states are exhaustively characterised and differentiated by their cognitive function. [supposition] (2) Realism about qualia: Conscious mental states have phenomenal contents. [premise] (3) Inverted spectra are possible: Mental states can have different phenomenal contents and yet play the same functional role. [premise] (4) Functionalism is incompatible with the possibility of inverted spectra scenarios. [from 1 & 3] (5) Functionalism is false. [conclusion] Certainly, this argument depends on two quite controversial premises, namely 2 and 3, which indeed lay at the core of the problem of consciousness. However, I consider the argument to be compelling. I think there are phenomenal contents that are not reducible in terms of representational contents or cognitive functions. Hence, I consider inverted spectrum cases (which concern in particular colour-qualia) to be possible, even though this possibility is hard to be proved if it can be proved at all. These are difficult questions that I will discuss later.81 Functionalism about consciousness can also be threatened by a version of the zombie argument. We can—it is claimed—conceive an entity where the functional architecture of human cognition is replicated and which, nevertheless, does not have the consciousness property—it has “absent qualia”.82 What this argument is purported to show is that functionalism does not account for the very existence of the consciousness property. While the inverted spectrum argument is aimed at showing that functionalism cannot account for differences between 81 82

See Chapter 6. See, e.g., Levine (1988).

63

phenomenal contents, the zombie argument (as does the China-body argument) goes further and concludes that functionalism cannot account for the existence of this kind of contents. Now, as I said concerning Strong AI, we must distinguish between the metaphysical and the physical readings of zombie arguments. When “functional zombies” are claimed to be metaphysical possibilities, this is supposed to show that the consciousness property cannot be identical to a functional property. The argument is that there is no contradiction in the idea of an entity functionally identical to a human being and yet not conscious. When functional zombies are said to be physically possible, this would show not only that the consciousness property is not a functional property, but also that, in the actual world, functional considerations do not give sufficient conditions for the attribution of consciousness. As I said before, in a subsequent chapter83 I will discuss the possible worlds strategy, which concerns metaphysical possibilities. Now, I will briefly address the question of physical possibility. So, are functional zombies physically possible? Like in the case of behavioural zombies, the short answer is that we do not know. We do not know if an entity different enough from human beings to be nonconscious could nonetheless realise the same cognitive functions that human beings. The possibility of functional zombies is subordinated to the multirealisability thesis: A given functional architecture can be implemented in different physical systems. Prima facie, multirealisability obtains. There are mechanic calculators as well as electronic ones; a given algorithm can be written in different computer languages and be run in machines with differing physical architectures; etc. However, note that the more complex the functional architecture, the more constraints on the physical characteristics a system must meet to be able to perform the function. The space of possible physical realisations of functional structures is constrained by at least three types of factors: Information, pragmatic and physical factors. First, in order to realise a structure with n possible informational states, the hardware must be able to be in at least n possible internal physical states. Second, a universal Turing machine is theoretically capable of computing any computable 83

See Chapter 5.

64

function, but these machines are supposed to have an infinite memory and to dispose of infinite time to perform the computations. In practice, every machine has limited storage and must process the information in a reasonable lapse of time. Third, any machine must be built on the basis of the laws of nature.84 For instance, no signal can travel at a speed higher than the speed of light (according to Relativity theory). Certainly, it is not easy to give a formalisation of the relation between the complexity of a functional architecture and the amounts of constraints that the corresponding realiser must meet. But the idea can be captured with the help of an analogy: Think about the characteristics of mechanical tools and the (noncomputational) functions they can accomplish. To perform a hammer function, a body just needs to be rigid, inelastic, and dense, with an appropriate simple form and dimension. To perform an axe function, a body must meet more constraints, for instance, it needs to have a very solid sharp border. To perform a jacking function, a body needs, in particular, to have some mechanical structure that amplifies force. To perform an engine function, a body requires some sophisticated mechanism that converts, say, chemical energy to kinetic energy, and must be able to endure high temperatures and friction. And some tools are not even possible; for example, no tool can perform a perpetual motion machine function according to the 2nd law of thermodynamics. Now, the functional architecture of the human mind is no doubt very complex. The number of constraints a physical system must meet to be able to implement its functional architecture and processing capacity is surely huge. As a result, it might turn out to be the case that only the human nervous system or a system very similar to it (in the relevant respects) meets all the requirements.85 If this is the case, and it happens that the human nervous system and any system very similar to it necessarily instantiate the consciousness property, functional zombies are not physically possible. Consider in particular the P-conscious mental states and suppose 84

85

Block (forthcoming) labels this constraint the “Disney Principle: that laws of nature impose constraints on ways of making something that satisfies a certain description.” On this see Shapiro (2004).

65

that they play the functional roles of A-consciousness.86 Given the complexity of human cognition, it might turn out to be the case that every physical system, in order to be able to realise the A-consciousness functions, must have a physical structure that happens to instantiate (with physical necessity) the (physical or supervenient) consciousness property. By analogy, note that for a physical system to be able to realise a sword it must have a physical structure which instantiates the (physical) rigidity property. In other words, the actual world could be so that, whenever a system can perform the A-consciousness functions, this system is so that there is something it is like to be it. Block (2007) proposed the A-consciousness/P-consciousness distinction primarily as a conceptual one. If one accepts this distinction, one is committed to the metaphysical possibility of entities instantiating the A-consciousness mental architecture of human beings, but without any P-conscious mental state playing an A-conscious role.87 Besides, Block discusses if in human cognition there are indeed cases of Aconscious mental states that are not P-conscious. If this were the case, there would be empirical evidence to support the physical possibility of functional zombies. It happens that there is an interesting psychological phenomenon called “blindsight”. Subjects who suffer from blindsight claim they are completely blind in some part of their visual field, and this is certainly a consequence of some damage that can be observed in their visual cortex. However, experiments show that they acquire information about the items belonging to the blind field: When asked to “guess” some features of those items, their answers turn out to be correct with a percentage well above chance. At first glance, it might thus appear that blindsight is a case where there is A-consciousness in the absence of Pconsciousness. However, this is controversial. Notice that a blindsight “perception” does not produce beliefs, by contrast with ordinary P86

87

For the distinction between access and phenomenal consciousness see Chapter 1. Note that since P-consciousness is not a functional notion (by contrast with Aconsciousness) it is not obvious how it could be incorporated in a functional framework. Indeed, the inverted spectrum argument partly hinges on this point. When restricted to visual perception, Block (2007) calls this hypothetical scenario “superblindsight”.

66

conscious visual perception. Blindsighters think they are randomly guessing, and do not use the information they acquire in appropriate ways; for instance, they do not spontaneously grasp a “blindlysighted” object. In fact, it is remarkable that despite the appealing conceptual difference that can be drawn between A-consciousness and Pconsciousness, in human cognition it seems that every A-conscious mental state role is played by a P-conscious mental state.88 It can be the case that this is just a coincidence, and functional zombies are physically possible. But, again, it can also be the case that, in the actual world, the functional structure of the human mind is so complex that only Pconscious mental states—in virtue of whatever properties of whatever kinds they have, which correspond to P-consciousness or physically entail it—can perform these functions. 2.2.2 Functions and information Even if it were the case that functional zombies are not physically possible, and in our world any entity which realises the functional architecture of the human brain also instantiates the consciousness property, this does not entail that functionalism about consciousness is true. I will present two arguments against the idea that the consciousness property is a functional property. Recall that the concern here is with computational functionalism, and note that the computational notion of function differs from the more general “causal role” notion and, in particular, from the biological notion. In causal role functionalism, there is a straightforward derivation of functions from the causal structure of physical processes.89 Regarding the biological function, this is a teleological notion framed in natural selection theory; biological functions are determined by the selective history of an organism. By contrast, the computational notion of function is closely related to the concept of “information”: Computational functions are performed by processing information. 88

89

According to the “global broadcasting theory” by Baars, the mental states that play the A-consciousness function are able to do so in virtue of being Pconscious. See, e.g., Kim (2005).

67

The functional architecture of a machine, in Turing machine jargon, is determined by its Turing table. It is a set of rules relating information inputs, information outputs, and internal informational states. In more intuitive terms, a functional architecture has to do with the way information is processed; it is an algorithm. Accordingly, “a functional property” is a property of an algorithm, which concerns the characteristics of the information processing described by the algorithm. Examples of functional properties are that the algorithm processes information serially; that it works with binary information units; that it includes some recursive, or some threshold, or some reverberation routine. Now, the question is what is the ontological nature of functional structures, functional properties, and, at the end, of information. Since there is multirealisability, functions and information processes cannot be identical to physical states and processes. But if physicalism is true, they do supervene on physical items. This seems to be correct: Any functional architecture is realised in some hardware, and whenever some function is being accomplished there are underlying physical processes going on. For instance, when computers process information there are electric current flows, mechanical processes, and so on. Indeed, engineers design the physical properties of a machine for it to be able to perform the computational functions they want it to perform. Similarly, whenever our cognitive system process, say, visual information, there are photochemical reactions in the visual receptors, electrochemical activities in the visual nerves and cortex, etc.90 Thereby, the possibility of the consciousness property being a functional property is not at odds with physicalism. Now, a given functional architecture can be realised by different adequate hardware, but the converse also happens to be true: A given hardware can be used to accomplish different functions. Moreover, a single type of physical process involving a given hardware can play different functional roles. For example, let us consider some Boolean function. Certainly, this function can be realised by logic gates made up with transistors or by logic gates made up with diodes. But, conversely, 90

Empirical research suggests that visual experience is given, somehow, by the activity of the ventral visual cortex, and not of the dorsal one. See Jeannerod & Jacob (2005).

68

a given logic gate can accomplish different functions in a circuit, e.g., a switch function or a filter function. Moreover, the same type of physical process occurring inside the logic gate chip can correspond to the accomplishment of different functional operations, like filtering or switching; it depends on the way the logic gate is related with the other elements of the circuit. So, not only multirealisability of functions obtains, but the converse of multirealisability also obtains. The physical entities constitutive of the hardware, their properties, and the relations between them, do not uniquely determine which function the hardware realises. In other words, functional properties are not intrinsic (or “internal”) properties of the hardware that realises them. True, the physical constituents of the hardware play a crucial role for the supervenience of some function. But the point is that something else is missing. The supervenience base of a function realised by a hardware S includes not only S but also physical systems other than S. The idea that functional properties are not intrinsic properties is sound. It makes no sense to attribute a function to a piece of hardware that is isolated from other pieces of hardware. A physical system S realises a function when it plays a role in the processing of a flow of information: S receives an information input, performs its function, and then gives an information output. These inputs and outputs are essential for the determination of the function that S performs, and obviously they involve physical processes occurring outside S. Now, recall that whenever S instantiates consciousness it instantiates an intrinsic property. Since the functional properties of S are not intrinsic properties of S, we can thus conclude that consciousness cannot be a functional property. The second argument against the thesis that consciousness is a functional property elaborates on the previous one. The idea is that the supervenience base of a function realised by some hardware also includes intentional agents. In fact, I will argue that “information” is an epistemological notion. There have been some interesting attempts to construe information

69

in an objective way.91 But there is the problem that it makes no sense to consider that there is any “information” in the absence of some epistemological subject who can interpret some physical state as an information state. Certainly, something can be a computer and store information even if, actually, nobody uses it, or nobody knows about its existence, or nobody knows how to interpret its language. But something cannot be a computer and be said to store information in a world not inhabited by intentional agents who could use it as a computer.92 Clearly, a physical system must have appropriate characteristics to be able to realise some information processing, and whenever there is a change in its informational states there is also a change in its physical state. But no physical state or process is, intrinsically, an information state or process. In a computer’s hard drive, RAM memory, or screen, what do we have? We have electromagnetic fields, electric charges, photoelectric reactions, and so on. There are no characters, no symbols, and there is no information. Computers do have intrinsic physical states, but they do not have intrinsic informational states. As Searle says “the deeper argument against computationalism is that the computational features of a system are not intrinsic to its physics alone, but require a user or interpreter.” (1997, p. 130). Moreover, any physical process can be viewed as an informationprocessing process when described in terms of parameters and variables. In Searle’s words: “[A]ny system of any complexity at all admits of an information processing analysis.” (2002, p. 110). The “input” is given by the initial state of the system, the “processing” corresponds to its evolution, and the “output” is given by its final state. Searle gives the example of a falling stone. The free fall of the stone can be viewed as a 91

92

See, e.g., Hofstadter (1979) and Dretske (1981). Recall that here the concern is computationalism. The biological notion of function permits the elaboration of an alternative concept of “information” which can be claimed to have an objective reference on other grounds. Dennett (1987) claims that machines have “derived intentionality”. They inherit their intentionality from the intentionality of the designers that conceived them to operate in accordance with their intentions. However, Dennett denies that minds have “original intentionality”. These derive their intentionality, in turn, from the natural selection “designer”. In any case, for Dennett any intentional state of a machine is a derived state.

70

(analogue) computation of the free fall equation. Indeed, any physical system P can be viewed as having information about another system Q if there is a causal relation connecting some properties of P with some properties of Q, and we dispose of a theory which describes these relations. From knowledge of the relevant properties of P and the relevant laws of nature, we can derive knowledge of some properties of Q. Now, if functions, functional architectures, functional properties, information, and so on, are subject-relative, it follows that, somehow, intentional subjects enter into the supervenience base of a function realised by a hardware S. Functions certainly supervene on physical items, and we can say that some hardware “realises” some function, but the supervenience base extends well beyond the realisation base. By contrast, once again, for an entity S’ to be conscious is for S’ to instantiate an intrinsic property. Hence, consciousness cannot be a functional property. In sum, physicalism requires functions to supervene on the physical. This seems to be the case given, firstly, that a hardware is necessary to realise them and, secondly, that there is multirealisability. But, given the converse of multirealisability, it is also the case that functions are not intrinsic properties of the hardware that is said to realise them. By contrast, whenever there is something it is like to be S this is a property of S, not of a physical system which includes S but is larger than S. Consequently, consciousness cannot be a functional property. 2.3 The internal character of consciousness The rejection of both Strong AI and computational functionalism is based on the remark that it is by definition that consciousness is an intrinsic (or “internal”) property. If one does not agree with the idea that consciousness is an intrinsic property, one is not taking “consciousness” to refer to the property defined by: “An entity S is phenomenally conscious if and only if there is something it is like to be S.” A physical property of a system including S but larger than S, or a property supervening on a base including S but also other physical systems, cannot be identical to the consciousness property instantiated in S.

71

Now, I argued against panpsychism,93 i.e., against the idea that consciousness is instantiated in fundamental entities. Accordingly, to say that when S instantiates consciousness it instantiates an intrinsic property, amounts to saying that this property is an internal property of a complex system. In this sense: (C3) The consciousness property has an internal character. Internal (or intrinsic) properties contrast with relational properties. For instance, the mass of a system, its electric charge, its density and its chemical composition, are internal physical properties, while its position, velocity, and acceleration, are relational.94 Similarly, the property of an individual F being a father is relational, while his property of being human is intrinsic. Of course, internal properties of a system S can be relational properties among constituents of S. For instance, the relative positions of the atoms in a molecule are internal properties of the molecule, and the relation of F being the father of S is an internal property of the pair {F ; S}.95 Recall that for a mental state M to be conscious is for M to instantiate the m-consciousness property, that for an entity S to be conscious is for S to instantiate the c-consciousness property, an that an entity is conscious iff she is in some conscious mental state.96 In short, an entity instantiates the c-consciousness property in virtue of being in some mental state that instantiates the m-consciousness property. Subsequently, I claimed that the consciousness property instantiated in a (conscious) subject—the c-consciousness property—has an internal 93

See Chapter 1. I take the dispositional physical properties, like transparency or electrical conductivity, to be relational ones. They are a function of the internal properties of a system, but concern its (potential) interactions with other systems. See Gnassounou & Kistler (2005). 95 Note that the dichotomies physical/supervenient and internal/relational are orthogonal. 96 In Chapter 1 I made the following distinction, following Rosenthal (2005): When it is an entity (or creature) that is P-conscious, it instantiates one token of the cconsciousness property; and when it is a mental state that is P-conscious, it instantiates one token of the m-consciousness property. 94

72

character. For these claims to be compatible, it may seem that an internalist view of mental states is required. It may seem that the conception of the consciousness property as an internal property involves a rejection of externalist views about the mental and, in particular, of externalism about representational content97 and of the “extended mind” hypothesis. 98 However, I think that this is not the case. I take externalism about representational content to be compatible with the internal character of consciousness. Externalism does not involve a commitment to an ontological identification of the vehicle of a mental state with its representational content. The representational content of a mental state M of a subject S can be external, and the taxonomy of mental states can follow external considerations, in spite of M having properties that are internal to S, i.e., properties that are physical properties of S or that supervene (exclusively) on physical properties of S.99 Similarly, the internal character of consciousness can probably be accommodated in the extended mind hypothesis (EMH). For EMH, although the physical realisation of the (conscious) mind of a subject S extends the body of S, the vehicle of a given mental state of S always involves to some extent physical constituents of S. Therefore, it can be claimed that the m-consciousness property of an extended (conscious) mental state of S is instantiated in, or (exclusively) supervenes on, these constituents of S. However, this solution is probably at odds with the spirit of the EMH. If the physical realisation of the mind of a subject S extends the body of S, it is to be expected for the properties (like consciousness) of the mental states of S to likewise have a physical realisation that extends the body of S. So, given that consciousness has an internal character, either one accepts that some properties of an extended mental state are internal to the subject, or one rejects the extended mind hypothesis at least for conscious mental states. In general, every physicalist picture of a mental state is going to involve some of the physical constituents of the corresponding subject. Otherwise, on which grounds would some mental state be associated 97

See, e.g., McGinn (1977) and Burge (1979; 1986). See Clark & Chalmers (1998). 99 Internal mental properties are often labeled “narrow properties”. 98

73

with a given subject? Mental states are embodied states and consciousness is an embodied property. Even if it is the case that mental states are to be individuated following externalist criteria, the vehicles of these states involve physical constituents of the corresponding subjects. Consciousness is realised in these physical constituents. 2.4 Conclusion Both Strong AI and (computational) functionalism take the consciousness property to be a supervenient property. Certainly, this possibility is in accordance with physicalism. But I argued, firstly, that the behaviour of an entity S supervenes on a base that includes not only S but also physical systems other than S. And secondly, that a function realised by some hardware H is not an intrinsic property of H. By contrast, consciousness has an “internal character”: It is an internal (or “intrinsic”) property of the conscious entity. Therefore, I concluded that consciousness is neither a behavioural nor a functional property and thereby rejected Strong AI and functionalism. In the next chapter I will, firstly, discuss an influential family of views that are compatible with the internal character of consciousness, namely, higher-order representation theories (HOR). Secondly, I will discuss “the unity of consciousness”. And Finally, I will address “the explanatory gap” and defend physicalism from the anti-physicalist conclusions that this question motivates.

75

CHAPTER 3: SUBJECTIVITY AND THE UNITY OF CONSCIOUSNESS 3.0 Introduction The first chapter led to the following conclusion: (C1) There is a property X, which is a physical property or a supervenient property, such that for an entity S to be (phenomenally) conscious is for S to instantiate X. Statement C1 is a metaphysical claim. Concerning the epistemological question of the attribution of consciousness I argued that: (C2) In order for a subject S to establish whether another entity W is conscious, she needs to appeal to an empirical theory proposing third-person point of view criteria for the attribution of consciousness. In the previous chapter I discussed two influential views on consciousness that take it to be a supervenient property. The first one, Strong AI, considers that to be conscious is to behave some way. The second one, computational functionalism, takes consciousness to be a functional role in the functional architecture of a cognitive system. I rejected both views on the following grounds: (1) Behavioural patterns concern relations between the behaving entity and the environment. (2) Whenever a system S has the property F of realising a function f, F is not an intrinsic property of S. And since: (C3) The consciousness property has an internal character. I concluded that: (3) The consciousness property cannot be a behavioural or a functional property.

76

Note that C3 establishes a constraint for any theory of consciousness. The consciousness property is not a relational property. The condition of S’s being conscious is not reducible in terms of relations between S and other entities. Any relation constitutive of the consciousness property instantiated in S, relates constituents of S. In this chapter, firstly, I will address an influential family of theories of consciousness that are compatible with its internal character, namely, higher-order representation theories (HOR).100 The mind can be viewed—under an atomistic picture—as a bunch of many interacting but distinct mental states, some of which are conscious. Now, it can be claimed that whenever the consciousness property is instantiated in a mental state this fact results from the existence of a representational relation between this state and another mental state of a higher-order. This is the general idea that characterises HOR theories. Secondly, I will address the question of the unity of consciousness, which has important metaphysical and epistemological implications. Consciousness has, in particular, “subject unity” and “phenomenal unity”. A single subject of experience cannot instantiate more than one token of the c-consciousness property.101 And whenever one is having a phenomenal experience, in spite of its phenomenal richness it possesses a “conjoint experiential character” (Bayne 2010, p. 11). If we take a subject as being in various simultaneous mental states that instantiate the m-consciousness property, these “will be subsumed by a single conscious state—the subject’s total conscious state.” (Bayne 2010, p. 16) and give rise to a single token of the c-consciousness property. Finally, I will discuss the “explanatory gap”. S. Kripke (1972) presented an influential argument against physicalism, based on the following idea: Psychophysical statements that identify physical phenomena with experiential psychological phenomena, if true, are necessarily true. However, they seem to be contingent. Subsequently, J. 100

The terminology I will use for HOR theories is taken from the taxonomy in Carruthers (2005). 101 Recall the following distinction: When it is an entity (or creature) that is Pconscious, it instantiates one token of the c-consciousness property; and when it is a mental state that is P-conscious, it instantiates one token of the mconsciousness property.

77

Levine (1983) claimed that the appearance of contingency is due to an explanatory gap in these statements: The physical phenomena do not seem to account for the existence and characteristics of the psychological phenomena. If the physicalist acknowledges that there seems to be a gap, he must explain and dispel this intuition in a way that safeguards physicalism. 3.1 Higher-order theories of consciousness Higher-order representation theories of consciousness, which include higher-order experience/inner-sense theories (HOE) and higher-order thought (HOT) theories, propose necessary conditions for a mental state to be conscious. The main idea is that some kind of representational relation between a mental state M and another mental state M’ is required in order for M to be conscious. Schematically: (HO) For a mental state M to be (phenomenally) conscious, M must be (actually or potentially) represented by another mental state M’. The main differences between alternative versions of HOR theories lie in the nature and characteristics of the link between the representing state M’ and the represented state M. Roughly, if the relation is of a perceptual kind (by analogy with the perception of objects) we have a HOE theory,102 and if it is of a referential kind we have a HOT theory. HOR theories elaborate the following intuition: When a mental state is conscious, one has access to (at least part) of its content. In particular, whenever there is something it is like to be in a mental state M, the subject in M knows what it is like to be in M, and this knowledge is no doubt of a special kind. Exactly in which way it is special is a controversial matter, but there surely is a cognitive faculty that “gives access”, in some sense, to the content of some mental states. Indeed, it seems difficult to conceive the situation of one being in a conscious mental state without knowing what it is like to be in that state. Accordingly, HOR theorists take the consciousness property to result from a kind of meta-representation faculty: For a mental state to 102

See, e.g., Lycan (1996).

78

be conscious is for it to be represented by another mental state. It is the access a higher-order mental state M’ has to the lower-order one M what explains the special kind of access one has to M, manifest in the “whatit-is-like-ness” of being in M. The most influential HOR theories are the HOT cognitive theories of D. Rosenthal (2005) and P. Carruthers (2005).103 Roughly, the idea is that a mental state is phenomenally conscious if there is a (actual or potential) thought about that mental state. Creatures that can only form first-order thoughts cannot be conscious. Creatures that can form second-order thoughts can be conscious of their first-order mental states. And creatures that can, in turn, form (at least) third-order thoughts enjoy introspection capacities or “self-consciousness”: They can be conscious of having conscious mental states. The following quotes illustrate these views. Rosenthal says: Conscious states are simply mental states we are conscious of being in. And, in general, our being conscious of something is just a matter of our having a thought of some sort about it. (...) Since a mental state is conscious if it is accompanied by a suitable higherorder thought, we can explain a mental state’s being conscious by hypothesizing that the mental state itself causes that higher-order thought to occur. (2005, pp. 26-27) Carruthers says: [...] the conscious status of a mental state or event consists in its noninferential availability to a ‘theory of mind’ or ‘mind-reading’ system capable of higher-order thought. And a conscious experience, in particular, will be an experience that is available to cause higher-order thoughts about the occurrence and content of 103

Carruthers (2005) presents his theory “as building on, but rendering higher-order, the ‘global broadcasting’ theory of Baars.” (p. 8). Baars (1997) characterises his “global broadcasting theory” as follows: “Consciousness seems to be the publicity organ of the brain. It is a facility for accessing, disseminating, and exchanging information, and for exercising global coordination and control.” (p. 7. Italics in the original).

79

that very experience. (2005, p. 93) The main difference between these two proposals concerns the necessity of the lower-order mental state actually causing the occurrence of the higher-order one.104 For Rosenthal, it is necessary to be in a higher-order mental state for the lower-order one to be conscious. For Carruthers, just the disposition to be in a higher-order mental state makes the lowerorder one conscious. His position is motivated by the adoption of some version of “consumer semantics”.105 I will not be concerned with the details of any particular HOT or HOE theory, but discuss the general idea stated in HO and use Rosenthal’s and Carruthers’ theories to support some claims. I will briefly address some problems I consider HOR theories face as accounts of phenomenal consciousness. Suppose there is a mental state M and another mental state M’ that appropriately refers to M. In virtue of this reference relation—according to HO—M instantiates the consciousness property. If there was not a reference relation between M and M’, there would not be an instantiation of the consciousness property in M. But—according to physicalism—the consciousness property instantiated in M has a physical nature. The presence or absence of this property is a fact of a physical nature, and the appearance or disappearance of this property is an event of a physical nature. So, the reference relation that accounts for these facts and events must also be of a physical nature and involve processes of a physical nature. Something that is not of a physical nature cannot account for the instantiation of something that has a physical nature. In short, for physicalism to be compatible with HO not only the mental states M and M’ must have a physical nature, but the intentional relation between M (the represented) and M’ (the representation) must also have a physical 104

Further differences concern the conceptual or not conceptual nature of the higher-order thought’s contents, and the relation between consciousness and sensory qualities. For Rosenthal, but not for Carruthers, higher-order thoughts always have conceptual content, and a mental state can have sensory qualities (qualia) even when it is not conscious. 105 “Consumer semantics”, in Carruthers’ words, stands for “either teleosemantics, or some form of functional or inferential-role semantics.” (2005, p. 93).

80

nature. Here appears a first difficulty with HO. Certainly, there have been many attempts to naturalise intentionality. In particular, there are the “causal theories of mental content” (CTR), based upon the idea that a mental representation “R” represents R because “R” was caused by R. Causality is clearly a naturalistic notion, and a causal relation seems to be necessary for “R” to represent R. However, there are two important problems the CTR face. The first one is that there are representations “R” that cannot be caused by some R, because this R is a fiction or an abstract object. For instance, there is no need of causation for “unicorn” to represent (possible) unicorns, or for mathematical terms to represent mathematical “objects”. The requirement of the existence of a causal relation between R and “R” seems to apply only for elementary representations. The second problem is that causality alone is not sufficient to uniquely determine reference and to account for the possibility of misrepresentation; this is known as “the disjunction problem”.106 In fact, it is necessary to appeal to something else in order to individuate representational contents. Typically, this additional ingredient is proposed to be some kind of function. For instance, F. Dretske (1995), R. G. Millikan (1993), and D. Papineau (1998) proposed various teleological notions of function, based on the fact that cognitive systems are the result of natural selection. Unfortunately, these suggestive attempts also face some difficulties. They are “bottom-up” approaches, i.e., they start by addressing the simplest cognitive systems and, on this base, they gradually seek to cope with more complex ones. But as the cognitive systems become complex, new problems appear for the causal accounts. Indeed, since consciousness in particular seems to involve higher cognitive processes, Horgan and Tienson (2002) argue that causal theories cannot explain what they call “phenomenal intentionality”. Probably, intentionality cannot be reduced, without reminder, in naturalistic terms. Even though—for a physicalist—everything has a physical nature, from this metaphysical claim it does not follow that intentional relations can be fully captured in naturalistic terms. Now, since HOR theories aim at explaining consciousness in 106

See, e.g., Fodor (1990) and Jacob (1997).

81

virtue of (actual or potential) representational relations, they inherit all the difficulties of the project of naturalisation of intentionality. Moreover, note that the reference relation that HOR theories use to account for consciousness is of a particular kind: It holds between mental states of a single mind, and not between a subject and an intentional external object. Unfortunately, this kind of reference relation is even more difficult to naturalise, because it is not possible to appeal to external elements like, e.g., interactions among several subjects.107 But suppose that the representational relation can be naturalised and, in particular, that one which is internal, i.e., between mental states of a single subject, can be reduced in terms of physical states and processes of the cognitive system. Now, consider what Rosenthal says: The appearances of consciousness reflect the contents of our HOTs; what it’s like for one to be in a particular mental state is a matter of how one’s HOT represents that state. (2005, p. 14) It might be true that a conscious mental state M is always represented by another mental state M’. Moreover, it might be true that the what-it-islike-ness of being in M is (at least partly) determined by the way M’ represents it. Indeed, the availability of some mental states for other mental states to represent them can explain different characteristics and capacities of our cognition, like introspection of mental contents and the production of related reports. But there is still the question of why a mental state M would be phenomenally conscious in virtue of being represented by another mental state. This is “the rock objection”: 108 The mere fact that a mental state M’ represents another mental state M does not seem sufficient to explain why there is something it is like to be in mental state M. Rosenthal just affirms this is the case: Intentional states are responsible for there being something it is like for one to be in qualitative states, and for differences in what it’s like for one to be in states of various sorts. (2005, p.14) 107

Davidson (2001) argues that a process of “triangulation”, which involves several subjects, is required in order to determine reference and intentional content. 108 See Goldman (1993) and Stubenberg (1998).

82

The same remarks can be made about Carruthers’ theory. He says: Such an account [HOT dispositional theory] [...] can explain how conscious experiences have a subjective dimension of what it is likeness. This is their higher-order analog content, in virtue of which they themselves (and not just the objects and properties that their first-order contents are of) are presented to us nonconceptually or in analog fashion. (2005, p.137. Italics in the original) Additionally, Carruthers claims that the mere availability to a HOT would render a mental state M phenomenally conscious: On this account [dispositional HOT theory], phenomenal consciousness consists in the dual perceptual content (both firstorder and higher-order) possessed by those perceptual states that are made available to HOT (given the truth of some or other form of consumer semantics). 109 (2005, p. 76) If it is mysterious why a mental state M would have the consciousness property in virtue of being actually represented by another mental state, it is even more mysterious why it would have it just in virtue of a potentiality. The consciousness property is either instantiated or not; it is not a dispositional property. If there is a process that is responsible for the instantiation of the consciousness property, it must actually obtain for this property to be instantiated. The instantiation of a physical or supervenient nondispositional property in a mental state cannot solely depend on potential reference relations that could be established with this mental state. Potential relations cannot be sufficient for an actual property or state to occur. Indeed, Rosenthal rejects the dispositional HOT theory partly because he considers that a HOT must actually occur 109

Note that a higher-order content is considered to be part of the content of the phenomenally conscious perceptual state, despite the dispositional character of the corresponding HOT. This does not seem very plausible, since the phenomenal properties of a perceptual state are not of a dispositional kind.

83

in order for the lower-order thought to become conscious. Additionally, there is the question of why would some mental states, but not others (the unconscious ones), cause (or be able to cause) HOTs.110 Before causing a HOT, a mental state is not conscious; so, why would not all the mental states (or none) cause (or be able to cause) a HOT and thus become conscious? What is the difference between the (unconscious) mental states that cause a HOT and those that do not?111 An answer to this question solely based in functional considerations is not satisfactory; it would dismiss the distinction between Pconsciousness and A-consciousness. In sum, HOR theories do not show why the consciousness property would result (with physical or metaphysical necessity) from a representational relation. Probably, A-consciousness results from some mental states being represented by other mental states. But HOR theories pretend to account for the fact that there is P-consciousness, and for the what-it-is-like-ness of a P-conscious state. I do not see in HOR theories a complete explanation of this. It might seem that every P-conscious mental state is, or must be able to be, accompanied by another mental state that represents it. In fact, whenever there is something it is like for one to be in some mental state, there is a sense in which one knows what it is like to be in that state, and one can attend to this what-it-is-like-ness.112 However, the situation might be analogous to the following case (Fodor 1983): One opens the door of a fridge, and sees that the light in its inside is on. Then, one forms the hypothesis that the light is always on. To verify the hypothesis, one opens the door, and whenever one does so the hypothesis is verified. However, the light is always off when the door is closed, and only turns on when the door is opened. Probably, a mental state M can be such that there is something it is like to be in M, independently of the existence of any other mental state 110

Recall that in Rosenthal's theory the lower-order thought causes the HOT. I am indebted to M. Kistler for this remark. See Rowlands (2001). 112 The “argument of transparency” (Harman 1990) questions if one can indeed attend to the what-it-is-like-ness of a perceptual experience, or if one can just attend to the properties of the intentional object being perceived. I will address this argument in Chapter 6. 111

84

M’ that does (or could) represent M. Certainly, to have a thought about M it is required to metarepresent M, and to describe M it is required to conceptualise the contents of M. But I do not see why a creature cannot be conscious despite not having any thoughts about her experiences and not being able to describe them. It could be the case that there is something it is like to be in some first-order mental states for creatures that cannot be in higher-order ones.113 Another problem with HOR theories is that, even though they are not incompatible with the unity of consciousness, they do not give an account of it. Under the atomistic picture of the mind underlying HOR theories, there are many distinct mental states some of which have the m-consciousness property. But whenever there is something it is like to be S, there is one way it is like to be S. It is not the case that, as a result of a subject S being simultaneously in n distinct conscious mental states, there are n simultaneous distinct ways it is like to be S. The mconsciousness property is (and must be) somehow unified. The several lower-order mental states that are conscious under the HOR picture must somehow merge their m-consciousness property. The unity of consciousness is one of its most salient characteristics, despite the conceptual difficulties it involves. As Bayne says: “Although there is no denying that phenomenal unity is a puzzling feature of consciousness, I take the existence of some such relation to be beyond doubt.” (2010, p. 11). The unity of consciousness establishes a constraint for any theory of phenomenal consciousness, and has important epistemological consequences. I will address this topic in the following section. 3.2 The unity of consciousness Whenever there is something it is like to be an entity S we have what I will label “an experiencer”. An experiencer is not a person, the self, or 113

I take it that, e.g., a dog, has experiences, even though it does not have the cognitive resources to think about the fact that it has them or to think about the what-it-is-like-ness of having them. Indeed, I take it that most of the time we human beings do not metarepresent our experiences either. Moreover, it has been argued that when we do we have further experiences: There is a phenomenology associated with (at least some) second-order thoughts about first-order conscious mental states, which are not in turn metarepresented.

85

something of the sort. An experiencer is a physical system, defined as follows: (Expr) Any minimal physical system that instantiates one token of the c-consciousness property is an experiencer. Accordingly, to one instance of the c-consciousness property corresponds one experiencer. The experiencer is “minimal” in the following sense: (Expr2) If a physical constituent of an experiencer E is subtracted, the resulting system no longer instantiates the c-consciousness property. For example, since the limbs of a human organism are not necessary for consciousness, they are not parts of the corresponding experiencer. In humans, empirical research strongly suggests that the experiencer is some part of the nervous system while it performs some neuronal activities.114 In principle, a physical entity W could instantiate more than one token of the c-consciousness property. In that case, we would have many experiencers in W. Each part of a physical system that instantiates a token of the c-consciousness property or constitutes the supervenience base for its instantiation is an experiencer. It has been argued that this is the case, for example, with split-brain patients: In each hemisphere of the brain there would be one instance of the c-consciousness property.115 Note that each minimal physical system that instantiates a token of the c-consciousness property cannot instantiate more than this single token. And this is so even if it is an open question whether two experiencers can partially overlap. The reason is that I take the cconsciousness property to have a physical nature, and I associate it oneto-one with experiencers. Between a physical entity W that instantiates one token of the property, and any physical entity W’ that instantiates 114 115

See Baars et al. (eds.) (2003). For a discussion see Bayne (2008).

86

two tokens, there must be a physical difference. If the c-consciousness property is a physical property, the difference between W and W’ directly concerns physical facts about them. And if the c-consciousness property is a supervenient property, there must be a (physical) difference between the supervenience bases corresponding respectively to one token in W and to two tokens in W’. Now, recall that “phenomenal contents” are the members of the class of possibilities for the what-it-is-like-ness of being in a conscious mental state, that “phenomenal characters” are the members of the class of possibilities for the what-it-is-like-ness of being a conscious creature, and that:116 (PC) If a subject S is taken to be at time t in a single mental state M, the phenomenal character of the experience of S at time t is identical to the phenomenal content of M. If a subject S is taken to be at time t in several mental states Mi, the phenomenal character of the experience of S at time t is the mereological sum of the phenomenal contents of the Mi into a conjoint experiential character. Bayne calls the requirement of aggregating, into a single phenomenal character, the phenomenal contents of the different conscious mental states a given subject could be in at a given moment, “the unity thesis”: Unity Thesis: Necessarily, for any conscious subject of experience (S) and at any time (t), the simultaneous conscious states that S has at t will be subsumed by a single conscious state—the subject’s total conscious state. (2010, p. 16) Observe that the necessity involved in the unity thesis is captured in the association of a single entity (or “creature”) with a single phenomenal character. If a physical entity W were simultaneously in two experientially disjoint conscious mental states M1 and M2, we would have, given PC, two phenomenal characters and hence two experiencers. 116

See Chapter 1.

87

Therefore, to one experiencer corresponds (by definition) one token of the c-consciousness property and one phenomenal character. Note that the c-consciousness property abstracts away from (the class of) different phenomenal characters: There are many possibilities for the what-it-is-like-ness of being S.117 Bayne distinguishes between “subject unity”, “representational unity”, and “phenomenal unity”, and claims that the unity of consciousness subsumes these three types of unity. He says: Consider [...] what it’s like to hear a rumba playing on the stereo whilst seeing a bartender mix a mojito. These two experiences might be subject unified insofar as they are both yours. They might also be representationally unified, for one might hear the rumba as coming from behind the bartender. But over and above these unities is a deeper and more primitive unity: the fact that these two experiences possess a conjoint experiential character. [...] there is something it is like to hear the rumba while seeing the bartender work. Any description of one’s overall state of consciousness that omitted the fact that these experiences are had together as components, parts, or elements of a single conscious state would be incomplete. (2010, p. 11. Italics in the original) In this quote “phenomenal unity” corresponds to the “conjoint experiential character”. In my terms, this unity amounts to the fact that a single phenomenal character might involve (and usually does involve) simultaneously many qualia,118 associated with different sensations in different modalities, different emotions, etc. That the phenomenal character of a conscious experience complies with phenomenal unity is a particularly salient characteristic of consciousness. In sum, each experiencer instantiates one instance of the cconsciousness property and, at every moment, his experience has a 117

Analogously, the m-consciousness property abstracts away from (the class of) different phenomenal characters: There are many possibilities for the what-it-islike-ness of being in a mental state M in virtue of being in M. 118 Recall that “qualia” are the constituents of phenomenal contents and characters, which individuate them.

88

phenomenal character that complies with phenomenal unity. The experiencer is the physical system that instantiates the c-consciousness property, and the phenomenal character is the specific what-it-is-likeness of being that system. Note, firstly, that the phenomenal character can change over time: The what-it-is-like-ness of being S at t1 can (and usually does) differ from the what-it-is-like-ness of being S at t2. The consciousness property is a dynamical property, by contrast with properties like the charge or the mass of an electron (a fundamental entity) that have a fixed value. I will say that the consciousness property can follow different trajectories in “phenomenal space”;119 consciousness is a “stream” (James 1890). Secondly, note that the consciousness property can appear and disappear, i.e., may be discontinuous over time. A physical system can be an experiencer at some moments but not at others; for instance, it is presumed that whenever a person is sleeping without dreaming she is not conscious. But then, given that the consciousness property is a dynamical property, how should we individuate experiencers through time? I will briefly discuss two cases: The individuation of experiencers (1) when a physical system instantiates the c-consciousness property in a continuous manner; and (2) when a physical system exhibits discontinuous instantiations of the c-consciousness property. 1- I propose to reidentify the experiencer as one and the same over the lapse of time. When 1 kg of water is for a lapse of time, say, in the liquid state, we reidentify this liquid water as the same water over this lapse of time even though the relative positions of the molecules, the temperature, and so on, can change. No doubt, the metaphysical question concerning the persistence of a body as one and the same over time is a tricky and difficult one. If some properties of the body change, if some of its constituents are detached and new components are added, and so on, it is not obvious how to justify the idea that there is, nonetheless, a single body persisting over time. However, this is a general metaphysical problem, i.e., a problem not specific (in a physicalist framework) to the question of consciousness. 2- Take a lapse of time Δt and divide it in 3 consecutive segments 119

See Chapter 6.

89

Δt1, Δt2, and Δt3. Suppose that a system P instantiates the cconsciousness property during Δt1, during Δt3, but not during Δt2. Should we identify the experiencer before Δt2 with the experiencer after Δt2? The obvious argument for a positive answer is given by the fact that it is the same physical “substance” that instantiates the c-consciousness property during Δt1 and Δt3. Indeed, we ordinarily reidentify human subjects of experience on the base of an “organism” that persist over time. Consider a person who suffered a coma120 and is awake again. We identify the subject of experience before the coma with the subject of experience after the coma. We do not consider that after the coma there is a new subject, and this reidentification is certainly based on a reidentification of the organism. Or consider someone who suffers, e.g., from multiple personalities or amnesia, and therefore might not reidentify herself, every time she is awake, as the same subject. Also in these cases, we reidentify the subject of experience as one and the same through the whole life of the corresponding organism. However, it can be counter-argued that, despite our intuitions, the experiencer at Δt1 is not the same as the experiencer in Δt3. For instance, take some water that is solid during Δt1 and Δt3, but liquid during Δt2. Are we going to say that the ice before Δt2 is the same as the ice after Δt2? Certainly, it is the same “substance”, the same water, that constitutes both ices; but it is reasonable to consider that the ice before Δt2 is not the same as the ice after Δt2 given the different history of both ices and the physical differences between them (e.g., differences between the crystal structures). This is a quite complex matter, which leads to an aspect of the problem of personal identity. There are recalcitrant dualist intuitions that make the link between the subjective point of view of an experiencer and the experiencer (the physical system) to look contingent. Different paradoxes can be constructed which exploit the weakness or incoherence of our intuitions concerning this relation.121 But for the physicalist, psychological phenomena are natural phenomena that involve physical systems. Therefore, she can and should exploit the possibility of identifying and 120 121

I suppose that a comatose person is not conscious. See, e.g., Williams (1973) and Parfit (1986).

90

reidentifying physical systems in order to identify and reidentify subjects of experience. Hence, and even though I am aware of the complexity of this question, I propose to reidentify experiencers on the base of a reidentification of physical systems that instantiate in a continuous or discontinuous manner the c-consciousness property. Finally note that, given physicalism, the unity of consciousness must result from physical properties of the experiencer. If one adopts the atomistic picture of the mind, the unity of consciousness must result from a relation, of a physical nature, between the totality of conscious mental states of a subject, that somehow merges their m-consciousness properties into a single c-consciousness property. I am not (naively) claiming that the unity of consciousness requires some sort of physical contiguity between the physical constituents of mental states (or between their supervenience bases). The point is just that, somehow, the unity of consciousness must result from a physical property of experiencers. And similarly, if one adopts the holistic picture of the mind, the global mental state must be such that the relevant physical entities composing it, or on which it supervenes, are physically related as to give rise to the instantiation of the c-consciousness property. I will address the question of the specific (physical) nature of the consciousness property in subsequent chapters. The unity of consciousness, plus the fact that the consciousness property has an internal character, serves to defend physicalism from one of the most influential critiques: Kripke’s modal argument and “the explanatory gap”. This is the topic I will address next. 3.3 The explanatory gap The internal character of consciousness and its unity have epistemological consequences that, I think, are usually overlooked. In particular, I suspect that the appearance of the so-called “explanatory gap” results from taking an epistemological condition as an ontological one, as O. Flanagan says: “The gap between the subjective and the objective is an epistemic gap, not an ontological gap.”(1992, p.221). In this section I will develop this idea. I will elaborate on the internal character of consciousness to defend physicalism from the antiphysicalist conclusions that the explanatory gap motivates.

91

Kripke (1972) famously claimed that every identity statement between terms that are rigid designators is, if true, a necessary truth. Therefore, if psychophysical statements (PPSs) of the sort “pain is the firing of C-fibres” are true at all, they are necessarily true.122 However, PPSs seem to be contingent: E.g., we (may) have the intuition that there is a possible world where there is pain and yet no C-fibres firing. J. Levine (1983) argued that the appearance of contingency of PPSs is due to an “explanatory gap”: They don’t explain why, e.g., pain feels the way it does, or even why there should be any feeling associated with Cfibres firings at all. Schematically, the Kripke-Levine argument is as follows: (1) If PPSs are true, they are necessarily true. [premise] (2) PPSs do not seem to be necessarily true, since there is an explanatory gap. [premise] (3) PPSs seem to be false. [conclusion from 1 and 2] Note that this argument does not conclude that PPSs are false, just that they seem (intuitively) false. In order to have a proof that PPSs are false, it is necessary to accept, in particular, that: (i) Psychological terms like “pain” are rigid designators. (ii) Natural kind terms like “C-fibres firing” are rigid designators. (iii) Identity statements between rigid designators hold in every possible world. (iv) There is a metaphysically possible world where there is pain and yet no C-fibres firing. All of these points are more or less controversial. In a further chapter I will discuss the validity of this kind of modal arguments, and in 122

Kripke (1972) argues that both “pain” and “firing of C-fibres” are rigid designators.

92

particular the acceptability of statements like iv.123 Here, I will rephrase the argument in the terms I have been using and see to what extent it threatens physicalism. On one hand, we have: (I) A mental state has the m-consciousness property if and only if there is something it is like to be in that state.124 There are many possibilities for the what-it-is-like-ness of being in a conscious mental state M, in virtue of being in M. I called the members of the class of these possibilities “phenomenal contents”, and I associated which each phenomenal content one instance of the mconsciousness property.125 Accordingly: (II) A mental state has the m-consciousness property if and only if it has some phenomenal content. On the other hand, physicalism entails that: (III) The m-consciousness property has a physical nature. From II and III, we then arrive to: (IV) A mental state has some phenomenal content if and only if it has a property (the m-consciousness property) of a physical nature. Given physicalism, there is no problem with a mental state having properties of a physical nature. The interesting and controversial feature in IV is that there is a bi-directional implication between having some phenomenal content and having a property of a physical nature. Clearly, this does not entail that phenomenal content is some property of a 123

See Chapter 5. Recall that: (Def) An entity (or “creature”) is P-conscious if and only if she is in some P-conscious mental state. A mental state is P-conscious if and only if there is something it is like to be in that mental state for the corresponding entity. 125 Recall that the consciousness property abstracts away from (the class of) different phenomenal contents. 124

93

physical nature. But the bi-conditional should be explained in physicalistic grounds, and one way to do so is by means of a bijective identification of having a given phenomenal content with the instantiation of some property of a physical nature. Hence, let us suppose that: (S1) Phenomenal contents are properties of a physical nature. This is the general form of a PPS. In the left we have ways it can be like to be a creature S, and in the right we have physical or supervenient properties (that can be instantiated in a creature S). Now, according to Kripke these statements seem to be false. Levine takes the reason to be that any theory that associates phenomenal contents with physical properties seems unable to explain why a given content is associated with a particular physical property and, more generally, why there are these associations at all. There is an “explanatory gap” in S1: (S2) We have the intuition that a phenomenal content cannot be identical to some property of a physical nature. If phenomenal contents are physical properties, on one hand we can explain that there is the bi-conditional relation stated in IV—there would be an identity between phenomenal contents and physical properties— but on the other hand there seems to be an explanatory gap. If phenomenal contents are not physical but supervenient properties, we will find the explanatory gap between these properties and the corresponding (physical) supervenience bases. And if phenomenal contents are not identical to certain properties of a physical nature, we are back to dualism. In short, physicalism is committed to the idea that phenomenal contents are properties (as everything else) of a physical nature. Now, we (may) have the intuition that physicalism suffers from an explanatory gap. Thereby, we (may) suspect that physicalism cannot be true. I will try to defend physicalism by weakening this intuition. Why we have the impression of an unbridgeable gap in PPSs? This

94

impression can arise from different sources. That PPSs cannot express metaphysical identities is, after all, an intuition. But, in particular, I think that: The explanatory gap seems unbridgeable, at least in part, because we take, explicitly or implicitly, the right-hand term of a PPS as referring to an ontologically objective property, and the left-hand term as referring to an ontologically subjective one (whatever this could mean). That this is often so can be illustrated with the following cases: - J. Searle (2002) explicitly states that consciousness is ontologically subjective. In terms of a PPS, C-fibre firing would be an ontologically objective physical phenomenon, while pain is an ontologically subjective one. This raises the question of how can something be real and caused by biological systems— as Searle claims conscious experience is—and yet be ontologically subjective? Indeed, it is not easy to give a positive account of “ontological subjectivity”. Defining it as the condition of being real, but only from a subjective point of view, is obscure. - D. Chalmers (1996) claims that there is a metaphysically possible world where every objective fact is the same as in the actual world and yet there is no (subjective) experience. In terms of a PPS, a world where there is (say) C-fibre firing but no pain. - N. Humphrey (2006) says that PPSs only establish correlations, which could not turn out to be identities. The reason is that the terms in both sides of the PPS do not share the same “dimension”. In terms of PPSs, I take this to mean that while C-fibre firings are objective physical phenomena, occurrences of pain are not.

95

I will argue that phenomenal contents can have a physical nature despite their subjective character. If they do, in both sides of a PPS we have objective physical properties and, therefore, the impression of a gap might, if not disappear, at least be undermined. The main point is that the subjectivity of conscious experience is due to an epistemological condition, which does not entail ontological subjectivity. This view is the converse of Searle’s view, for whom consciousness is ontologically subjective but epistemically objective; he says: [S]uppose we insist on giving an account of the world that is completely objective, not only in the epistemic sense that its claims are independently checkable, but in the ontological sense [...]. Once you adopt this strategy [...] it becomes impossible to describe consciousness, because it becomes literally impossible to acknowledge the subjectivity of consciousness. [...] such a reality [not merely epistemically objective but ontologically objective] has no place for consciousness, because it has no place for ontological subjectivity. (1992, pp. 99-100) Firstly, I will present what I take to be the underlying argument against the physical reality of the phenomenal contents of conscious mental states. This argument is based on the idea that for something to be real it must (at least in principle) be accessible from an intersubjective perspective, which is not the case for phenomenal contents. Secondly, I will show that this argument makes use of a criterion of reality that is not applicable in the case of subjective experience. Finally, I will develop some ideas about the unity of consciousness by addressing a plausible objection. 3.3.1 Subjectivity and reality According to J. Kim, “[i]t may well be that the [mind-body] problem is an inexorable consequence of the tension between the objective world of physical existence and the subjective world of experience.” (2005, p. 30). In fact, a tension arises with the attempt to reconcile the following statements:

96

(I) Every real entity and property has a physical nature. (II) Phenomenal contents are exclusively accessible from the subjective perspective. Statement I simply stands for physicalism, but II requires some clarification. It states the subjectivity of conscious experience, which follows from the definition of phenomenal consciousness. Recall that an entity S is conscious if and only if there is something it is like to be, intrinsically, that entity, and recall that the ways it can be like to be an entity constitute the class of phenomenal characters. Since only subject S can know what it is like to be her, only S can be said to have access to the phenomenal contents of her conscious mental state(s).126 Of course, if physicalism is right and phenomenal contents have a physical nature, there is a sense in which a subject W can have epistemic access to the phenomenal content of a conscious mental state of another subject S. To take the example of Kripke, by observing in the brain of S the phenomenon of C-fibres firing, W would be observing the pain of S. But note that the epistemic access W would have is of a different kind that the access that S (provided with introspection capacities) has. Indeed, this is the main point I will use in the subsequent argument: The asymmetry in the epistemic relation that S and W can establish with the conscious mental states of S, can explain the fact that the phenomenal contents of these states, as phenomenal contents, are exclusively accessible from the subjective perspective of S.127 The tension there is between statements I and II can be illustrated by the following line of reasoning: 126

Perry (2001) exploits the perspectival character of consciousness to defend what he calls “antecedent physicalism” against the most influential anti-physicalism arguments. Roughly, what Perry claims is that phenomenal contents are only subjectively knowable because they are a kind of “reflective contents”. 127 With a theory that provided third-person point of view criteria a subject W could attribute some phenomenal content to a mental state of another subject S. But the attribution by W of a phenomenal content to a mental state of S, does not amount to W experiencing the what-it-is-like-ness of being S. I will elaborate this point in Chapter 6.

97

(1) To have phenomenal contents is to have real properties. [supposition] (2) Given I, 1 implies that phenomenal contents have a physical nature. (3) Physical entities and properties have—by definition—an objective reality. [premise] (4) Every objective entity or property is—at least in principle— accessible from an intersubjective perspective. [supposition] (5) If phenomenal properties are real properties, they must be accessible from an intersubjective perspective. [conclusion from 1 to 4] (6) However, given II, phenomenal properties are exclusively accessible from a subjective perspective. How to resolve the tension illustrated by statements (5) and (6)? One may: i) Deny 1: Claim that phenomenal properties are not real properties (eliminativism). ii) Deny 4: Claim that some entities or properties may be (objectively) real and yet not be accessible from an intersubjective perspective. Prima facie ii might seem implausible. But since I rejected eliminativism,128 I will argue for this possibility. The main argument is that, concerning phenomenal contents, the entailment from the (epistemological) condition of being exclusively accessible from the subjective perspective to the (metaphysical) one of not being physical (or supervenient on physical items) is invalid. I will claim that phenomenal contents are only accessible from the subjective perspective in virtue of the exclusivity of the relation that each subject has to her own mental states. 128

See Chapter 1.

98

3.3.2 Intersubjective accessibility Whenever different subjects are involved in an epistemological relation with the same entity X, they have good reason to believe in the objective reality of this entity. It is unreasonable to deny the existence of, say, the moon, since every subject (under suitable conditions) can see it in the sky. But do we only have epistemic reasons to believe in the objective reality of some entity or property X when it is possible (in principle) to have intersubjective epistemic access to it?129 Certainly, if there is an entity or instantiated property X in a given place at a given time, it is possible (in principle) for different subjects to establish an epistemological relation to it. Accordingly, if a subject S claims he perceives an X, but no one else can perceive this putative something, there is good reason to suppose that the X does not exist (i.e., has no objective reality) and believe instead that S is hallucinating an X or misperceiving something (taking a Y for an X). Therefore, it seems right to state that we only have epistemic reasons to believe in the objective reality of some entity or property X when it is possible (in principle) to have intersubjective epistemic access to it. Let us then adopt the following “intersubjective accessibility” criterion of reality: (IACR) If a putative entity X is not intersubjectively accessible, there is no epistemic reason to believe in its objective reality. Now, are there any restrictions on the applicability of this criterion? For the IACR to be applied to a putative X, we need the different subjects to be in equivalent epistemic conditions to establish an epistemic relation with X. This seems to be an uncontroversial requirement, a restriction, on the rational applicability of the criterion. In fact, it makes no sense to demand intersubjective access to an X if it is not possible for an arbitrary

129

Scientific realists and antirealists disagree about the existence of unobservables. However, the problem posed by unobservables in the realism/antirealism debate is orthogonal to the present one. There, unobservability is not related to the exclusivity of a subjective access.

99

subject to be properly placed or equipped to access X.130 Now, if the putative X happens to be a property of a mental state of a subject S, there is clearly an asymmetry between the kinds of epistemic relations that S, on the one hand, and other subjects, on the other hand, may establish to X. Think about self-knowledge of the position of my limbs. I know it in a way that is exclusive to me. I do not have “to look” at my hand to know where it is, in contrast to the mechanism others subjects may use to find this out. Mental states being embodied, they stand in an exclusive kind of relation with the embodying system. Because of this epistemological asymmetry between the kinds of epistemic relations that a given subject S, on one hand, and other subjects, on the other hand, may establish to the mental states of S, the IACR should not be applied in the case of phenomenal properties of conscious mental states. It is not just the case that phenomenal properties happen to fail the IACR; phenomenal properties are outside the scope of the IACR. Accordingly, the IACR entailment from the epistemological condition of being exclusively accessible from the subjective perspective to the metaphysical one of not being objectively real is invalid when applied to phenomenal contents. To argue that phenomenal properties are not real, but fictions or illusions, a different criterion of reality is required. I will continue by discussing a possible reply by the eliminativist, based on the following idea: Self-knowledge of phenomenal content is not epistemically well grounded, and this is best explained by the claim that there are no such properties. More specifically, consider the following objection: Certainly, if conscious mental states had phenomenal contents, a subject S would occupy a particular and exclusive epistemic perspective towards these properties of her mental states. However, the exclusive subjective character of the epistemic 130

Note that even if a subject, in order to be able to establish an epistemological relation with an X, may require special equipment and skills, it is possible for her to establish this relation.

100

access to these putative properties should not preclude the possibility for each subject of observing, analysing and describing them in a satisfactory manner. In fact, our observation-based knowledge is always supported by subjective epistemic access. Each of us has a particular and exclusive perspective on the world, from which he relates to objective reality. Despite this, we can agree about the description we give of, say, a table, the moon or a brain. Now, psychology teaches us that no satisfactory descriptions of phenomenal contents can be given. Several well-known experiments show that we are usually wrong in our judgements about the phenomenal contents of our own conscious mental states.131 As E. Schwitzgebel (2008, p. 245) says “we are prone to gross error, even in favourable circumstances of extended reflection, about our own ongoing conscious experience, our current phenomenology. Even in this apparently privileged domain, our self-knowledge is faulty and untrustworthy. We are not simply fallible at the margins but broadly inept. Examples [...] include: Emotional experience (for example, is it entirely bodily; does joy have a common, distinctive phenomenological core?), peripheral vision (how broad and stable is the region of visual clarity?), and the phenomenology of thought (does it have a distinctive phenomenology, beyond just imagery and feelings?). Cartesian skeptical scenarios undermine knowledge of ongoing conscious experience as well as knowledge of the outside world. Infallible judgments about ongoing mental states are simply banal cases of self-fulfilment”. The impossibility of describing phenomenal contents in a consistent manner is best explained by the nonexistence of such properties.

131

The interpretation of the experiments used by eliminativists to support the claim that no satisfactory descriptions of phenomenal contents can be given is controversial. For the sake of the argument I will accept that in fact we are usually misled or laden by folk psychology concerning phenomenological descriptions.

101

Eliminativism takes the alleged impossibility of describing phenomenal contents in a consistent manner to be due to the nonexistence of such properties132 and not to very exceptional epistemological circumstances. My reply to this objection includes three points. First, psychological studies certainly show that the descriptions that one may give of the phenomenal contents of one’s conscious experiences are not to be uncritically trusted. But this is no argument against the metaphysical claim that phenomenal contents may be objectively real. The ontological objectivity of some entity does not depend on our ability to achieve agreement on a description for it. Second, the conceptual and methodological difficulties (perhaps insurmountable) one faces when trying to describe (putatively real) phenomenal contents can be accounted for precisely in virtue of the very particular epistemological relation that subjects have to their own conscious mental states. Moreover, if phenomenal contents happen to be nonconceptual contents133 there are good reasons to expect phenomenal knowledge not to be fully translatable in conceptual terms. Finally, and more importantly, the objection rests on the supposition that the relation between subjects and (the putative) phenomenal contents is analogous, or similar enough, to the relation between subjects and objects like chairs, the moon or brains. I believe this supposition is mistaken because—among other problems—it seems motivated by the homunculus illusion: The idea that there is the self on one side, and his mental states on the other, and that the former somehow “observes” the latter (with the “mind’s eye”?). I will conclude by developing this last point through a brief critique of observation models of self-knowledge of conscious mental states. The question of self-consciousness and self-knowledge is a difficult and important one. However, I will not discuss it deeply since this would surpass my present purpose. 3.4 The homunculus fallacy It is quite common to find in the literature expressions of the form “I have access to mental state x”—therefore x is conscious—and “I don’t 132 133

For instance, Dennett (1991) seems to argue in this direction. I will endorse this view in Chapter 6.

102

have access to mental state y”—therefore y is unconscious. This way of talking can mislead us into thinking that there is something “there” with this strange property of being only accessible to me. Moreover, one could reach the conclusion that phenomenal properties are ontologically subjective, whatever this could mean. This is the view I will criticise. For practical reasons I will, in what follows, adopt the atomistic view of the mind. The different claims I will make can be accommodated in the holistic view by replacing “mental state” with a component or part of a single comprehensive mental state. Think about the expression “I have access to some of my mental states”. What does this “I” amount to? If we put to one side all the mental states (conscious and unconscious) a person is in, what is supposed to remain in the other side as the “I” which does or does not have “access” to the mental states?134 Certainly, in a sense we can say that there is “access” to the phenomenal contents of our conscious mental states. By means of selfknowledge abilities, somehow, we can reflect on and describe our experiences. However, this should not mislead us into considering that there is a subject-object relation between the subject and his experiences. To be sure, there is no experience without an experiencer, and there is no experiencer without experiences; experiencer and experiences are not detachable. In G. Strawson’s (2009, p. 59) words: “There can’t be experience without a subject of experience simply because experience is necessarily experience for—for someone-or-something. Experience necessarily involves experiential ‘what it-is-like-ness,’ and experiential what-it-is-like-ness is necessarily what-it-is-like-ness for someone-orsomething”. By contrast, in an ordinary subject-object epistemological relation the relata are detachable.135 This is a substantial difference. 134

Dennett would say that, precisely, there is no “I”. There is no homunculus enjoying the “Cartesian Theatre” show. However, Dennett also talks as if there were a subject who would or not have access to some mental states or properties of mental states: “ […] you do not “have access to” the intrinsic qualities of your own experiences in any interesting sense, any more than outside observers do. You have access only to the relations between them that you can detect.” (2005, p. 81. Italics in the original). 135 Even when you are looking at yourself in a mirror, subject and object are detachable; you may not know that it is you that you are looking at. See Perry

103

The mind of a subject, at a given moment, can be identified with the totality of the mental states occurring at that moment.136 Using “mind” in this sense, we can then say that whenever a person is experiencing she is in some conscious mental states (as well as in some unconscious ones) that are a constitutive part of the corresponding mind. Likewise, whenever a person engages in self-knowledge activities (however they might proceed), since these are mental activities, they involve mental states that are also constitutive parts of her mind. Therefore, both self-knowledge activities and the phenomenal contents of conscious mental states belong to the same mind. In this sense, the relation between the experiencing subject and the experiences he might know about is an internal one. This is certainly a trivial point, but it has epistemic implications that may be overlooked. To say that the experiencer-experience relation is an internal one does not just mean that each person has a privileged first-person perspective towards her own conscious mind, like the one you would have towards an external object that you and only you (for whatever reason) could perceive. This is a metaphysical claim: The totality of mental processes and the totality of mental states (some of which can be self-known) belong to one and the same entity. Therefore, the subjective character of phenomenal contents can be explained by the fact that the experiences are constitutive parts of the mind. Someone’s experiences cannot be intersubjectively accessible in the same way that they are subjectively accessible: The subjective access is internal by contrast with the intersubjective one that is external. It can be replied that the mind is a composed system that can be divided in such a way that some part of it (some mental states) would have epistemological access to another part of it (other mental states). When I say “I have access to some of my mental states”, these mental states do not correspond to the whole mind. Some mental states could have access to other mental states, such that the “I” would be constituted by the former bunch of states. Accordingly we would have, after all, a 136

(1979). A mind can be considered as the complete history of someone's mental states, or as a (wide enough) time-slice of this history. In discussing conscious experience, I am using the second sense.

104

subject-object relation between some sort of epistemological subject and the phenomenal contents of the conscious mental states. There are three options when making the proposed division. First, suppose we put to one side all the conscious mental states. In that case, there remains no conscious “I” that would have “access” to some of his mental states. We would have an epistemological subject lacking any conscious mental states. This is quite problematic; for one thing, it is clear that epistemological activities—including self-knowledge— involve paradigmatic conscious mental states on the side of the subject. Second, suppose we separate only some conscious mental states, in order for the “I” to remain a conscious psychological entity. In this case, there would be conscious mental states towards which we have no access (since they belong to the “I”). This option also seems unacceptable. Even worse: We would have two conscious bunches of mental states belonging to a single mind; the unity of consciousness would be violated. Third, suppose that the “I” is constituted by mental states that are nonconscious (but can access other mental states that thereby become conscious), but that are potentially conscious since they can “switch roles” and quit the “I” to become themselves objects of conscious access.137 In that case the “I” is not going to be constituted, at any time, by conscious mental states. After “switching roles”, the former states constitutive of the “I” become conscious, but then the new “I” is not itself conscious. In what sense could the mental states that the “I” accesses be conscious, when the “I” itself is not? The “I” and the conscious subject would be dissociated. Recall that one of the most salient characteristics of consciousness is its unity, as indeed Kant claimed. We experience our conscious mind as a single subject having unified phenomenal experiences, even though we can be in several mental states at the same time. Every attempt to model self-knowledge of conscious mental states under a subject-object schema, where a conscious subject accesses some (conscious) mental states, seems committed to violate this unity. We plainly can reflect upon and describe our experiences, and the 137

I am indebted to M. Kistler for considering this possibility.

105

question of how we acquire self-knowledge of them is certainly difficult. Moreover, it is controversial whether self-knowledge, and in particular self-knowledge concerning phenomenal contents, is a genuine kind of knowledge and thus deserves this label.138 It is beyond my purposes here to present or defend some positive account of self-knowledge. I just want to say that, however self-knowledge of phenomenal contents may work, an observational model—with the “observation” turned “inward”—is implausible. Self-knowledge is reflexive in a strong sense: The object coincides with the subject. In short, talk of one having “access” to some of one’s mental states is misguided. Whatever the nature of the epistemological relation between a subject and his conscious mental states is, it is an internal one in the sense of obtaining within a single mind. The mind does not “access” in any ordinary sense of the term some of its mental states, for they are part of what it is. To think otherwise is to fall prey to the homunculus illusion. As Dretske says: “[...] although one has privileged information about the character of one’s own experiences, one does not look inward to get it. Knowledge about one’s own experiences is obtained not by experiencing one’s experience, but by simply having the experience one seeks knowledge of.” (1995, p.149. Italics in the original). Note that, to account for the subjective character of phenomenal properties, I am not simply relying on the distinction between two different epistemic perspectives: The first-person and third-person points of view.139 I based the argument in the difference there is between a relation K among a mind T and its constitutive mental states, and a relation K’ among a mind T and the mental states constitutive of another mind. I believe that the latter distinction has deeper epistemological consequences than the former, as my argument was intended to show. 3.5 Conclusion Firstly, I argued 138 139

that

higher-order

representation

theories

of

See Alter & Walter (eds.) (2009). McGinn (1989) claims that subjective experience has a first-person mode of epistemic access, while science always works from a third-person perspective. Therefore, consciousness is out of the scope of science.

106

consciousness (HOR) fall short as accounts of the existence of phenomenal consciousness. It might be the case that human beings are such that every P-conscious mental state is or can be represented by another mental state. But the occurrence or possibility of a higher-order mental state M’ representing a mental state M is not sufficient to account for the fact that there is something it is like to be in M. Secondly, I argued that any theory of consciousness (and in particular HOR theories) must account for its unity, which is an essential characteristic of it. I defined an “experiencer” as a physical system that instantiates one token of the c-consciousness property, and labelled “phenomenal character” the specific what-it-is-like-ness of being a given experiencer at a given moment. Since one experiencer is associated with one phenomenal character, consciousness has “subject unity”. And since a single phenomenal character (or phenomenal content) can involve many qualia, consciousness also has “phenomenal unity”. Finally, I discussed the “explanatory gap”. Kripke (1972) claimed that the statements that identify the property of having phenomenal content with some physical property do not seem necessarily true, and yet they should if indeed they are true. Levine (1983) claimed that the appearance of contingency is due to an explanatory gap between the terms of the identity. I suggested that the gap appears when we take the subjectivity of consciousness as an ontological condition, and not as an epistemological one. I argued that the exclusively subjective access there is to phenomenal contents can be explained by the very particular nature of the epistemological relation holding between a subject and his own mental states. Self-knowledge is reflexive in a strong sense: The object coincides with the subject. The property of having phenomenal content can be an objective property, despite the subjectivity of phenomenal experience.

107

CHAPTER 4: SUPERVENIENCE, EMERGENCE, AND ONTOLOGICAL NOVELTY 4.0 Introduction In the first chapter I defined (phenomenal) consciousness as follows: (Def) An entity (or “creature”) is P-conscious if and only if she is in some P-conscious mental state. A mental state is P-conscious if and only if there is something it is like to be in that mental state for the corresponding entity. Thereafter I argued for a physicalist metaphysics and for a realist stance towards (phenomenal) consciousness. I claimed that: (C1) There is a property X, which is a physical property or a supervenient property, such that for an entity S to be conscious is for S to instantiate X. When it is an entity (or “creature”) that is P-conscious, I say that it instantiates one token of the “c-consciousness property”; and when it is a mental state that is P-conscious, I say that it instantiates one token of the “m-consciousness property”. In the second chapter I claimed that: (C3) The c-consciousness property has an internal character. This means, firstly, that the condition of S’s being conscious is not reducible in terms of relations between S and other entities. Secondly, that any relation that might be constitutive of the c-consciousness property of S exclusively relates constituents of S. If this property is a physical one, it is exclusively instantiated in physical constituents of S; and if this property is a supervenient one, its supervenience base only involves physical constituents of S. As T. Nagel says, “[i]f mental processes are indeed physical processes, then there is something it is

108

like, intrinsically, to undergo certain physical processes.” (Nagel 1974, p. 445-446. Italics added). In the previous chapter I defined an “experiencer” as the physical system which instantiates one token of the c-consciousness property, and thereby complies with “the unity of consciousness”. In this chapter I will argue that the c-consciousness property is not a supervenient, but a physical property. Since I rejected panpsychism,140 I will claim that it is a property of complex physical systems—the experiencers—that emerges (in a particular sense) from their fundamental constituents. The emergent nature of consciousness will endow it with “ontological novelty” and with “original” causal powers. Recall that I avoided a commitment to a particular view of mental states. I only claimed that mental states have a physical nature: Either they are physical states or they are supervenient (on the physical) states. Therefore, to claim that the c-consciousness property is a physical property of the experiencer amounts, concerning the m-consciousness property, to claim that: (PME) If mental states are physical states, for a mental state M to instantiate the m-consciousness property is for it to instantiate a physical property. (SME) If mental states are supervenient states, for a mental state M to instantiate the m-consciousness property is for the supervenience base of M to instantiate a physical property. 4.1 Supervenience and ontology I will begin by recapitulating the metaphysical framework established so far. Physicalism says that: (PH) All the entities inhabiting the actual world, their properties, and all the facts and events involving them, have a physical nature. In PH the expression “physical nature” means: 140

See Chapter 1.

109

(PN) An item T has a physical nature if either it is a physical item, or it supervenes on a set of physical items. Recall that “an item” is any kind of entity, property, fact, event, or law ruling phenomena. A “physical item”, in particular, is any constituent (fundamental or not) of any physical system, whether one takes a timeslice of the system, or its history during a time interval. According to PN, every item belongs to one of two complementary ontological categories: The physical and the supervenient on the physical. Now, I defined the relation of supervenience between an item (the supervenient item) and a set of items (the supervenience base) as follows: (Sup) An item U supervenes on a set of items Pi if U is entailed by {Pi} with metaphysical necessity, and U ∉ {Pi}. Recall that Sup corresponds to a particular conception of “supervenience”; different authors understand different things by this term.141 I will reject the idea that the consciousness property (the cconsciousness property and thereby the m-consciousness property) is a supervenient property. The first reason I will advance is based on the fact that supervenient items are “ontologically derived” from the corresponding supervenience bases. To develop this idea I shall elaborate to a greater extent the concept of supervenience. The definition of supervenience appeals to the notion of metaphysical necessity, which is a modal notion that can be applied de dicto or de re. Frequently, modalities are modelled in terms of “possible worlds”. A metaphysically necessary statement, in particular, is said to be a statement that holds in any possible world. Accordingly, if it is metaphysically necessary that “fact A entails fact B”, then “fact A entails fact B” is true in any possible world. Similarly, if fact B follows from fact A with metaphysical necessity, this means that in any possible 141

About the different concepts of mind-body supervenience, see McLaughlin (1995).

110

world fact B follows from fact A. Note that here “a possible world” is, obviously, a metaphysically possible world, not a physically possible one. Roughly, the difference is that the physically possible world is constrained by the laws of nature that apply in the actual world, while a metaphysically possible world can have other laws (or no laws at all, if this makes sense).142 The idea of “a possible world” is much more problematic than it is usually thought to be. Despite being very useful to characterise “metaphysical necessity” and to produce semantic models of modal statements, it must be carefully used as a guide to draw metaphysical conclusions. Our intuitions about what is and what is not a metaphysically possible world should not be uncritically trusted.143 It is also common to equate “metaphysical necessity” with “logical necessity”.144 Accordingly, a metaphysically necessary statement is claimed to be one that cannot be denied without contradiction. So, if “fact B entails fact A” is metaphysically necessary, this means that the statement “it is not true that fact B entails fact A” contains a contradiction. Surely, we can say that there is a “logical entailment” or a “logical contradiction” between statements containing propositions related by logical or mathematical operators. For instance, “P and Q” logically entails “P or Q”, and “P and not-P” contains a logical contradiction. But it is not clear in what sense these notions could be applied to statements that do not establish logical relations between propositions. The relation there is between, e.g., a statement affirming the occurrence of a social item and a statement describing the occurrence of the supervenience base of this item, does not seem to be of a logical type. A statement affirming the event of scoring a goal in a football match can hardly be said to “logically” follow from a statement describing physical items, like the event of an object—the ball— 142

I will come back to the distinction between metaphysical and physical necessity in Chapter 5. 143 See Chapter 5. 144 Chalmers (2006) strongly relates metaphysical necessity with logical necessity: “[...] the metaphysically possible worlds are just the logically possible worlds (and [...] metaphysical possibility of statements is logical possibility with an a posteriori twist).” (p. 38. Italics in the original).

111

trespassing the goal’s line. Furthermore, it is not clear how could the notion of logical necessity be applied de re. In what sense can the existence of an entity, the instantiation of some property, or the occurrence of an event, “logically” follow from other items? In what sense the event of scoring a goal could “logically” follow from items like the event of the ball trespassing the goal’s line? Therefore, I think it is better not to equate logical necessity with metaphysical necessity. Metaphysical necessity is a relation that can be used de re and de dicto, while I take logical necessity to have a clear sense only when applied de dicto and for specific types of statements. Nevertheless, logical necessity can be related to metaphysical necessity in the following way: (LN) If a statement “U” representing an item U can be logically deduced from a set of statements “Pi” representing the set of items Pi, U follows from P with metaphysical necessity. Thesis LN is plausible, but involves many controversial or problematic points. In particular, to justify the inference from logical entailments to metaphysical entailments, it must be shown that logical relations between statements reflect metaphysical relations between the truthmakers of the statements. And note that the converse of LN is not true: For U to follow from {Pi} with metaphysical necessity it is not required for “U” to be logically deducible from “{Pi}”; it is not even required for U and Pi to be symbolically representable. Intuitively, to say that every supervenient item is entailed by physical items with metaphysical necessity means that, when every physical item is fixed, by that means, without any mediation by further items, every supervenient item is fixed as well. Consider the set {P} of all the physical items, and the set {S} of all the supervenient items, ever existing or obtaining in the world. The relation of supervenience between {S} and {P} is such that {P} uniquely determines {S}. Moreover, every element in {S} is uniquely determined by a subset of {P}, namely, its supervenience base. And the way this “determination” works does not appeal to any law of nature, or to any sort of process connecting {P} with {S}. It is in that sense that the entailment involved in the supervenience relation has metaphysical necessity.

112

Certainly, it is not easy to formulate the notion of “metaphysical necessity” in a straightforward manner. This might be a fundamental notion, in the sense of not being fully expressible in terms of more basic notions. Despite the difficulties, I hope that the idea of metaphysical necessity is fairly clear for present purposes. Here, the important remark is that the metaphysical necessity involved in the supervenience relation entails that the supervenient is “ontologically derived” from the supervenience base. The claim that U supervenes on a set of items Pi does not mean that there is a process, of any sort, which produces something ontologically new, namely U, out of {Pi}. The existence of a supervenient entity, or the instantiation of a supervenient property, does not require the existence of any metaphysical substance that is not already contained in the supervenience base. But does this mean that every supervenient item is ontologically reducible in an eliminative way in terms of physical items?145 A hardcore physicalist might claim that, indeed, the supervenient items are not real; only the physical are. He might argue that supervenient items are not objective items, but fictions, or abstract constructs, or “concepts”, or something ontologically mind-dependent. By contrast, a moderate physicalist might claim that supervenient items, despite being ontologically derived, are real items. He agrees that, certainly, a property like “biological fitness” is relative to a teleological conception of natural history, and an entity like “a nation” is relative to social organisations and institutions. However, he might argue that the fact that some species is well adapted, and the event of a war starting between two nations, are components of reality. He may point out that there are truth-makers for statements about supervenient facts and events, which inherit objectivity in virtue of the supervenience relation they hold with physical items. In any case, both the hardcore and the moderate physicalist agree that supervenient items are ontologically derived from physical items. Borrowing a metaphor by S. Kripke, we can say that “god”, having created and determined every physical item, did not need to do any further work to create and uniquely determine every supervenient item. 145

After an “eliminative” reduction of U in terms of the reduction base {Pi}, U is claimed to not exist; only {Pi} exists.

113

Given that supervenient items are ontologically derived from physical items, one might wonder what the point is of having ontologies of supervenient items. But it happens that these ontologies serve essential epistemological purposes that are out of the scope of physics. Not only the social sciences, but also the natural sciences, go beyond the scope of physics and essentially include in their taxonomies supervenient items. In fact, reality can be organised in different ways, and following different theories. Depending on the particular goals or interests at hand, one can use social, functional, chemical, biological, or other kinds of criteria. For instance, an object classifies as a “chair” following functional criteria, a substance classifies as an “oxidiser” following chemical criteria, and a human being classifies as a “bachelor” following social criteria. The reason why physics cannot accomplish the epistemological duties of other sciences is not just pragmatic, but theoretical: From the claim that every supervenient token is ontologically derived from physical items, it does not follow that supervenient types can be described in terms of physical types. Here are two examples to illustrate this: (1) Each virus is identified with a RNA (or DNA) molecule (plus a protein coat), and individuated on the base of the biochemical properties of the molecule. But note that no molecule is intrinsically, i.e., solely in virtue of its physical properties, a virus. Without living organisms that some molecules can infect, there are no viruses; something is a virus only relative to an appropriate environment. It is thus clear that the supervenience base of a virus goes beyond the RNA (or DNA) molecule, even though this base essentially includes the molecule. Viruses are not identified with the corresponding supervenience bases, and are not individuated following the boundaries of the corresponding supervenience bases. Similarly, biological entities like “hearts”, “cells”, and “animals”; biological properties like “alive”, “predator”, and “fit”; and biological events like “infection” or “death”; are also supervenient. (2) In economy there are entities like “money” or “banks”, and properties like “rich” or “bankrupt”. Clearly, the existence of these items does not force us to postulate the existence of entities of a nonphysical nature. But, at the same time, it is clear that something is not a given

114

amount of money just in virtue of being some physical entity with some physical properties, despite the fact that we identify, say, a printed piece of paper, with a 100 dollars bill. Only relative to social agents, social institutions, and so on, facts and events involving money can obtain.146 The supervenience base of social entities and properties is, no doubt, extremely complex. But, regarding physicalism, the important points are that no social item can exist or obtain without some physical items existing or obtaining, and that the physical entails the social with metaphysical necessity. The second reason I will advance to reject the idea that the consciousness property is a supervenient property, is based on the fact that supervenient (on the physical) items do not have “original” but “derived” causal powers: The causal powers of a supervenient item are just the aggregate of the causal powers of the physical entities constituting its supervenience base. Certainly, this lack of original causal powers is closely related to the fact that the supervenient is ontologically derived from the physical. One can meaningfully attribute causal powers to supervenient entities and properties,147 especially if one is a moderate physicalist. One can say that a virus caused the death of an organism, or that an economic crisis caused a company to go bankrupt. However, since every supervenient item is ontologically derived from the physical, every causal relation between supervenient items supposes a physical interaction (causal or not, but physical) between their supervenience bases. Indeed, the very definition of supervenience states that every fact involving a supervenient entity or property is metaphysically entailed by physical items. As a result, the fact that a supervenient entity or property has causal powers must be derived, without mediation by any law of nature or physical process, from the physical supervenience base. As an 146 147

See Searle (1995). Note that the causal powers of a given entity result from the causal powers of their properties. In fact, causality requires (amongst other conditions) the existence of nomological relations between events: For the event c in system C to cause the event e in system E, there must be a nomological relation between c and e. But nomological relations are given by the laws of nature, and laws of nature express relations between properties.

115

example, think about the supervenient property of being alive. We can say that a creature breathes because it is alive. However, the property of being alive does not cause anything in addition to whatever results from the metabolic activity of the creature and its interaction with the environment. In short: If a supervenient property had original causal powers that would mean that it has powers that its supervenience base does not have. But this does not make sense: The occurrence of the supervenience base metaphysically entails the instantiation of the supervenient property; thus, any causal powers of the property are metaphysically entailed by the supervenience base. Certainly, the question of what is for an entity or property to have causal powers depends on the underlying conception of a causal relation. Unfortunately, there is no general agreement for a definition of causality; this is a tricky and controversial issue. First, the verb “to cause” and the derived terms can be used in different senses in different contexts. An economic crisis does not “cause” a company to go bankrupt in the same sense that a virus “causes” the death of an organism or the movement of a magnet “causes” the induction of a current. Second, there is no agreement on a definition of causality even when restricted to physical causation, i.e., to causation between physical entities.148 Every available definition seems to exclude relations we would like to classify as causal, or conversely to include anomalous ones. Since a comprehensive discussion of this topic would take me too far from the present subject, I will adopt the criteria for causality advanced by M. Kistler (2006a). This view only applies to the domain of natural science, but in the discussion about mental causation it is causation in the physical sense that is at stake. So: (CR) An event c causes an event e if and only if: (i) There is a nomological relation between c and e, (ii) c temporally precedes e, and (iii) there is a transfer of energy from c to e. The different alternative conceptions of causality generally include or 148

Russell (1912a), like many others such as Sklar (1974), even argues that the concept of causality is not used, or at least not needed anymore, in the formulation of physical theories. See Price & Corry (eds.) (2007).

116

entail i and ii. Causal relations have a dynamic character, and require a regularity that is given by laws of nature. But it is clear that i and ii alone are not sufficient to characterise causality. For one thing, they do not rule out events that we take to have a common cause. The additional and characteristic element in conception CR is given by point iii: Two events are relied as cause and effect if and only if there is an amount of energy, of electric charge or of any other quantity that is conserved in virtue of a law of nature, which is transferred between these events. To avoid any circularity, one must take as “transfer” just the presence of a given amount of a conserved quantity in each one of the two events. (Kistler 2006a, p. 64. Translation from French revised by M. Kistler) This requirement iii can be used to show that supervenient entities and properties do not have original but derived causal powers. Energy is a physical quantity, and transfer of energy results from physical interactions between physical entities. Thus, if one says that a supervenient entity transfers or receives energy, strictly speaking one is referring to a physical process involving its supervenience base. In sum, the supervenient entities and properties are ontologically derived and have only derived causal powers. This follows quite directly from the definition of supervenience. In the next section I will argue that, consequently, the consciousness property cannot be a supervenient property. 4.2 Consciousness and supervenience   If consciousness were a supervenient property it would (i) be ontologically derived from its (physical) supervenience base and (ii) have only derived causal powers. This possibility is—I will argue—at odds with the realist stance I adopted towards consciousness. Suppose that consciousness were a supervenient property. Then, the physical entities in its supervenience base would not have the consciousness property. Neither an individual constituent of the base, nor a part of the base, nor the whole base, would instantiate the consciousness property. Otherwise, this property would be a physical

117

one and not a supervenient one. Therefore, if consciousness were a supervenient property, on one hand we would have supervenient entities that are such that there is something it is like to be them, and on the other hand we would have the corresponding bases of supervenience which include no physical body that is such that there is something it is like to be it. This does not make sense. In fact, supervenience does not have the resources to produce the consciousness property. From a physical system P such that there is nothing it is like to be P, the property of there being something it is like to be an entity S which supervenes on P cannot follow with metaphysical necessity. Moreover, recall that the supervenient items belong to taxonomies that are just alternative ways of organising the (physical) world. The shift from a physical description where there is a physical system P to an alternative description where there is an entity Q which supervenes on P, cannot produce the fact that there is something it is like to be Q even though there is nothing it is like to be P. The consciousness property cannot be a supervenient property. There is the question of what are the criteria to attribute consciousness to an entity and, concerning this, the shift from a physical taxonomy towards a supervenient one can make an important difference. The epistemological criteria to attribute consciousness could include supervenient items that, correspondingly, are only manifest under supervenient taxonomies. But concerning the truth of “there are entities such that there is something it is like to be one of them” the shift from a physical taxonomy to a supervenient one cannot make a difference. If this statement is true, it is independently of any description one could give of entities and properties. Consequently, and once again, the consciousness property must be a physical property. I am not saying that it is not acceptable to attribute consciousness to a supervenient entity. Indeed, we do attribute consciousness, for instance, to persons, and this makes perfect sense despite the fact that “persons” are social (and therefore supervenient) entities. What I am saying is that we cannot attribute the consciousness property to a supervenient entity S unless we also attribute it to a physical entity included in the supervenience base of S. The situation is analogous to the one concerning causal powers: It makes sense to attribute causal

118

powers to supervenient entities but, strictly speaking, it is the base of supervenience that has the powers. Just as supervenient entities have derived causal powers, we can say that conscious supervenient entities derive their consciousness property from their supervenience bases. In the case of persons, they derive the consciousness property from the part of their biological bodies that constitutes what I called “the experiencer”. The previous argument does suppose that, by contrast with supervenient properties, physical properties can be said to be instantiated in a system S irrespectively of any description we could give of S. But I do not think this is too controversial. Physical properties are supposed to be objective in this strong sense. There are many problems concerning the truth of accepted scientific theories, and their referring to the physical world.149 But, for a physicalist, the world is populated by physical entities which instantiate physical properties independently of the scientific representations we could have about them. Were our theories completely false, and would the theoretical terms lack reference, there would nonetheless be physical entities with physical properties populating the world. In sum, the metaphysical entailment involved in the supervenience relation is too weak to account for something like the instantiation of the consciousness property at the supervenient level and yet not at the physical level. Additionally, if consciousness were a supervenient property it would not have original causal powers. This conflicts with a realist stance about consciousness, especially if one adopts “the causal criterion of reality”. Kistler formulates it as follows: (CCR) [F]or an [object or property] to be real it is necessary and sufficient that it is capable to make a difference to causal interactions. (2002, p. 57) Suppose that consciousness is a supervenient property instantiated in an (supervenient) entity Q. Take {Pi} as the supervenience base of Q and, thereby, of its consciousness property. Then, (i) there is something it is like to be Q but there is nothing it is like to be (an element or subset of) 149

See, e.g., Van Fraassen (1980) and Churchland & Hooker (1985).

119

{Pi}, and (ii) the causal powers of Q are just the aggregate of the causal powers of {Pi}. In that case, the fact that there is something it is like to be Q makes no difference concerning the causal powers of Q. The consciousness property of Q would not make any contribution in any causal relation in addition to the contributions given by the elements of {Pi}. Following CCR, we can then conclude that the consciousness property is not a real property or that, if it is real, it is so only in a weak sense.150 Only {Pi}, which ex hypothesi does not include the consciousness property, is real. Of course, one can reject the CCR because, e.g., one takes it to be a methodological principle and not a metaphysical one.151 But, intuitively, it is not clear anyway in what sense consciousness would be a real property, a natural phenomenon, something which makes an important difference if instantiated, if this property does not make any difference in the causal history of the world. The realist about consciousness cannot prove that consciousness exists and has (original) causal powers. But, given that he believes consciousness to be real, he might believe as well that it is because consciousness has original causal powers that he has this belief.152 This is the possibility I shall argue for. 4.3 Microphysicalism I claimed that every entity that populates the world has a physical nature, and that some of these entities are such that there is something it is like to be them—they instantiate the consciousness property. I also denied that the consciousness property is a supervenient property, and thereby I arrived at the following claim: (C1’) There is a physical property X such that for an entity S to be (phenomenally) conscious is for S to instantiate X.

150

The hardcore physicalist would claim indeed that consciousness—if supervenient—is not a real property. The moderate physicalist would consider it to be real, but in the “weak” sense: It is real and has causal powers but in a derived manner. 151 For a defence of the metaphysical validity of the CCR see Kistler (2002). 152 Chalmers (1996) considers that a “zombie” can form the belief that there is something it is like to be him. I will criticise his view in Chapter 5.

120

Previously I rejected panpsychism, i.e., the idea that fundamental physical entities instantiate the consciousness property.153 Therefore, the conscious entities involved in the previous statement are meant to be complex physical entities: (C4) The consciousness property is a physical property of (some) complex physical systems. The question now is what is the relation between the properties of complex physical systems and the properties of fundamental physical entities, which I will label “high-level properties” and “fundamental properties” respectively. In fact, Chalmers’ zombie argument against C4, and to some extent Kim’s argument against mental causation, precisely hinge—I will argue—on this issue.154 They adopt or presuppose a widely spread metaphysical view of the physical world that I have called “microphysicalism”.155 To defend C4 and make it compatible with the possibility of mental properties having causal powers, microphysicalism must be questioned. In this section I will elaborate the metaphysical picture given by microphysicalism. It claims that: (m-PH) All the entities inhabiting the actual world, their properties, and all the facts and events involving them, have a microphysical nature. In m-PH the expression “microphysical nature” means: (m-PN) An item T has a microphysical nature if either it is a fundamental physical item, or it supervenes on a set of fundamental physical items.

153

See Chapter 1. See Chapter 5. 155 See Chapter 1. Coleman's (2006) notion of “smallism”, i.e., “the view that all facts are determined by the facts about the smallest things, those existing at the lowest 'level' of ontology” (p. 40), roughly corresponds to microphysicalism as long as “determined” is taken in a metaphysical sense. 154

121

The “fundamental physical items” are: (1) The fundamental entities and the fundamental properties they instantiate; (2) the fundamental laws of nature; and (3) boundary conditions. Concerning 1, I take an entity to be fundamental if it is simple/indivisible, and a property to be fundamental if it is constitutive of a fundamental entity. For instance, in contemporary physics the Standard Model of Particles proposes an ontology of fundamental entity types, which classifies them in two categories (fermions and bosons), and specifies their fundamental property types (mass, electric charge, and spin). Regarding 2, I take fundamental laws to be the ones that directly relate fundamental entities, in virtue of their fundamental properties. For instance, the Standard Model of Particles proposes that there are four “fundamental interactions” (electromagnetic interaction; weak interaction; strong interaction; and gravitational interaction) plus “symmetries” which have “associated conservation laws”. Point 3 is required because a physical system is not fully determined by the set of entities it contains and the laws of nature that apply to it. Some spatiotemporal conditions are required. Take, for instance, a classical world that is ruled by Newton’s dynamics and the forces of gravity and electromagnetism, and inhabited by some charged massive bodies. A complete description of a physical system in this world must include “boundary conditions”. At a time to, the dynamics of the system for t > to (where “t” is another time), properties like kinetic energy, and events like electromagnetic radiation, depend on the relative positions and velocities of the bodies at to. Statement m-PH looks appealing, especially if one prefers “desert landscapes”, and is easily taken as being in perfect agreement with contemporary natural science. However, I will argue against m-PH after briefly elaborating the metaphysical picture it proposes. Suppose we want to create an exhaustive inventory of the items that ever existed or obtained in the world. If m-PH is true, when we include all the fundamental items the inventory is complete. It might be epistemologically useful to include items that are not fundamental. But these would be redundant from the ontological point of view. As a metaphor, microphysicalism can be said to be a 2D metaphysics: The physical world is exclusively inhabited by the fundamental entities,

122

behaving in accordance to the fundamental laws. The only “real” level is the lowest level of the physical ontology. The view of m-PH is a metaphysical picture of reality. As PH, it provides a general physical supervenience base (GSB) for the objects of study of the social sciences and of the natural sciences other than physics. In this respect, it behaves exactly as PH. Viruses, money, the property of being alive, and the event of getting married, are ontologically derived from physical items, and do not have original causal powers. But m-PH is more austere than PH about the cardinal of the GSB. The GSB of m-PH is exclusively composed by fundamental items. High-level items (and in particular high-level physical items) are taken by m-PH to be ontologically derived and to inherit their causal powers from fundamental items. In what follows, given that I claimed that the consciousness property is a physical property, I will focus on the implications of m-PH for the realm of physical entities and properties. From m-PH it follows that every high-level item supervenes on fundamental items. Certainly, the relation of supervenience between physical (or microphysical) items and the objects of sciences other than physics (what I called “supervenient items”) is very intricate. But the relation of supervenience there is—according to m-PH—between highlevel physical items and the fundamental items is relatively simple. First, high-level physical items locally supervene156 on fundamental items. Secondly, the relata in the supervenience relation are both physical and thus are individuated following the same type of criteria (physical criteria). Accordingly, the individuation of a high-level physical item coincides with the individuation of its supervenience base. As a result, the “ontological derivability” of a high-level physical item from fundamental items is tantamount to an eliminative reduction of the former in terms of the latter. When I say that, for m-PH, high-level physical items are reducible in an eliminative way in terms of fundamental items, this is to be taken primarily in an ontological, and not in an epistemological, sense. The concept of reduction is used in the epistemological sense, for instance, when we talk of the reduction of a theory A in terms of another (more 156

See Kim (1993).

123

fundamental) theory B.157 In the ontological sense, an item U can be eliminatively reducible in terms of items Pi independently of the existence of a theory that reduces a description of U in terms of a description of {Pi}.158 The view m-PH is, primarily, an eliminative metaphysics: Every high-level physical item is taken to be eliminatively reducible in terms of fundamental items. As a consequence, every theory about high-level physical items is expected to be reducible in terms of theories about fundamental items. Now, since causality is a relation between events, and for m-PH every event is a fundamental event or is eliminatively reducible in terms of fundamental items, causal interactions between high-level entities, and “downward” causal relations (causal relations going from “highlevels” to the “fundamental level”) are also eliminatively reducible in terms of fundamental items. The adoption of m-PH, in addition with statement C4, leads to the following thesis about the nature of consciousness: (m-PH+C4) The consciousness property is a high-level (physical) property, which supervenes on fundamental items. Clearly, m-PH+C4 contradicts the conjunction of (i) the realist stance I adopted about consciousness with (ii) the rejection of panpsychism. Statement m-PH+C4 means that the consciousness property is eliminatively reducible in terms of fundamental items. Now, fundamental items only include fundamental entities, none of which, given ii, instantiates the consciousness property. Therefore, m-PH+C4 entails that the consciousness property is not a real property, in contradiction with i. Strictly speaking, for m-PH only the fundamental properties are real and, given ii, the consciousness property is not among them. 157

If high-level types can be mapped into the more fundamental level types the reduction is said to be “homogenous”; otherwise, it is said to be “heterogeneous”. See Nagel (1961). 158 How reduction of theories works, what types there are, if there are successful examples, if biology and chemistry can be reduced in terms of physics, etc., are primarily epistemological issues.

124

Since m-PH+C4 is the conjunction of m-PH and C4, at least one of these conjuncts is false. Contemporary physics is commonly taken as giving a strong support for m-PH. Nevertheless, I believe physics can be used to argue precisely in the opposite direction. In the following sections of this chapter I will try to do so, thereby supporting C4. 4.4 The levels of reality Physics is an empirical science and, therefore, its ontology is open to substantial revisions. But let us take it that reality is such that the general atomistic metaphysics underlying physics is true: Physical entities are either fundamental entities or complex entities composed by fundamental entities, and the laws of physics determine the way these entities interact with each other as a function of their properties. To be specific: - Fundamental entities are those characterised by not being composed by further entities; they can instantiate various fundamental properties, but are simple in the sense of not being divisible. - Fundamental properties are attributed to individual fundamental entities, and are constitutive of them. - Complex entities are those composed by fundamental entities, and can instantiate various high-level properties. - High-level properties are those attributed to complex entities as a whole. - Fundamental laws are the ones that directly relate fundamental entities, in virtue of their fundamental properties. In a fundamental law the only properties that are involved are fundamental ones, and possibly structural properties of spacetime. - High-level laws are the ones that relate a complex entity with other (complex or fundamental) entities, in virtue of at least one of its high-level properties.159 Note that some physical properties can both be attributed as high-level 159

Note that high-level laws include “vertical laws”, i.e., laws relating high-level properties with fundamental properties.

125

properties or as fundamental level properties; for instance, take electric charge. We attribute electric charge to an electron or a quark (fundamental entities), or to a capacitor (a complex system). Other properties, by contrast, are exclusively attributed as high-level ones. For instance, solidity, temperature, electrical conductivity, and transparency, are always attributed as high-level properties; it makes no sense to attribute them to a particle. Hereafter I will use the following abbreviations: “HP” stands for “high-level property”, “FP” for “fundamental property” and “FL” for “fundamental law”. In this section I will introduce and motivate, in an informal manner, an alternative to m-PH, namely, the metaphysics of “ephysicalism” (for “emergentist physicalism”). To begin, think about a claim like “there are no rigid bodies, since physics has shown that the volume occupied by supposedly rigid bodies is almost completely empty and, moreover, every body is made out of atoms that are in permanent motion”. I believe there is something deeply wrong with this claim. To be sure, contemporary physics explains rigidity by proposing the existence of fundamental entities, atoms, molecules, and so on, which do not (and cannot) instantiate this property. Clearly, several properties like solidity, liquidity, and temperature, are HPs, and only complex entities composed of many molecules can be said to instantiate them. But then, should we conclude that these HPs are not real physical properties, by contrast with the FPs? Does the acceptance of physical theories that account for the instantiation of these HPs by appealing to fundamental items involve a commitment to an eliminative reduction of them? Maybe not. The main motivation for e-physicalism is the intuition that even though there surely are tight relations between FPs and every HP, and even though we can better understand high-levels with models of the lower-levels, many HPs properties are real on the same footing that the FPs are considered to be. Diamonds are rigid, water is liquid, and the sun has a surface temperature of about 5778 K. These HPs, despite being entailed by fundamental items, seem to be real in some strong sense and to have original causal powers. They seem to be “ontologically novel”, i.e., not eliminatively reducible in terms of FP. Kistler argues in this direction, and claims that even macroscopic physical properties that can

126

be conceptualised as dispositional, like transparency and conductivity, are—in my terms—ontologically novel. He says: [M]acroscopic properties such as a vase’s property of being fragile or my current property of having the intention to write the word ‘disposition’ can be causally efficacious in bringing about their manifestations although they can also be conceived of as dispositional properties. [...] According to the major theories defended at present, such properties are either epiphenomenal and thus causally inert or efficacious only by being identical with microscopic properties, which also constitute their reduction base. [...] it is coherent and plausible to consider the dispositional macroscopic property itself as causally responsible for its manifestation [...] Its reduction is the object of an independent enterprise; however, the construction of such a reduction does not justify the identification of the reduced property with the reducing property.160 [...] the causal efficacy161 of macroproperties does not lead to an unacceptable overdetermination of their effects: the microproperties in the reduction base cause these effects only indirectly, by lawfully162 determining the existence of the macroscopic property. This provides a justification for following the intuition that my present act of typing the word ‘disposition’ has been caused by my decision to do so, this decision being a macroscopic mental property that is not identical with any microproperty of my brain. (2007, p. 138-139) E-physicalism proposes, metaphorically speaking, a 3D view of physical reality. In addition to the two dimensions of the fundamental level, there is a vertical dimension constituted by high-levels. This 3D physical 160

If “the construction of [...] a reduction does not justify the identification of the reduced property with the reducing property” the reduced property has what I call “ontological novelty”. 161 “Causal efficacy” corresponds to what I call “original causal powers”. 162 That there is a lawful determination of the macroproperties by the microproperties is central in e-physicalism, and is crucial to justify the idea that some HPs have ontological novelty.

127

world is inhabited both by fundamental entities (at the bottom) and by complex entities with ontologically novel properties. Between the levels there are, of course, close relations. But not every item of a high-level can be eliminatively reduced in terms of fundamental items. Philosophers known as the “British emergentists” proposed in the th 19 and early 20th centuries the existence of “emergent” items.163 They had a “layered” view of nature, and considered that each level is the object of a specific natural science. At the bottom was the level of fundamental physics, on top of it the level of chemistry, then the level of biology, and finally the level of psychology. Even though they acknowledged the existence of relations between levels, they thought that in each level there are properties or phenomena—the emergent ones—which, in some sense, are not derived from lower levels. Mill (1843) and Broad (1925) claimed that there are high-level causal interactions which are not a priori entailed by low-level processes; Alexander (1920) argued that there are high-level qualities which cannot be expressed in terms of fundamental items. These views received serious objections and, in times of logical positivism, were almost unanimously rejected. For critics, British emergentism was just motivated by the observation that some properties do not seem to be explainable by taking into consideration phenomena at lower levels. Indeed, emergentists never gave a satisfactory explanation for the instantiation of some “strongly emergent” properties, and emergence thus remained mysterious. Additionally, emergentism was considered to be refuted by scientific theories that appeared in the early 20th century and gained wide acceptance. These new theories showed the existence of nomological relations between levels, and explained high-level properties in terms of lower-level ones. For instance, the chemical properties of the hydrogen molecule H2 were shown, with quantum mechanics, to be derivable from the physical properties of the hydrogen atom, and the condition of “being alive” was related to metabolic phenomena explained by organic chemistry. Certainly, an alternative to microphysicalism must not only grant some HP with ontological novelty and original causal powers, but must 163

For a historical review see McLaughlin (1992).

128

acknowledge the existence of a tight relation between HP and FP. I will argue that the view I labelled “e-physicalism” has these virtues. 4.5 Emergence E-physicalism claims that some high-level items result from the fundamental level with nomological, and not with metaphysical, necessity. Accordingly, the metaphysics of e-physicalism can be characterised by the following statement: (e-PH) Some high-level (physical) items emerge from fundamental items. In e-PH, high-level items are said to emerge from fundamental items— which constitute the “emergence base”—in the following sense: (Emg) An item U emerges from a set of items Pi if U is entailed by {Pi} with nomological necessity. By “nomological necessity” I mean a necessity mediated by laws of nature. What is the ontological status of the laws of nature is a very controversial question. But, for the purpose at hand, let us just accept that the laws of nature, and specifically of physics, are whatever determines the structure of spacetime and rules every interaction amongst the physical entities inhabiting it.164 Nomological necessity is metaphysically contingent in the following sense: There are metaphysically possible worlds ruled by laws of nature other than the ones that hold in the actual world.165 This is why the search for the laws 164

Alternatively, one might say that the laws of nature rule how the (fundamental or complex) physical entities interact, and that they are absolute in the sense of being invariant with respect to spacetime translations. “Invariance” means, roughly, that the same laws apply at any point in time and at any location in space. But note that, if one considers spacetime to be itself determined by some laws of nature (as General Relativity claims), one falls into a circularity when holding that the laws are spacetime invariant. 165 According to Kistler (2005) individuals cannot be reidentified across possible worlds that do not share the same laws of nature. In this sense, the laws of nature of the actual world are metaphysically necessary. Were the laws different, the physical kinds would also be different. See Chapter 5.

129

of nature is an empirical enterprise. Now, the subset of metaphysically possible worlds ruled by the same laws that the actual world constitutes the set of “physically possible” worlds.166 Accordingly, the entailment from fundamental items to high-level items proposed by e-physicalism does not hold in every metaphysically possible world but only in the physically possible ones. Clearly, if e-PH is true then m-PH is false. Not every high-level item would supervene on fundamental items. Now, since C4 claims that the consciousness property is a HP, in what follows I will focus on the following corollary of e-PH: (e-PHc) Some high-level properties (synchronically) emerge from fundamental items. There are different views and definitions of emergence, which can substantially differ between them and from Emg. In particular, P. Humphreys (1997) proposes a diachronic view of emergence. He claims that there is a physical (nomological) process, which he represents by the “fusion operator”, which produces HPs out of FPs and FLs. By means of this process, the fundamental entities cease to exist as separate entities to produce new “fused” entities equipped with new original causal powers. Whether Humphreys is right is both an empirical and a metaphysical question. The fusion operator must be shown to correspond to a specific physical process; the properties produced by fusion must be shown to have original causal powers; and the claim that, after the fusion, the fundamental entities cease to exist as separate entities must be properly motivated. Humphreys makes a suggestive case for his view based on the phenomenon of quantum entanglement and its relation to the macroscopic phenomena of phase transitions. Similarly, R. Penrose (1994) supports the idea that there is a diachronic nomological relation between quantum properties and macroscopic properties. In fact, he argues for the possibility of there 166

In Chapter 5 I will develop the question of the differences between metaphysical and nomological necessity, and between metaphysically and physically possible worlds.

130

being fundamental167 noncomputable168 laws of physics, which would rule the physical process underlying the collapse of the wave function. Certainly, Penrose’s proposal is directed to a solution of the “measurement problem”169 of quantum mechanics, and not to the development of an emergentist metaphysics of the physical world. However, if Penrose is right and there are laws relating the quantum and the macroscopic level m-PH is false. The macroscopic properties would not supervene on FPs and FLs. There would be vertical laws, going from the fundamental level to the macroscopic level,170 which would account for the appearance of the HPs. Despite the possibility of there being diachronically emergent physical properties, I will not argue for it. Statement e-PHc proposes a synchronic view that I take to be less controversial, simpler, and yet to provide a metaphysical picture where the consciousness property can be ontologically novel and have original causal powers. In fact, e-PHc is not meant to be interpreted as committed to the existence of any physical process apart from the ones proposed by contemporary physics, and does not require to consider that fundamental entities cease to exist as separate entities at any point in time. But note that a synchronic view of emergence does not exclude the dynamics of the fundamental entities in the emergence base. Even if one takes a time slice of a physical system, the entities of the system have properties like instantaneous velocity. In the following section I will argue, on the basis of contemporary physics, in favour of e-PH. Certainly, physics is an empirical science and therefore any theory, in particular a contemporary one, can be false. 167

Here a law is said to be “fundamental” if it is not metaphysically entailed by other laws. 168 Computable functions are functions whose values can, in principle, be computed by a Turing machine, i.e., algorithmically. Every law of physics proposed by currently accepted theories is computable. By contrast, a noncomputable law is one that cannot be represented by a computable function. Some self-referring functions are classical examples of noncomputable functions. For instance “the halting problem function”, which is a function which receives as input a computer programme and decides whether the programme finishes running or continues to run forever, was proved by Turing (1936) to be noncomputable. 169 The measurement problem leads to the famous Schrödinger's cat paradox. 170 It is an open question at what level the “frontier” is between the quantum world and the classical world.

131

Accordingly, one cannot definitely prove m-PH to be false. Indeed, if contemporary theories are completely wrong (which is not very likely), m-PH and other views, including e-PH, are founded on a vacuum. But, it is also clear that the metaphysical picture given by m-PH is motivated by contemporary theories. Moreover, any support a contemporary theory could give to m-PH requires supposing that this theory is true to an important extent. Thus, it is perfectly legitimate to use contemporary physics to argue against m-PH. If we had a “complete” physical theory that would truthfully describe every physical item, we could give a definitive answer to the question of whether m-PH is true. But, as Humphreys says, “certain metaphysical questions cannot be answered (yet) because we do not know enough.” (1997, p. 122. Italics in the original). Indeed, it is probable that we will never know enough. 4.6 The case for emergence   From m-PH follows the following corollary: (m-PHc) Every HP is entailed by fundamental items with metaphysical necessity. If m-PHc is true the defender of microphysicalism will claim that, strictly speaking, HPs do not exist. He will say that only the FPs are real, since the HPs are ontologically derived from fundamental items. Moreover, he can use the CCR to support this claim since HPs, being supervenient, do not have original causal powers. Statement m-PHc contrasts with the proposal of e-physicalism, which has the following corollary: (e-PHc) Some HPs are entailed by fundamental items with nomological necessity. If e-PHc is true, there are some HPs that are not eliminatively reducible in terms of fundamental items. There would be emergent properties, characterised by being ontologically novel and having original causal powers.

132

It is important to note that if “fundamental items” only include FP, m-PHc is obviously false and e-PHc is obviously true. Consider the following claims: (m-PHc’) Every HP is entailed by FPs with metaphysical necessity. (e-PHc’) Every HP is entailed by FPs with nomological necessity. Statement m-PHc’ is false since it does not take into account any of the nomological relations between the fundamental entities. For instance, consider again171 one litter of water and its property of being liquid. The liquidity of water, clearly a HP, does not supervene on the mereological sum of approximately 3,35.1025 H2O molecules. There are laws of nature governing the interaction between the H2O molecules, according to the physical conditions, which somehow determine the liquidity of water. The liquidity of water, if it is a supervenient property, supervenes on a larger supervenience base: The aggregates of H2O molecules plus the nomological relations between them. If the fundamental items only include FP, the liquidity of water is an emergent property in the sense of e-PHc’. Indeed, despite the fact that e-PHc’ is obviously true, it can be used to argue that every HP has ontological novelty. Since HPs follow from the FPs with nomological, and not metaphysical, necessity, they cannot be reduced exclusively in terms of FPs. Moreover, HPs have causal powers that the FPs do not have. These causal powers are new because they result from the FPs plus the nomological relations between these FPs. For instance, the liquidity of water would be ontologically novel and have original causal powers. Mereological sums of H2O molecules, clearly, are not liquid, and do not have the causal power of driving a water-mill (as liquid water does). But m-PH should be taken as proposing m-PHc and not m-PHc’. In m-PHc, the supervenience base is composed by “fundamental items” which include not only the FP but also the nomological relations 171

I used this example in Chapter 1.

133

between the fundamental entities. Accordingly, to falsify microphysicalism and in particular m-PHc, it must be shown that there are HP which are not metaphysically entailed by a set of fundamental items. It must be shown that there are emergent properties in the sense of e-PHc, i.e., HPs that cannot be instantiated without the mediation of laws that are not fundamental. B. McLaughlin makes this point when he states what would be an emergent property, if there were any. He says: If P is a property of w, then P is emergent if and only if (1) P supervenes with nomological necessity, but not with logical necessity, on properties the parts of w have taken separately or in other combinations; and (2) some of the supervenience principles linking properties of the parts of w with w’s having P are fundamental laws. [...] A law L is a fundamental law if and only if it is not metaphysically necessitated by any other laws, even together with initial conditions.172 (1997, p. 93) In order to argue against the validity of m-PHc and for e-PHc, I will discuss two examples that show a relation between different levels that does not hold with metaphysical necessity. Obviously, m-PHc is not only refuted by showing a “metaphysical gap”, i.e., the absence of a metaphysical entailment, between the fundamental level and some highlevel. Any metaphysical gap between different levels will do. The first example against m-PHc is taken from Kistler (forthcoming). It shows that the relation between the micro-level of the molecules of a gas and the macroscopic property of temperature is nomologically mediated. The second example is given by the nonlocal correlations in EPR systems. It has two advantages: (1) It concerns the relation between the fundamental level and the “first” one of the high-levels; and (2) the existence of these correlations is strongly supported by the empirical evidence. The discussion to follow will suppose the validity of the following principle “NL”, which concerns the realm of physics: 172

Note that McLaughlin uses “fundamental law” in a different sense that I do. I defined a fundamental law as one that relates exclusively fundamental properties. The “emergence laws” I will introduce latter, are fundamental in McLaughlin's sense.

134

(NL) If a physical item U, represented by the statement “U”, is entailed with metaphysical necessity by a set of physical items Pi, represented by the set of statements “Pi”, the statement “U” can be logically deduced from the statements “Pi”. A clarification and justification of this principle would take a very long digression, with little chances of success. However, I think it is important to make it explicit when we use mathematical and logical criteria for drawing metaphysical conclusions. Indeed, I do not take NL to have general validity.173 I just suppose it to be valid in the realm of physical science. 4.6.1 The kinetic theory of gases The kinetic theory of gases is a paradigmatic case of reduction of a macroscopic property in terms of microscopic properties. The macroscopic property is the temperature of a gas, and the microscopic property is the kinetic energy that each molecule constitutive of the gas has. On one hand, there is a law in thermodynamics that describes the relation between the temperature, the volume, and the pressure of an “ideal gas”. This law was established experimentally, and showed to apply to most gases over a wide temperature range. It says: (TD) PV = n R T In TD, “P” represents the pressure of the gas, “V” the volume it occupies, “T” its temperature, “n” the number of moles, and “R” the “gas constant”. Then, TD means that the product of the volume times the pressure is proportional to the temperature. On the other hand, in statistical mechanics considerations about the motion of the molecules of a gas yield the following result: (SM) PV = (2/3) N K 173

Note that NL is the converse of the statement LN previously introduced.

135

In SM the variable K represents the average kinetic energy per molecule, and N the total number of molecules. Thus, SM means that the product of volume times pressure is proportional to the kinetic energy. Comparing TD with SM, it becomes clear that the temperature of a gas is proportional to its kinetic energy: n R T = (2/3) N K With kB = (n/N) R we obtain: (KT) K = (3/2) kB T

“kB” is the Boltzmann constant

Now, the question is how to interpret equation KT. Prima facie, it seems that the property T supervenes on the motion of the molecules of the gas. The property of a gas having a temperature would be identical to a mathematical combination of properties of the molecules, namely, the average kinetic energy per molecule (modulo a scalar). However, a careful analysis of the procedure leading to SM shows that the gas is supposed to comply with a macroscopic constraint. The standard expression for the kinetic energy of a simple174 body is: k = (½) m v2

“v” is speed of the body

This is a property that, no doubt, we can attribute to microscopic entities. But the kinetic energy K that is used to deduce SM is the average kinetic energy per molecule: K = (½) m 2

is the average speed of the molecules

This quantity is obtained by taking an average over two variables. First, results from taking an average over the speed of the molecules at a given point in time. This is not too problematic. At any moment, each molecule is supposed to have a definite speed and this 174

By a “simple” body I mean a body that is not rotating and has not internal vibrations. When a body is rotating or vibrating, the expression of k must also include the energy corresponding to these movements.

136

simplification does not change the mathematical results. But secondly, also results from taking an average over time. To construct this average, it is supposed (1) that the speed of each molecule remains constant through time and (2) that the average speed of the molecules remains constant through time. Certainly, these suppositions must be justified. The speed of a given molecule can fluctuate over time, for instance given (inelastic) collisions and then, in principle, the average speed of the molecules at a given point in time could change through time. Suppositions 1 and 2 must be supported by some physical principle. There must be some property or law that ensures these statistical regularities. The justification of 1 and 2 rests on the fact that the gas is taken to be in thermal equilibrium. This is a concept of thermodynamics, related to the conservation of energy in isolated systems: When there are no energy flows between the container enclosing the gas and the environment, the temperature inside the container remains constant in time and uniform in space. Now, this property cannot be derived from the properties of the individual molecules that compose the gas, taken separately or in mathematical combinations. Thermal equilibrium is a macroscopic property. And without supposing that the complex system has this property, that is, without supposing that there is a physical property that produces the uniformities 1 and 2, the deduction of SM from TD given by the kinetic theory of gases is not possible. In sum, the kinetic theory of gases does not show that a macroscopic property, i.e., temperature, is identical to mathematical combinations of microscopic properties. It shows that there is a nomological relation between the temperature of a gas and the kinetic energy of its molecules, given by equation KT. Possibly, a proof of KT can be obtained exclusively in terms of microscopic properties and laws. But we do not know. So far, every derivation of KT used the constraint of thermal equilibrium or something equivalent.175 Therefore, the kinetic theory of gases, contrary to appearances, does not give a definitive support to m-PH. Certainly, this 175

For a detailed discussion of the reduction of thermodynamics in terms of statistical mechanics see Kistler (forthcoming).

137

theory does not involve the fundamental level. But if m-PH is true the supervenience relation must hold between every level, in particular, between the level of the molecules and the level of the gas. The next case to be discussed involves the fundamental level. And while the kinetic theory of gases is a classical theory, the next case belongs to quantum mechanics. Certainly, the interpretation of quantum mechanics is very controversial and there are many different proposals. The orthodox one is the Copenhagen interpretation, but others like the many-worlds interpretation or D. Bohm’s interpretation are also influential. Since a discussion of the interpretation of quantum mechanics would take me too far from my present purpose, I will follow the simpler strategy of directly discussing the EPR case. I will present it with a minimal usage of mathematical equations, without quantum formalism, and without requiring training in quantum mechanics. 4.6.2 The EPR Paradox and Bell’s theorem In a famous article,176 A. Einstein, B. Podolsky and N. Rosen claimed that quantum mechanics is an incomplete and thus unacceptable theory. This “EPR argument” was based on (1) the principle of “locality” and (2) a realist requirement of “completeness”: (1) The principle of locality rejects the possibility of “spooky actions at a distance” (Mermin 1985) and is certainly intuitive: “[...] anything that happens at a given location has only local effects. Nothing that happens at A can have any effect on the state of affairs at B.” (Greenstein & Zajonc 1997, p. 112). Einstein (1948) said about locality: The following idea characterises the relative independence of objects far apart in space, A and B: external influence on A has no direct influence on B; this is known as the Principle of Local Action, which is used consistently only in field theory. If this axiom were to be completely abolished, the idea of the existence of quasienclosed systems, and thereby the postulation of laws which can be checked empirically in the accepted sense, would become impossible. 176

See Einstein, Podolsky & Rosen (1935).

138

Consider two disjunct closed volumes in physical space, A and B. Locality says that a change in the properties of the physical systems enclosed in A cannot determine any change in the properties of the physical systems enclosed in B, unless there is some signal going from A to B. By a “signal” I mean some waves, or some particle, or something of the sort. No doubt, locality is an intuitive principle. It is hard to imagine how there could be nonlocal relations, and this principle underlies the classical picture of the physical world. Indeed, one of the most important critiques to Newton’s theory of gravitation was the fact that it seemed to violate locality. It was thought that there must be some physical vehicle for the force of gravity, but in Newton’s theory there is no account of how the mass of one body interacts with the mass of another body through the space separating them and, moreover, gravitational interaction is taken to be instantaneous. (2) The requirement of completeness is framed in a realist view of scientific theories. It says that a theory, to be acceptable, needs not only to be empirically adequate but must also provide a true177 and complete model of the physical world. Every type of physical entity, property, and every law, must be truthfully represented by the theory. In the EPR paper this requirement is stated as follows: “[E]very element of physical reality must have a counterpart in the physical theory.” (1935, p. 777. Italics in the original). Now, the group EPR provides a “principle of physical reality”: “If, without in any way disturbing a system, we can predict with certainty [...] the value of a physical quantity, then there exists an element of physical reality corresponding to this quantity.” (1935, p. 777. Italics in the original). This principle gives, of course, sufficient but not necessary conditions for the attribution of reality. Einstein, Podolsky and Rosen imagined a potentially real experiment178 and discussed the solution given by quantum mechanics. Their “EPR system” used two particles and the properties of position 177

Truth is conceived as some kind of correspondence or isomorphism between models and physical systems. 178 By a “potentially real” experiment I mean one that can be realised if the technological means are available, since there are no theoretical reasons that prevent it.

139

and momentum, which are “complementary variables” and so are subject to Heisenberg’s uncertainty principle. Later, Bohm proposed a new but equivalent version of an EPR system, with two electrons and their “spin” property.179 Bohm’s version illustrates in a simpler and more intuitive manner the situation and, in particular, makes more explicit the requirement of locality. Furthermore, Bohm’s version can also be adjusted to use other particles and properties, like photons and “polarisation”, and can be generalised to include more than two particles.180 I will then discuss the EPR paradox in Bohm’s version, as is commonly done. The experimental setup is as follows:

There is a source of electrons, and two Stern-Gerlach magnets. These systems are spatially separated, and are not connected by any further system or related by any signal. The space between them is empty. Electrons are particles with spin S = ±½. This means that a measurement of the spin of an electron will always produce one of two possible values, namely +½ or -½, with equal probability. The situation is analogous to that of a flipping a fair coin: There are two possible outcomes, with the same probability of obtaining. The Stern-Gerlach magnets are devices that measure the spin of a particle. Each magnet has an orientation that determines the direction of measurement. When a particle of spin ½ enters the detector, it is deflected either in the direction of orientation, or in the opposite 179 180

See Bohm (1951). See Greenberger, Horne, Shimony & Zeilinger (1990).

140

direction. A row of the EPR experiments, divided in three consecutive intervals, proceeds as follows: - (Δt1) Both magnets are oriented in the same direction α. The source contains a system of two entangled electrons with total spin ΣS=0. - (Δt2) The source emits the electrons. Electron e1 travels towards magnet M1, and electron e2 travels towards magnet M2.181 - (Δt3) Electrons e1 and e2 arrive at magnets M1 and M2 respectively, and each one is deflected either in direction α, or in direction -α. Thereby, a measurement (in direction α) of the corresponding spins S1 and S2 is obtained. I will label “M1/e1” and “M2/e2” the events of M1 and M2 interacting with e1 and e2 respectively.182 Quantum mechanics predicts two possible outcomes for this EPR experiment: {S1 = +½ ; S2 = -½} or {S1 = -½ ; S2 = +½}. The combinations {S1 = +½ ; S2 = +½} and {S1 = -½ ; S2 = -½} never occur. For instance, {S1 = -½ ; S2 = +½} means that e1 was deflected in direction –α, i.e., in a direction opposite to the orientation of the magnets, while e2 was deflected in direction α, i.e., in the direction of orientation of the magnets. Quantum mechanics also predicts that each one of these two possible outcomes has the same probability of obtaining. The statistical distribution of the results is normal (like the one of flipping a fair coin). In principle, these predictions are not surprising. The initial system we have during Δt1 is composed by two electrons and has a total spin ΣS= 0. Then, it is expected to obtain in Δt3 results such that S1 + S2 = 0. The EPR paradox concerns the interpretation of the experiment. According to the Copenhagen interpretation of quantum mechanics, during Δt2 the electrons e1 and e2 do not have a definite 181

The propagation of the electrons takes some time. This is an important factor for the interpretation of EPR experiments. 182 In technical terms, these events correspond to the measurements that produce the “collapse of the wave function”.

141

value for their spins. Each electron is in a “superposition” of states: e1 is in a superposition of [S1 = +½ ; S1 = -½], with each of these sub-states having the same probability of obtaining; the same for e2. In other words, before the measurements M1/e1 and M2/e2 occur, the variables S1 and S2 do not have any definite value. Moreover, if no measurement is made S1 and S2 never take a definite value. But then—argues the EPR group—how is it possible that, in Δt3, whenever magnet M1 measures S1 = +½ magnet M2 always measures S2 = -½ and, conversely, whenever M1 measures S1 = -½ magnet M2 always measures S2 = +½? How can we explain this correlation? It can be maintained that during Δt2 the values for S1 and S2 are not determined. However, there is a physical relation between e1 and e2 which ensures that only the outcomes {S1 = +½ ; S2 = -½} and {S1 = -½ ; S2 = +½} occur. Somehow, at the moment of the measurements (during Δt3) there is a physical relation between M1/e1 on one side, and M2/e2 in the other side, which accounts for the correlation. The problem with the previous solution is that this physical relation would be nonlocal. First, M1/e1 and M2/e2 are spatially separated and, ex hypothesi, no physical system connects M1 with M2. Second, the measurements M1/e1 and M2/e2 can be made simultaneously.183 Therefore, no signal could have travelled from M1/e1 to M2/e2 or vice-versa. In fact, there is a principle of relativity theory that says that nothing can travel at a speed greater than the speed of light, let alone instantaneously. As an analogy for this violation of locality, consider the following situation: One subject flips a coin in Paris, and another subject flips a coin in London. Whenever the coin in Paris falls face-up the one in London falls facedown, and conversely. To explain the result, one would look for some signal going from Paris to London, or vice-versa, that ensures the correlation. But if there is no signal and, nevertheless, there is the correlation, locality is violated. There would be a “spooky action at a distance”. 183

There is no requirement for the measurements to be made one after the other and, in any case, the order of the measurements does not affect the prediction.

142

Since the solution that appeals to a physical relation between M1/e1 and M2/e2 violates locality, it must—according to the EPR group—be rejected. The outcome of M1/e1 depends exclusively on the interaction between M1 and e1, same for M2/e2, and these events only have local effects. Now, note that just by realising the measurement M1/e1 one can predict with absolute certainty the result of M2/e2 (and conversely). If one finds S1 = -½ one can predict the result S2 = +½, and if one finds S1 = +½ one can predict the result S2 = -½, and this without in any way disturbing e2. So, according to the principle of physical reality, S2 is a real property of e2 that takes a definite value (same for e1 and S1). But orthodox quantum mechanics, as we saw, does not attribute to e2 this real property. It says that, if no measurement is made, S2 does not take a definite value. Therefore, following the requirement of completeness, the EPR group concludes that quantum mechanics is not a complete theory. Einstein, Podolsky and Rosen were thinking of an alternative, intuitive, and simple interpretation of the experiment. The idea is that, during Δt2, electrons e1 and e2 have a definite value for their spins— they are not in superposed states. If the outcome of the experiment is {S1 = +½ ; S2 = -½} this shows that, before the measurement, e1 already had the (intrinsic) property of spin S1 = +½ and, similarly, e2 already had the (intrinsic) property of spin S2 = -½. Analogously for the outcome {S1 = -½ ; S2 = +½}. Accordingly, the measurements in Δt3 just revealed the value that the variables S1 and S2 already had when the electrons were travelling towards the detectors. Electrons have definite values for their spins, independently of the occurrence of any experiment that measures these values. Interpretations of the EPR experiment like the previous one are called “hidden variable theories”. They are characterised by denying the existence of nonlocal correlations between M1/e1 and M2/e2. Instead, they consider that e1 and e2 have intrinsic properties—corresponding to the “hidden variables”—that determine, separately, the outcomes M1/e1 and M2/e2, in a manner that accounts for the synchrony between these outcomes. M1/e1 and M2/e2 just reveal the definite value of real intrinsic properties; the synchrony between M1/e1 and M2/e2 just results

143

from the manifestation of preexisting definite values. N. Bohr replied to the EPR paper by elaborating the Copenhagen interpretation of quantum mechanics.184 He did not argue for the possibility of nonlocal correlations, but claimed that quantum theory was not to be interpreted in a realistic way. For him, this theory was just an algorithm to predict (macroscopic) experimental results, and not a representation of the microworld. Bohr defended an instrumentalist view of science and rejected the requirement of completeness: Theories are only required to be empirically adequate, and empirical adequacy is always tested at the macroscopic level. If quantum mechanics makes correct predictions for EPR systems, it is fulfilling its purpose. For many years it was considered that no experimental evidence could give support to one of the parts in dispute. The EPR argument was not meant to question the empirical adequacy of quantum mechanics. They agreed with its predictions; the discussion was purely conceptual. But, in the sixties, J. S. Bell published a seminal paper185 where he presented “Bell’s inequalities”. This theorem showed that there was a difference between the predictions of orthodox quantum theory and hidden variables theories, for generalised EPR systems. The possibility to empirically test if indeed there are “spooky actions at a distance” was open. If the principle of locality were shown to be violated, the argument against the completeness of quantum mechanics would lose its grounds. Bell considered a slightly different set-up for the EPR experiment:

184 185

See Bohr (1935). See Bell (1964).

144

In this generalised setup, the EPR system is the same as before: It consists of two electrons, coming from a common source (where ΣS = 0), and travelling in opposite directions towards two Stern-Gerlach magnets. But now, the magnets are not oriented in the same direction. Between the orientation of magnet 1 and magnet 2 there is an angle γ. Under these conditions, quantum mechanics provides the following prediction: (QM) < (σx . α) (σz . β) > = - cos γ The terms (σx . α) and (σz . β) represent the measurements M1/e1 and M2/e2, where the variables {σx ; σz} represent the spins of e1 and e2, respectively. Variables {σx ; σz} are not supposed to have a definite value before the measurement. Only after the interactions of the electrons with the magnets the spins take a definite value. The variables {α ; β} represent the orientations of magnets M1 and M2 respectively; they form an angle γ. All these variables {σx ; σz ; α ; β} are unitary vectors, and the dot between them represents the scalar product. Thus, (σx . α) and (σz . β) can take two possible values: +1 and -1. If we have (σx . α) = +1 this means that e1 was found to have its spin in direction α (the orientation of M1). If we have on the contrary (σx . α) = -1 this means that e1 was found to have its spin in the direction -α (opposite to the orientation of M1). Similarly for (σz . β), e2 and M2. Accordingly, for each row of the experiment there are four possible results: {(+1 ; +1) ; (+1 ; -1) ; (-1 ; +1) ; (-1 ; -1)}. For instance, the result (-1 ; +1) corresponds to (σx . α) = -1 and (σz . β) = +1. This means that e1 was deflected in the direction opposite to the orientation of M1, while e2 was deflected in the direction of orientation of M2. The product (σx . α) (σz . β), in turn, can only take two values: +1 and -1. If [(σx . α) (σz . β)] = +1, this means that both electrons were deflected either in the direction of orientation of the magnets, or in the opposite direction. If [(σx . α) (σz . β)] = -1, this means that while one of the electrons was deflected in the direction of orientation of the magnet, the other one was deflected in the opposite direction.

145

The expression < (σx . α) (σz . β) > represents the statistical distribution of the product between (σx . α) and (σz . β). Each one of the four possible combinations of the factors (σx . α) and (σz . β), namely {(+1 ; +1) ; (+1 ; -1) ; (-1 ; +1) ; (-1 ; -1)}, has an associated probability that depends on the angle γ (the angle between α and β). Let us see three examples. First, take γ = 0. This corresponds to the original EPR setup where the magnets are oriented in the same direction. Equation QM entails that < (σx . α) (σz . β) > = -1. This coincides with what we saw: Always, while one of the electrons is deflected in the direction of orientation of the magnet, the other one is deflected in the opposite direction. Second, take γ = π/2. Equation QM entails that < (σx . α) (σz . β) > = 0. This means that, statistically, there is no correlation between the results in M1/P1 and M2/P2. Sometimes we have [(σx . α) (σz . β)] = + 1, sometimes [(σx . α) (σz . β)] = -1, and both situations are equiprobable. Finally, take γ = π/4. Equation QM entails that < (σx . α) (σz . β) > ½ = 2 /2 ≈ 0,7. This means that it is more probable to obtain [(σx . α) (σz . β)] = +1 than [(σx . α) (σz . β)] = -1. Approximately 2 times out of three both electrons are deflected either in the direction of orientation of the magnets, or in the opposite direction. Bell’s theorem shows that no local hidden variables theory (LHT) can make the same prediction for the generalised EPR setup that is provided by quantum mechanics. In other words, no LHT can produce equation QM. A LHT is a theory which: (1) Supposes locality: The product (σx . α) is exclusively determined by the interaction between e1 and M1. Likewise, the product (σz . β) is exclusively determined by the interaction between e2 and M2. (2) Takes the product (σx . α) as exclusively determined by a law that relates intrinsic properties of e1 with the orientation of a surrounding magnetic field. The quantity “σx” is taken to be a hidden variable. Analogously for (σz . β) and the quantity σz. Recall that a “hidden variable” is a term that represents an intrinsic

146

property that has a definite value. Hidden variables are taken to exist, with a definite value, independently of any measurement to determine this value. For instance, the property of e1 having S1 = -½ during Δt2 would correspond to a hidden variable. Bell’s theorem has the advantage of applying to any LHT. It does not make suppositions about which physical properties could correspond to the hidden variables. The quantities σx and σz are only required to correspond to intrinsic properties of the electrons. They can correspond to any fundamental physical property electrons could instantiate, and even to functions relating different fundamental properties. For instance, σx can be the spin of the electron e1 (like in the solution of the EPR group), or a hypothetical property H that e1 could happen to instantiate, or a combination of intrinsic properties of e1 including its spin, its charge and H (and analogously for σz). Bell’s theorem is the following inequality: (BT) | P (α , β) - P (α , β’) | + | P (α’ , β) - P (α’ , β’) | ≤ 2 In BT two possible orientations for each magnet are considered: α and α’ are orientations of M1, and β and β’ are orientations of M2. The function P represents the left hand of equation QM: P (α , β) = < (σx . α) (σz . β) > P (α’ , β) = < (σx . α’) (σz . β) > Etc. Accordingly, and since α . β = cos γ, equation QM can be rephrased as: (QM) P (α , β) = - α . β Bell proved that every LHT is constrained by equation BT. Any prediction of a LHT for the quantity in the left, for any values of {α ; β ; α’ ; β’}, is going to be minor or equal to 2. Hence, if the experimental implementations of EPR systems show results that violate Bell’s inequalities, there is empirical evidence against any LHT. Now, as expected, quantum mechanics does violate inequality BT

147

for certain values of {α ; β ; α’ ; β’}. Let us label {θ1 ; θ2 ; θ1’ ; θ2’} the angles corresponding to directions {α ; β ; α’ ; β’}. Now, take for example {θ1= 0 ; θ2 = 3π/8 ; θ1’ = -π/4 ; θ2’ = π/8}. According to QM, we have: P(α, P(α, P ( α' , P ( α' ,

β ) = 1/21/2 β' ) = -1/21/2 β ) = 1/21/2 β' ) = -1/21/2

Then, the left side of BT gives: | P (α , β) - P (α , β’) | + | P (α’ , β) - P (α’ , β’) | = (2) (2½) Clearly, quantum mechanics violates Bell’s inequality since (2) (2½) > 2 Several EPR experiments to test Bell’s inequalities were realised by A. Aspect and his colleagues.186 They reproduced a generalised EPR scenario, and registered the results for several values of all the variables. In particular, they varied the distances between each magnet and the source.187 First, they showed a systematic violation of Bell’s inequalities and, second, that the predictions of quantum mechanics obtain with great accuracy. Even though, “[s]ubtle and delicate objections have been raised to the design of and assumptions underlying the Aspect experiments [...] most workers in the field believe that these objections are not fatal.” (Greenstein & Zajonc 1997, p. 143). Note that what Bell’s theorem and “the Aspect experiments” strongly suggest is that theories that assume the existence of hidden variables and locality are false. Indeed, a nonlocal hidden variables theory can reproduce the same predictions of quantum mechanics, as is 186

See: Aspect, Grangier & Roger (1981); Aspect, Grangier & Roger (1982); Aspect, Delibard & Roger (1982). The first experiments to test Bell's inequalities were by Clauser et al. (1969) and Freedman & Clauser (1972). All of them showed a violation of Bell's inequality and gave support to quantum theory. 187 By changing the distances, they also changed the time each electron takes to arrive to the magnets. Thereby, the measurement events M1/e1 and M2/e2 were realised not only when they happen to be simultaneous.

148

the case with the quantum theory of Bohm (1951). Bohm elaborated a theory that provides the same predictions as quantum mechanics in every domain, but supposes the existence of hidden variables. His theory has the virtue of providing deterministic trajectories for quantum particles, instead of supposing the existence of superposed states. However, Bohm’s theory is nonlocal in a very deep sense. He postulates the existence of a “quantum potential”188 that is such that “the nonlocality contained in Bohm’s theory extends over arbitrarily large distances.” (Greenstein & Zajonc 1997, p. 148. Italics in the original). Moreover, his theory “goes beyond simple nonlocality, and calls upon us to see the world as an undivided whole.” (Greenstein & Zajonc 1997, p. 148). It is important to note that, if nonlocal correlations resulted from causal relations between M1/e1 and M2/e2, this would violate the principle of relativity that says that no body, signal, or causal influence can travel faster than light.189 However, the nonlocal correlations of quantum mechanics have been shown to be compatible with special relativity.190 It is not only because the correlations between M1/e1 and M2/e2 are (or can be) simultaneous that they do not classify as causal. There is no transmission of energy between M1/e1 and M2/e2; no “mark” (Salmon 1984) can be transmitted by means of nonlocal correlations. Another way to see this, is to note that no “information” can be transmitted from M1/e1 to M2/e2: From the results M1/e1 it is not possible to deduce the results M2/e2, except for the particular cases α = ± β (the original EPR setup).191 188

There is no empirical support, so far, for the existence of the quantum potential. Some physicists proposed the existence of “tachyons”, which are particles that travel at the speed of light, or faster without an upper limit. They use them to account for EPR experiments without violating locality. However, there is no evidence of the existence of tachyons and giving up relativity is a very high price to pay. 190 See, e.g., Shimony (1986). 191 It can be claimed that in the particular cases α = ± β there is information flow. However, if there were an information flow in these cases, it should be an information flow in every case. The prediction of quantum mechanics for the EPR system does not attribute any particular property to the cases where α = ± β. The function (QM) P (α , β) = - α . β is continuous and (iteratively) derivable for every value of α and β. 189

149

In sum, there is strong empirical evidence for the existence of nonlocal correlations in nature. According to Greenstein and Zajonc, “the lesson [...] is that it is not appropriate to think of these states as consisting of two separate particles with their own individual properties. Rather, the two particles are tangled together into a seamless unity.” (1997, p. 150). 4.6.3 Microphysicalism and nonlocality For Mermin “Einstein maintained that quantum metaphysics entails spooky actions at a distance; experiments have now shown that what bothered Einstein is not a debatable point but the observed behaviour of the real world.” (1985, p. 38). However, I do not want to claim that there surely are nonlocal relations in nature. First, there are many discussions about the interpretation of quantum mechanics. Indeed, Copenhagen’s interpretation and antirealist views of science prevent a realist reading of Aspect’s experiments. Second, this would suppose that there are crucial experiments, against the Duhem-Quine thesis.192 Nevertheless, m-PH is supposed to get support from contemporary physics, and any scientific support for a metaphysical thesis requires a realist stance towards scientific theories. If a realist reading of Bell’s theorem plus Aspect experiments supports Emg, the view m-PH is seriously undermined. I will argue that this is the case. Recall that m-PH and e-PH have the following corollaries concerning the relation between HPs and fundamental items: (m-PHc) Every HP is entailed by fundamental items with metaphysical necessity. (e-PHc) Some HP are entailed by fundamental items with nomological necessity First, note that nonlocal properties are, by definition, HPs. These are properties of a system, which cannot be instantiated by a fundamental constituent. Second, note that the nonlocal property of EPR systems is not 192

See Duhem (1906) and Quine (1951).

150

metaphysically entailed by fundamental items. Recall equation QM: (QM) < (σx . α) (σz . β) > = - cos γ A LHT is, as I said, a theory which: (1) Supposes locality: The product (σx . α) is exclusively determined by the interaction between e1 and M1. Analogously for the product (σz . β). (2) Takes the product (σx . α) as exclusively determined by a law that relates intrinsic properties of e1 with the orientation of a surrounding magnetic field. The quantity “σx” is taken to be a hidden variable. Analogously for (σz . β). According to Bell’s theorem, a LHT cannot make the prediction of quantum mechanics, given by QM. Quantum mechanics, by contrast with a LHT, is such that:193 (~1) The products (σx . α) and (σz . β) are correlated; there is a mutual dependence. In particular, the variables σx and σz are correlated or “entangled” (α and β, corresponding to the orientation of the magnets, are just parameters with fixed values). (~2) The variables σx and σz do not represent fundamental properties, i.e., intrinsic properties instantiated in the constituents e1 and e2 of the EPR system. 193

Recall that Bohm's interpretation of quantum mechanics makes the same predictions that orthodox quantum mechanics, and yet is a hidden variables theory. So, Bohm's theory is compatible with (2). But Bohm's theory not only is incompatible with (1), but also proposes a “quantum field” that produces nonlocal correlations at arbitrarily large distances, and integrates the physical systems inhabiting the universe into an undivided whole. An analysis of the implications for microphysicalism of Bohm's particular view would be a too long digression. But note that Bohm's metaphysics are plainly contrary to the spirit of microphysicalism; for him, the universe is not just the aggregate of fundamental entities causally interacting with each other.

151

Therefore, EPR systems have a property P, represented by the variables {σx ; σz}, such that (σx . α) is correlated with (σz . β) in accordance with QM. Now, P is not a conjunction of fundamental properties. Otherwise, σx and σz would be hidden variables. And neither is P the result of a fundamental law. Fundamental laws directly relate fundamental properties, but σx and σz are not fundamental properties. If P resulted from a fundamental law, this law would relate hidden variables. Consequently, P is not deducible from fundamental items, i.e., from fundamental properties and fundamental laws (plus the pertinent boundary conditions). This means, following NL,194 that: The nonlocal property P of EPR systems is not entailed from fundamental items with metaphysical necessity. In other words, P does not supervene on fundamental items. Third, note that even though the nonlocal property P of EPR systems is not metaphysically entailed by fundamental items, there is a nomological relation that does entail P. From the fundamental constituents of the EPR system, i.e., electrons e1 and e2, and the law that determines how a magnetic field influences their trajectory (plus the pertinent boundary conditions), the property P is nomologically entailed. In fact, equation QM plus Bell’s theorem, as we saw, entail the existence of the nonlocal property P. Fourth, note that nonlocal properties are to be taken as real, according to the CCR. Let “c” be the event corresponding to the emission of the two electrons by the source, and let “e” be the conjoint event {M1/e1 ; M2/e2} that shows nonlocal correlations. Firstly, there is a nomological relation between event c and event e, and c precedes e. After the emission of a pair of electrons, the interactions M1/e1 and M2/e2 occur following the pattern given by equation QM. Secondly, there is a transmission of energy from the source of electrons to the 194

Recall that: (NL) If a physical item U, represented by the statement “U”, is entailed with metaphysical necessity by a set of physical items Pi, represented by the set of statements “Pi”, the statement “U” can be logically deduced from the statements “Pi”.

152

Stern-Gerlach magnets. The detectors of the direction of deflection of the electrons flash in virtue of a transmission of energy from the electrons (by means of a collision).195 Now, the causal relation between c and e involves (given Bell’s theorem) a nonlocal property P, which is causally responsible for a property of e: The correlations there are between M1/e1 and M2/e2. Alternatively, we can say that because the EPR system has the nonlocal property P, event c causes event e. Hence, property P has causal powers: The power of producing the correlations between M1/e1 and M2/e2. If we accept the CCR, P is to be taken as a perfectly real property. In sum, Bell’s theorem plus Aspect’s experiments provide evidence against m-PH and support e-PH. Certainly, nonlocal effects have only been observed at the atomic level. But this is a virtue of this example against m-PH. There is a gap in between the quantum world and the classical world, evidenced by Schrödinger’s cat Paradox. There seem to be deep differences between the way macroscopic and atomic objects behave. Clearly, this gap can be exploited to claim that macroscopic items cannot be reduced in terms of microscopic items. But the EPR case shows that even at the atomic level there seem to be HP which are ontologically novel and have original causal powers. Quantum mechanics provides other theorems and phenomena that can be used to discuss the validity of m-PH. In particular there is the Kochen-Specker theorem,196 which is based on a second theorem developed by Bell against LHT.197 However, the Kochen-Specker theorem it not empirically testable. And other experiments, like the family of “double-slit” experiments, can also be interpreted (though in a less straightforward manner than EPR cases) as showing the existence of nonlocal correlations.198 4.7 Emergence laws In sum, both microphysicalism and e-physicalism take sets of fundamental items as the bases that entail HPs. But while 195

In general, every device that detects X requires to be perturbed by X, and this involves an energy transfer. 196 See Kochen & Specker (1967). 197 See Bell (1987). 198 See Greenstein & Zajonic (eds.) (1997).

153

microphysicalism takes all these entailments to have metaphysical necessity, e-physicalism considers some of them to have nomological necessity. Accordingly, for the former every HP is a supervenient property, while for the latter there are also emergent HPs that I will call “e-HPs”. Let us call “e-laws” (for “laws of emergence”) the laws that account for the instantiation of the e-HPs. Since e-HPs are not metaphysically entailed by sets of fundamental items, it is clear that elaws are not metaphysically entailed by sets of fundamental items either. In particular, they are not aggregates of FLs; e-Laws are not reducible in terms of fundamental items. Accordingly, supervenient HPs are metaphysically entailed by fundamental items; by contrast, e-HP are metaphysically entailed by fundamental items plus e-laws. In other words: e-HPs emerge from fundamental items, and supervene from a set of fundamental items plus at least one e-law. Let us go back to the cases I used to argue for e-physicalism and, in particular, to the following equations: (KT) K = (3/2) kB T (QM) < (σx . α) (σz . β) > = - cos γ Equation QM represents a property of the EPR system. Now, Bell’s theorem shows that QM cannot be exclusively deduced from fundamental items. QM essentially involves a HP (a nonlocal property). Accordingly, QM is an e-law: It relates a HP with a set of fundamental items (which includes in particular two electrons). Equation KT does not directly relate a HP—temperature—with fundamental items. K is an average of the kinetic energy per molecule, and molecules are not fundamental entities. But if we added the relation (of emergence or supervenience) between the level of the molecules— which includes in particular their dynamics—and the fundamental level, we would have a nomological relation between a HP (temperature) and fundamental items. E-laws entail irreducible properties of complex systems in a synchronic way. Therefore, the best way to conceive them is as

154

constraints on the behaviour of fundamental entities.199 For instance, KT establishes a constraint for the behaviour of ideal gases. Each molecule has at least three degrees of freedom,200 and there are fluctuations in its instantaneous velocity. However, when the system is isolated there is a statistical constraint on the overall behaviour of the molecules: Temperature remains constant. Similarly, QM imposes a constraint on the behaviour of the electrons when they interact with the Stern-Gerlach magnets. For each row of the EPR experiment many possibilities are open: Both directions of deflection are possible for each electron. However, there is a constraint for the overall behaviour of the system, which is manifest in the experimental results. Now, it could be thought that e-laws are always statistical constraints on the behaviour of the emergence base. However, the EPR case contradicts this idea: When both magnets are oriented in the same direction, the constraint is manifest in each row of the EPR experiment. It is only when they are not oriented in the same direction that the constraint is manifest only in the statistical characteristics of the results. In sum, e-laws (1) produce the instantiation of e-HPs on the corresponding emergent base and (2) operate as constraints on the behaviour of this emergence base. Note that between the fundamental level and a given e-HP there could be several e-laws. For example, consider a gas and its temperature. From the fundamental level to the molecular level there could be an elaw that establishes nonlocal correlations between the fundamental entities of the gas. Only because of these laws we would have gas molecules. Then, from the molecular level to the macroscopic level there could be an e-law that determines the thermodynamic state of the gas. Only because of this law we would have a gas in thermal equilibrium. 199

I take from Kistler (forthcoming) the idea of conceiving “vertical laws” as constraints. A “vertical law” is a law relating an emergent property with properties at a lower level. The constraint applies to the overall behaviour of the entities belonging to the lower level. 200 There are additional degrees of freedom associated with rotations of a molecule and internal vibrations. The number of potential degrees of freedom depends on the structure of the molecule.

155

Consequently, from the fundamental level to the temperature of the gas we would have the composition of (at least) two e-laws. Of course, I am not saying that this is exactly how things are. There could be more elaws (or none if m-PH is true) between the fundamental level and the temperature. These are empirical questions. The different levels of e-physicalism can be constructed as follows: -The fundamental items constitute the fundamental level. The second level is constituted by the e-HPs that only have fundamental items in the emergence base. The third level is constituted by the e-HPs that only have fundamental items and eHPs of the second level in the emergence base. The fourth level is constituted by the e-HPs that only have fundamental items and eHPs of the second and third level in the emergence base. Etc. -Between adjacent levels there are “simple” e-laws, and between nonadjacent levels there are “composed” e-laws, i.e., e-laws that result from the aggregate of simple ones. How many levels there are in nature, what is the form of the e-laws, what e-HP there are, and so on, are empirical questions. Now, as systems become more complex it becomes more difficult to show that a given HP is not reducible solely in terms of fundamental items. But at least for EPR systems, Bell’s theorem and Aspect’s experiments provide good reasons to believe in the existence of an e-law, an e-HP, and thus a second level of physical reality. 4.8 Conclusion I argued that consciousness does not supervene on physical items. It is a physical property of the conscious entity, which emerges from its fundamental constituents. Here, emergent properties are conceived as resulting with nomological necessity from the emergence base, as nonreducible in terms of the fundamental constituents of this base, and with causal powers of their own. This thesis, which corresponds to the “e-physicalism” view, is in

156

conflict with “microphysicalism”, i.e., with the idea that every property of a complex physical system supervenes on fundamental items. Therefore, I argued against microphysicalism and showed, through the elaboration of two examples—one in classical physics and one in quantum physics, the plausibility of e-physicalism. Thereby, I propose the following conjecture for the physical nature of consciousness: (C5) Consciousness is an emergent property of experiencers. I cannot prove that consciousness is an emergent property, but the previous discussion points in this direction. Therefore, I will assume that indeed the consciousness property is an emergent property and use this idea to confront Kim’s argument against the causal powers of mental properties, and some influential objections to physicalism, namely, Chalmers’ zombie argument and Jackson’s knowledge argument. In the next chapter I will elaborate the notion of an “experiencer”, criticise “the possible worlds strategy”, and discuss, in particular, Chalmers’ and Kim’s arguments.

157

CHAPTER 5: E-PHYSICALISM AND CONSCIOUSNESS 5.0 Introduction In the previous chapter I argued that the consciousness property is not a supervenient but a high level physical property (HP), i.e., a property that can be instantiated in complex physical systems but not in fundamental entities. To defend the possibility of the consciousness property being ontologically novel and having original causal powers I rejected microphysicalism and advanced e-physicalism. For the former, no HP has these virtues. But for the latter, there are emergent properties (e-HP) that do. Thereby, I suggested that consciousness is an emergent property. In this chapter, firstly, I will elaborate the notion of experiencers, i.e., of the minimal physical systems that instantiate the c-consciousness property.201 Secondly, I will discuss “the possible worlds strategy” and D. Chalmers’ zombie argument against physicalism. Finally, I will address the argument by J. Kim (2005) against the causal powers of mental states. The discussion will be framed into the e-physicalism view. 5.1 Experiencers So far, I argued that consciousness emerges from fundamental items, I claimed that it is an internal property, and I defined experiencers as the minimal physical systems that instantiate one token of the cconsciousness property. These claims converge on the following statement: (C6) The emergence base of consciousness is given by (1) the fundamental constituents of the experiencer, (2) the fundamental laws and (3) boundary conditions.

201

Recall that, when it is an entity (or “creature”) that is P-conscious, I say that it instantiates one token of the “c-consciousness property”; and when it is a mental state that is P-conscious, I say that it instantiates one token of the “mconsciousness property”.

158

Some bodies, given their physical constituents and structure, happen to be able to instantiate the consciousness property. However, only when they occupy some specific dynamical states, they actually become experiencers. For instance, take a physical body P that instantiates the cconsciousness property during a lapse of time Δt but not before nor after it. What makes of P an experiencer S during Δt by contrast with other times is the fact that only during Δt P is in some specific dynamical state. This dynamical state, which we may label “the conscious state”, encompasses several sub-states. In fact, not only can the c-consciousness property appear and disappear, but also the what-it-is-like-ness of being S changes (or might change) during Δt. I will say that the consciousness property can follow different trajectories in “phenomenal space”.202 As a very rough analogy, think about an engine that is in the “on state” during Δt and in the “off state” the rest of the time. During Δt, at any time slice the engine is in the on state, and occupies a sub-state that depends, among other variables, on how many revolutions per minute (rpm) it makes. We can then say that during Δt the engine follows some trajectory in the space of possible “on sub-states” (which starts and ends with a null value for the rpm variable). The following questions are empirical: (1) Which constituents and what structures are required for a body P to be able to instantiate the consciousness property; (2) what dynamical state P must occupy to instantiate it; and (3) what are the laws of nature that entail the emergence of consciousness. For the case of human beings (and some close species), some partial answers have been proposed by neuroscience. I will not discuss nor endorse any of them. Just note that three representative ones that I previously mentioned,203 namely “the hypothesis of the 40hz” of F. Crick (1994), the “dynamic core hypohesis” of G. Edelman and G. Tononi (2000), and “the conscious mental field” of B. Libet (2004), suggest that the experiencer is given by parts of the thalamocortical system when engaged in some specific dynamics. It is to be noted that scientists often talk as if subjective experience was something different from a neuronal structure engaged in 202 203

See Chapter 6. See Chapter 1.

159

some activity. But for e-physicalism, consciousness is not a correlate, or something caused, by brain activity (or the activity of any other body). It is a physical property instantiated in physical bodies while they are in some dynamical state. The only sense in which the consciousness property is “produced” is by means of the occurrence of the emergence base. Let us call the e-law (the law of emergence) that entails the instantiation of the consciousness property “the consciousness law”. Clearly, the consciousness law might be a composed e-law.204 There is no reason to think that there is a (simple) e-law that directly relates the fundamental level with the consciousness property. The emergence base of the consciousness property can (nomologically) entail various emergent properties that in turn (nomologically) entail the consciousness property. For instance, consider a human brain and recall that it is an extremely complex system. Between the fundamental level and the molecular level, then between the molecular level and the neuronal level, then between the neuronal level and the level of collective synchronised firing activity, and so on, there could be different (simple) e-laws. Obviously, and importantly, experiencers are not isolated systems. Like every physical system they are nomologically related with other systems. Which sub-state the c-consciousness property occupies depends on the interaction between the experiencer and the environment. For instance, the what-it-is-like-ness of seeing a ripe tomato is nomologically related with physical properties of the tomato. A red and a green tomato produce different stimuli and thus (in normal conditions) determine the state of the experiencer in different ways. And not only is the experiencer not isolated but, as happens with any physical system, the interaction with other systems is required for it to achieve some states and follow some dynamics. For instance, an engine requires a starting system to start off, fuel and oxygen to keep running, and varies its rpm depending on the activity of external systems like an accelerator. 5.2 The possible worlds strategy In this section I will discuss a type of philosophical argumentation that I labelled “the possible worlds strategy” (PWS). The PWS is particularly 204

See Chapter 4.

160

exemplified in Chalmers’ “zombie argument”, but the critique of it applies, in general, to arguments that draw metaphysical conclusions by means of the “conceivability” of possible worlds. First, I will briefly state the difference between metaphysical and physical possible worlds. Then, I will focus on the conceivability of counterfactual scenarios and their metaphysical import. 5.2.1 Physical and metaphysical possibility There is the idea that there are two main levels of necessity and possibility: The metaphysical and the physical. The constraints given by physical possibility are subsumed under the constraints due to metaphysical possibility, but not the other way around. I will discuss the difference between metaphysical and physical necessity/possibility, taking these modalities to apply not only de dicto but also de re. Take “a world” to be some set of items.205 For example, a world “NwW” could be constituted by two simple bodies in a Euclidean space, with the property of having a given mass, and with trajectories ruled by Newton’s dynamics and the law of gravity. Now, a world is a “metaphysically possible world” (mtphW) if it complies with some metaphysical constraints. Which are these constraints is a controversial question. But for one thing, if a set of statements is a representation of a mtphW (or of part of it), or determines a set of mtphWs, it seems it must respect the principles of logic.206 The actual world (W), obviously, is a metaphysically possible world. And so is, presumably, the NwW, since it seems that the statements that describe it are consistent. Now, a statement is true with metaphysical necessity if it is true in every mtphW. Amongst the mtphWs there is a subset that contains the actual world W and all the worlds that share with it the laws of nature. This is the subset of “physically possible worlds”. Now, a statement is true with physical necessity if it is true in every phW (but not in every mtphW). For instance, if the actual world W were a Newtonian world, then NwW would be a physically possible world (phW), and each body would 205

Recall that “an item” is any kind of entity, property, fact, event, or law ruling phenomena. 206 See Chapter 4.

161

follow, with physical necessity, a conic trajectory207 relative to the other body. In general, the mtphWs and the phWs are conceived on the base of W rather than from scratch. From W some items are proposed to be subtracted or added, and thereby a world W’ is obtained. If W’ complies with the metaphysical constraints it is metaphysically possible. And if W’ also shares with W the laws of nature it is physically possible. We can imagine different possible worlds, and we can use these hypothetical scenarios for epistemological purposes. But I will argue that it is not at all trivial to determine if an imagined world is or not metaphysically possible. In fact, it is not clear which are the metaphysical constraints that delimit metaphysical possibility. To appeal to logic is enlightening but not sufficient. First, logic applies to propositions;208 second, it is controversial whether logic is analytic (Quine 1951) and in what sense it could be objective (Heyting 1956). These are difficult problems that are often underestimated or overlooked. In sum, the actual world W belongs to the set of phWs, which in turn belong to the set of mtphWs, and a statement p can be evaluated at these two levels. At the metaphysical level, the truth-value of p is evaluated for worlds that comply with the metaphysical constraints. At the physical level, the truth-value of p is evaluated for worlds that also comply with the laws of nature of W. 5.2.2 The conceivability of possible worlds There are some types of statements that are claimed to be such that, if true, they are true with metaphysical necessity. Accordingly, if p is one of these statements, but there is a mtphW where it is false, we can conclude that p is false in every mtphW and, in particular, in W. This is the general structure of the “possible worlds strategy” (PWS). The successfulness of the PWS requires (i) the validity of the idea that there are statements which, if true, are metaphysically true, and (ii) the validity of the method to establish that “there is” a mtphW where a given statement is false. Concerning i, S. Kripke (1972) claimed that identity statements 207 208

“Conics” include ellipses, parabolas and hyperbolas. See Chapter 4.

162

between “rigid designators” are such that, if they are true, then they are true with metaphysical necessity. Amongst rigid designators, Kripke included proper names and also, with H. Putnam (1975), natural kind terms. Concerning ii, the usual strategy is to appeal to “conceivability”. If one can “conceive” a given world W’, it is claimed that “there is” this possible world W’. The purpose of the PWS is to establish the falsity of some identity statement in order to draw metaphysical conclusions. Concerning in particular identities of the form “the consciousness property = X”, where X stands for a natural kind term, the PWS proceeds as follows: (1) “The consciousness property” and “X” are rigid designators. [premise] (2) Identity statements between rigid designators, if true, are true in every metaphysically possible world. [premise] (3) If “the consciousness property = X” is true, it is true in every metaphysically possible world. [from 1 and 2] (4) If a world Wx is conceivable (without contradiction), then Wx is metaphysically possible. [premise] (5) A world W’ where “the consciousness property = X” does not hold is conceivable. [premise] (6) Since W’ is conceivable, W’ is metaphysically possible. [from 3 and 4] (7) There is metaphysically possible world where “The consciousness property = X” is false. [from 5 and 6] (8) “The consciousness property = X” is false. [from 3, and 7] I will accept the Kripke/Putnam semantics and, given that I take the consciousness property to be a physical property, I will thus accept 1 and 2. However, it is not obvious that the rigidity of some terms has the metaphysical import that the PWS supposes it has. In principle, the rigidity of a term concerns the normativity of language use. There are some terms that preserve their reference when used in counterfactual scenarios, because this is the way they work in normal situations. To find out if a term is rigid, we test if it should maintain its reference across imagined scenarios. But in the PWS rigidity is supposed to be

163

much more than just a question of normativity. Rigidity is supposed to tell us something about the actual world. The normativity associated with rigidity is supposed to reflect a kind of direct access to noumena. If rigidity does not have the metaphysical import that the PWS supposes it has, conclusion 8 does not follow. The most problematic steps are 4 and 5. What is for a world to be “conceivable” and, more important, what is the relation of conceivability with metaphysical necessity? Concerning the second question, I suppose that either it is presumed (i) that philosophers have the capacity to avoid contradictions when conceiving a possible world, or (ii) that something contradictory is not conceivable. In both cases, there are difficulties that should be addressed. Indeed, D. Stoljar’s "epistemic view" on the problem of consciousness consists on a critique of our capacities to imagine or “conceive” different possibilities and drive acceptable conclusions from them. For him, our ignorance prevents—in my terms—the PWS to be reliable. He says: [P]rogress on the philosophical issues [about phenomenal consciousness] will be achieved only if we keep ignorance in mind. On the one hand, the reasoning that is central to the philosophical questions starts from imagination or conceivability. We imagine possibilities and draw apparently objectionable conclusions from what we imagine. On the other hand, ignorance has an impact on imagination. What in our ignorance we believe ourselves to imagine and what we in fact imagine are distinct. In particular, the possibilities we believe ourselves to imagine do or would have objectionable consequences, but we do not imagine them in the relevant sense, and those possibilities we do genuinely imagine don’t imply anything objectionable. (2006, p. vi) As I previously said, the “conception” of a world W’ usually consists in imagining the actual world W being modified in some way. For instance, one subtracts from W some item. But how can we know if the obtained world W’ is metaphysically possible? And how can we determine if W’ is physically possible? As far as we know, W is populated by entities that interact with each other in many ways. The events concerning one

164

entity are related to events concerning other entities. So, if we subtract from W some item and pretend W’ to be a phW, presumably other items have to be modified as well. And if we subtract from W some item and do not want to modify any other item, presumably W’ is not going to be a phW. Now, we know that a world that shares with W the laws of nature can be metaphysically possible (since W is). But for a world that is not a phW it is not trivial to determine if it is, nonetheless, a mtphW. I will develop these points next. A preliminary remark: M. Kistler (2005) argues that it is not possible to reidentify physical entities of the actual world in metaphysically possible worlds that are not physically possible. The reason is that it is essential of a physical property that it complies with a given set of physical laws. Corresponding to different laws there are different properties and thereby different entities. Consequently, worlds that are not physically possible are of no utility to drive conclusions about items of W. If accepted, Kistler’s argument is definitive against the PWS. Nevertheless, for the sake of the PWS, I will assume that some entity of W is reidentifiable in W’ when W’ is not a phW. Suppose one takes out of W an item U and thereby pretends to obtain a physically possible world W’. If U is an entity, the history of all the entities that interacted with U must be modified as well. Otherwise, W’ would be violating the laws of nature. If U is a property, the behaviour of all the entities Ei that instantiated U, and of all the entities that interacted with the Ei, must also be modified. Otherwise, W’ would be violating the laws of nature. If U is an event, all the causal chains where U participated in must be modified. Otherwise, W’ would be violating the laws of nature. And so on. Clearly, it is not easy to determine the resultant world W’. I doubt one can “conceive” W’ with all the implications of the absence of U, unless one is a Laplacian demon. Now, suppose one takes out of W an item U, and nothing more, and acknowledges the fact that the resulting world W’ is not a phW. One only requires W’ to be metaphysically possible. I will argue that there is no guarantee that W’ is indeed metaphysically possible. In order for W’ to coincide with W in every item, except in U (and in the laws), the laws of nature in W’ must be such that every item in W’

165

is going to behave as if U was present. For instance, if U causes F in W, in W’ there would be F but without a cause. If U is caused by G in W, in W’ there would be G but without causing U. Now, it is not obvious that these are metaphysically possible scenarios. On one hand, for items not involving U the laws in W’ must produce exactly the same phenomena that the laws of W. But on the other hand, for U and all the items related with U the laws in W’ must produce phenomena quite different from the ones produced by the laws of W. So, to show that W’ is metaphysically possible, it must be shown that there is a set of laws of nature for W’ which meets these requirements. It can be replied that even though it is difficult to prove that there is a set of laws of nature such that W’ and W only differ by the absence of U (and by having different laws), we can be certain that this set exists because we can “conceive” W’ without contradiction. But how can we know that our “conception” is indeed coherent? Firstly, the laws of nature must form a harmonic set. They have to be “consistent”, in a sense similar to the one that applies to axiomatic systems. Now, the actual world W is very complex. The quantity of constraints that this set must comply with to reproduce W without U is huge. Secondly, the laws of nature must be complete: They must reproduce all the items of W except U. And recall that it is essential for a law of nature to be absolute (without exceptions) and to apply everywhere at every time. In short, from the “conception” of a physically or metaphysically possible world to the claim that then “there is” such a world there is a big step. If one wants to construct a world W’ out of W by subtracting (or adding) an item U, without changing the laws of nature, W’ is going to differ from W in items other that U. And if one wants W and W’ to differ exclusively by the absence of U, the laws of nature in W’ must be different from the laws of nature in W. But then, there is no guarantee that W’ respects the metaphysical constraints. 5.3 Metaphysical zombies Chalmers (1996) proposed an influential argument for the claim that consciousness is not a physical property, which leads to dualism or panpsychism. The idea is that a world identical to the actual world in all its physical aspects but where there is nothing it is like to be any entity is

166

“conceivable”. Following the PWS, Chalmers concludes that, in the actual world, consciousness is not a physical property. He says: 1. In our world, there are conscious experiences. 2. There is a logically possible world physically identical to ours, in which the positive facts about consciousness in our world do not hold. 3. Therefore, facts about consciousness are further facts about our world, over and above the physical facts. 4. So materialism is false. (1996, p. 123) The question is if “there is” a metaphysically possible world which is an exact replica of W but where the consciousness property is absent. For a physicalist, there is no such world. He takes the consciousness property to be a physical property and, therefore, an exact replica of W in every physical aspect is an exact replica tout court. A world where there is the (instantiated) consciousness property is physically different from a world where there is not this property. But, to be sure, for Chalmers this point begs the question. He proposes a “construction” of the zombie world that he thinks must be accepted by the physicalist as legitimate, and which leads to his proof against physicalism. He says that “high-level facts are entailed by all the microphysical facts” (1996, p. 71) where this “entailment” has logical necessity. The zombie argument, in my terms, can then be reconstructed as follows: (i) The consciousness property is a physical property. [supposition] (ii) Panpsychism is false, i.e., the consciousness property is not a FP. [supposition] (iii) Every physical property is a FP or supervenes on fundamental items. [premise] (iv) The consciousness property supervenes on fundamental items. [from i, ii and iii] (v) The consciousness property is ontologically derived from fundamental items. [from iv] (vi) But the consciousness property is not ontologically derived from fundamental items. [premise]

167

Since there is a contradiction between v and vi, one (or both) of the suppositions must be false: (vii) Either the consciousness property is not a physical property, or panpsychism is true. [conclusion] I have been arguing for both i and ii. In order to avoid the contradiction between v and vi, I must then reject one (or both) of the premises iii and vi. In a previous chapter209 I argued for premise vi. Supervenience does not produce ontologically novel properties, and for that reason I rejected the thesis that consciousness is a supervenient property. Thus, in point vi I agree with Chalmers’ view. But in the same chapter I argued against premise iii. This premise is a corollary of microphysicalism, which I rejected. Therefore, by rejecting microphysicalism I thereby rejected Chalmers’ zombie argument. However, I will continue with some additional remarks about the conceivability of the zombie world. For e-physicalism, if (1) we include the laws of nature of W in W’, then W’ cannot be a zombie world. And if (2) we do not include all the laws of nature of W in W’, then W’ is not a world physically identical to W.210 In fact, (1) If the laws of W are the same as the laws of W’, an exact physical replica of the fundamental level of W in W’ will include the consciousness property. In fact, for e-physicalism consciousness emerges from fundamental items in virtue of a law of nature that I labelled “the consciousness law”. In general, an exact replica of all the fundamental items of W, if this replica has the same laws that W, includes every item that supervenes on these fundamental items, but also every item that emerges from these fundamental items. (2) If the laws of W are not the same as the laws of W’, a replica of the fundamental level of W in W’ is not an exact physical replica of W. The zombie world W’ is not physically identical to the actual world 209 210

See Chapter 4. Searle (1997) argues against Chalmers that in order to "obtain" a zombie world it is necessary to change the laws of nature.

168

W, since the difference in the laws entails differences concerning the physical phenomena that obtain in W and W’ respectively. In particular, if W’ lacks the consciousness property then it lacks a physical property and, therefore, physically differs from W. That Chalmers’ endorses microphysicalism is clear in the following passage regarding the “construction” of the zombie world: In conceiving of a microphysically identical world, we conceive of a world in which the location of every last particle through-out space and time is the same. It follows that the world will have the same macroscopic structure as ours, and the same macroscopic dynamics. (1996, p. 73) Suppose that a world W’ is identical to W at the fundamental level. Now, whether W’ is also identical at high-levels depends on the laws of nature holding in it.211 If W’ does not include the e-laws, then its macroscopic structure is different from the macroscopic structure of W: In W’ there are no e-HPs, which are physical properties with causal powers.212 And if, conversely, W’ includes the e-laws, then the consciousness property is instantiated exactly as it is in W. But it could be replied: Even if there were e-laws, if W’ is identical to W at the fundamental level, then W’ is identical to W at every level. The reason is that W’ is not supposed to be a time-slice replica of the fundamental level of W, but a spatiotemporal replica. Now, is it conceivable to have a spatiotemporal replica at the fundamental level and yet differences at the higher-levels? Prima facie, an exact replica in W’ of the complete history of the 211

Chalmers does not seem to be clear about the role that fundamental laws play for the supervenience relation and for the occurrence of facts about microphysical entities. He says: “I am making the [...] claim that high-level facts are entailed by all the microphysical facts (perhaps along with microphysical laws).” (1996, p. 71. Italics added). 212 Recall that “e-HPs” stands for emergent high-level properties. See Chapter 4.

169

fundamental level of W might seem to be an exact replica at every higher-level. However, I will argue that, if there are emergence laws in W but not in W’, W’ can only be an exact replica of W by chance. It is not metaphysically necessary for W’ to be an exact replica of W also at higher-levels. Take two different subsequent times t1 and t2. At t1 and t2, the actual world W is in states W(1) and W(2) and another world W’ is in states W’(1) and W’(2). Ex hypothesi, between t1 and t2 the evolution of the fundamental level of W’ is an exact replica of the evolution of the fundamental level of W. Now, consider the relations between the states W(2) and W(1), and between the states W’(2) and W’(1). In W, the state W(2) is entailed from W(1) by means of the fundamental items at t1 plus the e-laws that synchronically “guide” the evolution of its fundamental entities from t1 to t2.213 Now, given that there are not e-laws in W’, the fact that W’ mirrors the evolution of W between t1 and t2 could only happen by chance. The evolution of the fundamental entities of W’ between t1 and t2 is as if they were following the same constraints that the ones of W follow, but there are not such constraints. So, while the entailment of W(2) from W(1) happens with nomological necessity,214 the entailment of W’(2) from W’(1) is neither nomologically nor metaphysically necessary. Now, the laws of nature that determine the temporal evolution of physical systems, whatever they are, are ingredients of physical reality. Even if the fundamental level of W’ mirrors the fundamental level of W at every time, W’ is not an exact physical replica of W. If W’ were an exact physical replica of W, despite the fact that the latter evolves with nomological necessity by contrast with the former, the following argument would be valid: (i) There is a metaphysically possible world W’, where every 213

Recall that e-laws are systemic properties that establish constraints for the behaviour of fundamental entities. 214 That there is a nomological relation between two physical states of a physical system at different moments does not mean that determinism is true. Nomological relations can involve stochastic laws that involve some degree of randomness.

170

physical entity follows the same trajectories as in the actual world W, and where ex hypothesi there are no laws of nature ruling these trajectories. [premise] (ii) Physical reality is nothing over and above physical entities behaving some way; it doesn’t include laws of nature as an ingredient. [premise] (iii) The worlds W and W’ are physically identical. [from i and ii] (iv) There are no laws of nature ruling the trajectories of the physical entities in W. [from i, ii, and iii] This argument is not compelling. It is true that we do not see laws of nature in addition to physical systems behaving some way, as premise ii says. But the hypothesis that there are no laws of nature (whatever they are) ruling the dynamics of physical systems is obviously very problematic. These dynamics, no doubt, exhibit multiple regularities, instead of pure randomness. The most reasonable position is to consider that laws of nature, somehow, are ingredients of physical reality, and therefore premise ii is false. Two worlds that coincide in every item but not in the laws of nature are not identical. In sum, if there are emergent properties, a world W’ which is an exact replica at the fundamental level of W is not an exact replica of W at every level. And this is so even if W’ is a historical replica of W. Premise 2 of Chalmers’ argument is not valid. It could be replied that a world W’ that differs from W only by the absence of the consciousness law, will be indistinguishable from W. W’ would be a zombie world: Every physical entity in W’ exhibits the same behaviour it has in W, but there is nothing it is like to be any entity. But recall that the consciousness property, being ontologically novel, has causal powers. Thereby, a world W’ without the consciousness law (and the consciousness property) will differ from W in a multitude of items, namely, in every item directly or indirectly related to the consciousness property. Now, Kim developed an influential argument against the causal powers of mental states that, in principle, could be applied to the consciousness property. This is the topic I will address next. The response to Kim’s argument will also be based on the rejection of

171

microphysicalism. 5.4 Kim’s “supervenience argument”     Kim (1993; 1998; 2005) elaborated different versions of “the supervenience argument” against the causal powers of mental properties. It applies to a conception of “the mental” in accordance with the following principles: Irreducibility. Mental properties are not reducible to, and are not identical with, physical properties. (2005, p. 34) Supervenience. Mental properties strongly supervene on physical/biological properties. That is, if any system s instantiates a mental property M at t, there necessarily exists a physical property P such that s instantiates P at t, and necessarily anything instantiating P at any time instantiates M at that time. (2005, p. 33) From these principles, together with “Exclusion”,215 “Closure”216 and “Edward’s dictum”217, the supervenience argument concludes that mental events do not have causal powers; only physical events do. If there are mental events, they are epiphenomenal: They have a cause218 but do not produce any effect. Clearly, the supervenience argument does not apply to the conception of the consciousness property proposed by e-physicalism. This property is not “mental” in the sense of Kim’s “irreducibility” and “supervenience”. This property is a physical property, and is mental in

215

"Exclusion" says: “No single event can have more than one sufficient cause occurring at any given time—unless it is a genuine case of causal overdetermination.” (2005, p. 42). 216 "Closure" says: “If a physical event has a cause that occurs at t, it has a physical cause that occurs at t.” (2005, p. 43). 217 "Edward's dictum" says: “There is a tension between “vertical” determination and “horizontal” causation. In fact, vertical determination excludes horizontal causation.” (2005, p. 36). 218 The cause of (the instantiation of) a mental property M is the cause of (the instantiation of) its physical supervenience base P (in Kim's sense of “supervenience”).

172

the following sense:219 (PME) If mental states are physical states, for a mental state M to instantiate the m-consciousness property is for it to instantiate a physical property. (SME) If mental states are supervenient states, for a mental state M to instantiate the m-consciousness property is for the supervenience base of M to instantiate a physical property. For e-physicalism, the principles of “irreducibility” and “supervenience” are replaced by the following ones: E-irreducibility. The consciousness property is not reducible to, and not identical with, fundamental physical properties. Emergence-C. The consciousness property emerges from fundamental physical items. That is, if any system s instantiates the consciousness property M at t, there necessarily exists a set of fundamental items Pi which are part of s at t, and necessarily anything including Pi at any time instantiates M at that time. But even though the supervenience argument does not apply to ephysicalism, it is based on a principle that is plainly incompatible with it: Edward’s dictum. There is a tension between “vertical” determination and “horizontal” causation. In fact, vertical determination excludes horizontal causation. (2005, p. 36) Not only e-physicalism allows for the coexistence of vertical determination and horizontal causation, but requires the existence of vertical determinations for the occurrence of some causal relations. If we accept Edward’s dictum we have to reject e-physicalism. 219

See Chapter 4.

173

In the supervenience argument (2005, pp. 39-44) Edward’s dictum is used in “Stage 1”. And even though this stage does not apply to ephysicalism, by replacing “supervenience” with “emergence” an analogous case that does apply can be formulated. It would proceed as follows: M and M* are instances (or instantiations) of the consciousness property, and {Pi} and {Pi*} are the corresponding emergence bases. Suppose that: (1) M causes M*. (2) For some set of fundamental items Pi*; M* has {Pi*} as its emergence base. Certainly, statements 1 and 2 are compatible with e-physicalism. Now, there seem to be a tension between them: (1) and (2) together give rise to a certain tension when we consider the question “Why is M* instantiated on this occasion? What is responsible for, and explains, the fact that M* occurs on this occasion?” For there are two seemingly exclusionary answers: (a) “Because M caused M* to instantiate on this occasion,” and (b) “Because {Pi*}, a[n] emergence base of M*, is instantiated on this occasion.” (2005, p. 39. With “supervenience” replaced by “emergence” and “P*” by “{Pi*}”) I will argue that e-physicalism avoids the conflict between 1 and 2. To the question “What is responsible for, and explains, the fact that M* occurs on this occasion?” the answer is: Both M and {Pi*} are responsible for the instantiation of M*. The property M synchronically emerges from its emergence base {Pi}; the property M* synchronically emerges from {Pi*}; and M and {Pi} together cause M* and {Pi*}. In fact, for e-physicalism the fundamental set {Pi} evolves into the fundamental set {Pi*} by means of emergent high-level properties that constrain its evolution. Without these constraints there is no nomological relation between the set {Pi} and the set {Pi*}. In other words, neither {Pi} nor M, at t1, cause {Pi*} or M*, at t2. The physical state at t1, with

174

all its levels, causes the physical state at t2, with all its levels. For e-physicalism, in general, causality can involve several levels in the cause and in the effect. A causal relation between two events is conceived as follows: Firstly, suppose there is a physical event involving a system P at t. P changes from P(t-δt), which is the state immediately before t, to P(t), which is the new state at t. Let us call this event δP(t). Analogously, let us call δQ(t+Δt) an event at t+Δt involving another physical system Q. Secondly, suppose that P and Q have (at least one) emergent properties. Then, the event δP(t) not only involves fundamental items. From P(t-δt) to P(t) the fundamental level of P evolves under the constraints that the relevant e-laws impose. Similarly for the event δQ(t+Δt). Finally, suppose that there is a nomological relation between the events δP(t) and δQ(t+Δt), and that some conserved physical quantity was transferred from P to Q. According to the definition of causality,220 the event δP(t) causes the event δQ(t+Δt). We thus have two systems, P and Q, involved in a causal relation that essentially includes emergent properties. The corresponding events δP(t) and δQ(t+Δt) obtain in virtue of e-laws that guide their production. And there is no conflict between “vertical” determination and “horizontal” causation: (i) At time t, the fundamental level of P determines (synchronically) the higher-levels of P. (ii) At time t+Δt, the fundamental level of Q determines (synchronically) the higher-levels of Q. (iii) An event involving the fundamental level of P and its emergent properties, i.e., δP(t), causes an event involving the fundamental level of Q and its emergent properties, i.e., δQ(t+Δt). Certainly, laws of nature relate individual properties belonging to specific levels. However, this does not mean that other properties at other levels are irrelevant. Suppose that: (i) P(t) and Q(t+Δt) have n and m levels respectively. (ii) P(t) and Q(t+Δt) have, in particular, emergent properties pi and qj respectively. 220

See Chapter 4.

175

(iii) pi belongs to level i and qj belongs to level j. (iv) There is a law of nature L(pi;qj) that relates pi and qj. Even though L establishes a direct nomological relation between pi and qj, through this relation L establishes a further relation between all the levels of P(t) and of Q(t+Δt). The reason is that the property pi emerges from the levels {1, 2, ..., i-1} and is constrained by the properties at levels {i+1, i+2, ..., m}. Analogously, qj emerges from the levels {1, 2, ..., j-1} and is constrained by the properties at levels {j+1, j+2, ..., n}. In sum, the picture of the consciousness property given by ephysicalism is not subject to Kim’s argument against the causal powers of mental properties and, furthermore, avoids the tension between vertical determination and horizontal causation that leads to Kim’s verdict. 5.5 Conclusion For e-physicalism the consciousness property is an emergent property: It is instantiated in complex physical systems, but not in fundamental ones, and has original causal powers. The emergence base of consciousness is given by (1) the fundamental constituents of the experiencer, (2) the fundamental laws and (3) boundary conditions. Firstly, I argued that an exact physical replica of the actual world cannot be a “zombie world”. If e-physicalism is true, Chalmers’ zombie argument is not valid. Furthermore, Chalmers’ argument requires an entailment from the “conceivability” of a world to the possibility of this world. But, in general, it is not trivial to determine whether a conceived world is metaphysically, or physically, possible. Secondly, I argued that the claim that the consciousness property has original causal powers is not threatened by Kim’s argument against the causal powers of mental properties. If e-physicalism is true, the consciousness property has original causal powers and there is no tension between vertical determination and horizontal causation.

177

CHAPTER 6: E-PHYSICALISM AND PHENOMENAL QUALITIES 6.0 Introduction The view of phenomenal consciousness I have been elaborating, which I labelled “e-physicalism”, is based on the following definition: (Def) An entity (or “creature”) is P-conscious if and only if she is in some P-conscious mental state. A mental state is P-conscious if and only if there is something it is like to be in that mental state for the corresponding entity. Recall that the “phenomenal content” of a (P-conscious) mental state M of a subject S is the what-it-is-like-ness of being S in virtue of S’s being in M; and that the “phenomenal character” of the experience of a (Pconscious) subject S is the what-it-is-like-ness of being S. Therefore, it follows from Def that: (C7) A mental state is conscious if and only if it has phenomenal content. I claimed that consciousness is a physical property. When necessary, I call it the “c-consciousness property” when it is instantiated by an entity (or creature), and I call it the “m-consciousness property” if it is instantiated by a mental state. Thus, given C7: (PMQ) If mental states are physical states, a mental state has phenomenal content if and only if it instantiates a physical property (the m-consciousness property). (SMQ) If mental states are supervenient states, a mental state has phenomenal content if and only if its supervenience base instantiates a physical property (the m-consciousness property).

178

As I previously said,221 the bi-conditional relation between having phenomenal content and having a physical property does not entail that phenomenal contents are physical properties. However, given physicalism, it is natural to suppose that: (S1) Phenomenal contents are physical properties. I took “qualia” to refer to the constituents of phenomenal contents and phenomenal characters, which individuate them. Indeed, I will identify phenomenal contents and characters with “structures”—in a sense to be specified—of qualia. 222 Hence, from S1 it follows that: (S1’) Qualia are physical properties. Statements S1 and S1’ are, no doubt, very controversial. I will be content if I can, at least, make them plausible. In this chapter I will, firstly, argue against the identification of phenomenal content with representational content. The argument will be based on the distinction between the representational content and the physical vehicle that conveys that content. The idea is that phenomenal content is a (physical) property of the vehicle, while representational content is external (or relational). Secondly, I will advance some speculations about the biological function of phenomenal contents, and about the physical structure of “phenomenal space”. The proposal is to conceive the consciousness property as one that can take different “values” in a multidimensional space. Each dimension of this space is a “modality” of qualia, and a phenomenal character is one position in this space. Finally, I will address F. Jackson’s “knowledge argument” from the e-physicalism point of view. I will claim that the truth of physicalism 221 222

See Chapter 3. Recall that: (PC) If a subject S is taken to be at time t in a single mental state M, the phenomenal character of the experience of S at time t is identical to the phenomenal content of M. If a subject S is taken to be at time t in several mental states Mi, the phenomenal character of the experience of S at time t is the mereological sum of the phenomenal contents of the Mi into a conjoint experiential character.

179

does not entail the possibility of capturing everything that can be known about the physical world in scientific theories. The discussion concerning “phenomenal contents”, “qualia”, and similar categories like “phenomenal qualities” and “phenomenal properties” is extensive and complicated. It involves many arguments, thought experiments, and empirical information, and there is wide disagreement, in many respects, among the authors. For example, some take qualia to be, by definition, properties of a nonphysical nature; others, consider that qualia can have a physical nature. Some take qualia to be intrinsic properties of subjective experience or sense data; others, to be properties of the external objects we perceive. Some take qualia to be properties of conscious mental states; others, consider that unconscious mental states can also have qualia. Some take exclusively perceptual states to involve qualia; others, consider that further types of mental states and processes also involve them.223 Given the complexity of the topic I will not advance a thoroughly elaborated theory of phenomenal contents. A comprehensive discussion would extend the scope of the present work. I will be content with sketching a picture that follows from, or at least is coherent with, the view of consciousness that I have been elaborating. My aim is to make explicit the commitments and consequences of e-physicalism for the question of the nature of phenomenal contents and qualia. If ephysicalism happens to entail unacceptable consequences in this area, it is to be revised or rejected. 6.1 Vehicles and content – an analogy Let us accept, following F. Dretske (1995), the following conception of a representational relation: The state “R” of a system S represents R if and only if (1) R and “R” are related by nomological dependence and (2) the function of S is to indicate R. From 1, the state “R” is said to convey “information” about R. In 2, the notion of “function” is a teleological one. If S is an artificial representational system, it was designed to 223

For instance, mental attitudes like desiring, epistemic activities, and mental acts like remembering, have been claimed to have a distinctive phenomenology. See, e.g., Horgan and Tienson (2002).

180

represent some property R. Thus, its function clearly is to represent R, and the representational content it conveys, if it performs its function correctly, is R. Later, S can also be used to represent another property R’. In this case, it acquires the new function of representing R’. If S is a natural representational system, it was not designed with an established purpose. So, it is not obvious what determines the specific function of S, i.e., what is S supposed to represent, and what is its representational content when it correctly performs its function.224 However, given that natural systems result from a natural selection processes, a plausible hypothesis is that the “proper function” (Millikan 1993) of S is determined by its phylogenetic history. The evolutionary history determines what S is supposed to represent and, accordingly, what its representational content is in “normal conditions”, i.e., in the conditions that obtained during the selection process. Now, like artificial systems, a natural system S can acquire the new function of representing R’ if (i) a state “R’ ” of S is nomologically related with R’ and (ii) S is systematically used to represent R’. In other words, S can acquire a new function through and in virtue of its ontogenetic history. Now, take a simple compass C. It is composed of a rigid support, holding a magnetised (with the appropriate polarity) arrow that can rotate horizontally. This system C represents the direction of North: First, the Earth's magnetic field, under normal conditions,225 causes the arrow to rotate until it is aligned with it; there is a covariance between the direction of the arrow and the location of the (magnetic) North Pole. Second, the function of C is to represent the direction of North, since it was designed with this purpose. The compass C is a “vehicle” of representation, and the “representational content” C conveys is what C represents and the way it represents it to be. This content can be characterised as “the direction of North is αi”, where “αi” is the direction the arrow points to.226 Note that, 224

See, e.g., Macdonald & Papineau (2006). The compass was designed to indicate the direction of the North Pole when placed under some conditions. These are the “normal conditions”. 226 Obviously, the representational content conveyed by the compass is not symbolic but analogous. The proposition “the direction of North is αi” expresses symbolically the analogous representational content. 225

181

obviously, this content is not internal: The North—a cardinal point—is external to C. However, this content has a perspectival component: The direction of North is relative to the location of C. Compass C can convey a representational content because it can occupy different physical states related to what it purports to represent. The angle between the direction α of the arrow and some reference line in the compass’ support can take any value from 0 to 2π, and α is nomologically related to the direction α’ of the North Pole. For each value αi corresponds a unique representational content Ri of the form “the direction of North is αi”. Under normal conditions we have α = α’; if α ≠ α’, the compass misrepresents the direction of North. Even though each value of α determines a unique representational content “R”, it is clear that the position of the arrow cannot be reduced in terms of what it represents. A physical state of C and the representational content it conveys are, obviously, different items. Indeed, α is not sufficient to determine “R”. C conveys a representational content “R” because it has the function of representing α’. Since the vehicle of representation and the representational content are different items, the following possibilities obtain: (a) The position of the arrow can represent something different from “the direction of North is α”. (b) The content “the direction of North is α” can be conveyed by different vehicles. (c) The position of the arrow can be used for nonrepresentational purposes. As an example of a, think about the compass C being used to determine the direction of induced magnetic fields around it. If these fields are strong enough, the Earth’s magnetic field becomes negligible and C reliably indicates the direction of the induced field. In this case, C conveys a representational content of the form “the orientation of the induced magnetic field is α”, instead of one of the form “the direction of

182

North is α”. So, it is clear that a given position of the arrow can determine different representational contents. A change in the purpose a device is used for (its function), changes the representational content conveyed by its physical properties. Possibility b obviously obtains. There is no limit for the number of physically different compasses that can be designed. But consider in particular the possibility of an “inverted compass” C’. C’ and C share all their physical properties, except that their arrows are magnetised with contrary polarities. Now, suppose that the convention is the inverse of the established one: It is the back of the arrow, not the head, what is meant to point to the north. Clearly, the representational content conveyed by C, under the ordinary convention, is the same that the representational content conveyed by C’, under the contrary convention. As an example of c, consider that the compass C is used for a magic trick. The fact that the arrow can move without any observable force acting on it can be used, in appropriate circumstances, as an instance of abnormal phenomena. In this case, the position of the arrow is not used to represent anything. Clearly, physical properties of the vehicle C, like its weight, its shape, or its rotative arrow, can be used for nonrepresentational purposes. Now, consider a compass CC designed in a way that the direction of North is not given by the direction of an arrow but by the value of an intrinsic property Q. First, Q can take several values. Just as a capacitor can take an electric charge between, say, 0 and 50 nF, Q can take values between 0 and 2π in some continuous scale. Second, the values that Q takes covary, under normal conditions, with the location of the North Pole relative to the compass. Each value of Q determines a unique north direction. Clearly, the compasses C and CC convey the same representational content, even though they are different vehicles of representation. With C, the direction of North is obtained from the direction of the arrow. With CC, it is obtained (by a subject or mechanism) by measuring or evaluating Q or another magnitude nomologically related to Q. Even though I argued that the consciousness property is a physical property, I think it is extremely implausible for a compass, or a thermostat, or something of the sort, to instantiate the consciousness

183

property. I said that consciousness certainly requires a highly complex system undergoing some complex dynamics. But just for illustrative purposes, suppose that the property Q is the c-consciousness property, and that for each specific value Qi of the property Q there is a specific what-it-is-like-ness of being CC, i.e., a specific phenomenal character. Each Qi is a different quale, and since Q is, ex hypothesi, a physical property, for Q to take a specific value Qi is for CC to be in a specific physical state. The quale Qi, i.e., the what-it-is-like-ness of being CC at a given moment, represents, as does the direction of the arrow of C, the direction of North. Firstly, there is, ex hypothesi, a relation between the physical property Q and the magnetic field of the Earth “eM”. The what-it-is-likeness of being CC co-varies with the direction of eM (or of another magnetic field), since eM causes the property Q to take a specific value Qi . This fact, results from a law of nature that directly relates Q and eM, or from a mechanism (ruled by laws of nature) that does so. Secondly, the function of Q is, ex hypothesi, to represent the direction of North. CC was designed with property Q and in such a way that the co-variation between it and the direction of the North Pole obtains. Now, it can be claimed that the property Q of CC could represent the direction of North without there being anything it is like to be CC. But this is not so: Each Qi is a quale. There are not two things: A physical property that co-varies with the direction of the North Pole, and the strange fact that there is something it is like to instantiate this physical property. Each Qi is a quale; a quale is a physical property; and the property Q happens to be nomologically related to other physical properties (like the direction of the surrounding magnetic field). The vehicle of representation is the quale, not something else that happens to have a quale. If you take out of CC the what-it-is-like-ness of being CC, you take out the property that represents the direction of North. Consider the following question: “What is the reason for the existence of a nomological relation between Q and the direction of a magnetic field?” This question is similar to: “Why two charged bodies attract each-other?” This is just the way the world (Wc) inhabited by CC is. This world is such that it includes Q among the intrinsic physical properties, just like mass and electric charge. And just as there are

184

nomological relations involving electric charge, there are nomological relations involving Q. Intrinsic properties have characteristic and unique physical behaviours, given by the relevant laws of nature. Q is among these properties; it complies with specific laws of nature; and it has specific causal powers. So, the system CC is related with other systems in virtue of instantiating Q. Because there is something it is like to be CC, and depending on the specific what-it-is-like-ness of being CC, the system CC interacts in a specific way with other systems. Could there be a “zombie” compass ZC, in the world Wc? If ZC is a system that coincides in every physical property with CC but lacks property Q, then ZC is not a compass, and cannot establish with the other systems all the relations that CC has with them. And if ZC physically differs from CC, whether it can behave exactly as CC in every respect is an empirical question; the answer depends on the available physical furniture in Wc and the laws of nature ruling it.227 The analogy between the compass CC and the human representational systems involving consciousness is far from perfect. But as the analogy suggests, for e-physicalism (1) phenomenal contents and, in particular, each quale, are physical properties; and (2) phenomenal contents are different from representational contents. In fact, for ephysicalism phenomenal content, when it has a representational character, is a property of the vehicle that conveys the corresponding representational content, and this vehicle can convey this content in virtue of having this property. I will develop these points next. 6.2 Phenomenal content and representational content There are good reasons to hold that the phenomenal content of some conscious mental states has a representational character. In particular, the phenomenal content of a conscious mental state resulting from a perception process seems to represent the perceived item. But is the phenomenal content identical to a representational content? The different versions of “representationalism”228 share the idea that

227 228

See Chapters 2 and 5. See, e.g., Dretske (1995) and Tye (1995).

185

phenomenal contents are nothing more than representational contents.229 The opponents of representationalism consider that even though phenomenal contents accomplish, at least in some cases, representational functions, they are not reducible to representational contents. I will argue, in opposition to representationalism, that phenomenal content can be dissociated from representational content. In fact, ephysicalism is incompatible with representationalism. Recall that: (PMQ) If mental states are physical states, a mental state has phenomenal content if and only if it instantiates a physical property (the m-consciousness property). (SMQ) If mental states are supervenient states, a mental state has phenomenal content if and only if its supervenience base instantiates a physical property (the m-consciousness property). Accordingly, the instantiation of a physical property is sufficient for a mental state M to have phenomenal content. It is not required for the M to have the function of representing something. By contrast, recall that for a state “R” of a system S to represent R it is necessary for S to have the function of representing R. I consider that the phenomenal contents of (conscious) perceptual states do have a representational character. Firstly, they are causally determined by external systems.230 For example, colour-qualia co-vary with the reflectances of the objects visually sensed.231 Secondly, the proper function of these contents is arguably to represent properties of the external items at their causal origin. But I think there are phenomenal contents that do not represent anything, and that even in the case of perceptual states a dissociation 229

Importantly, the representational contents that are claimed to be identical (or to supervene on) phenomenal contents are generally taken to be nonconceptual. 230 “External systems” includes the parts of the body of the corresponding subject that do not belong to the corresponding experiencer. 231 However, it has been argued that, at least in some cases, there is no one-to-one association between some specific type of phenomenal content and some specific type of physical property. See, e.g., Gray (2003).

186

between phenomenal content and representational content is possible. N. Block (2007) calls this view “phenomenism”.232 I will briefly propose a few cases that are supposed to show that: (1) The same phenomenal representational contents.

content

can

have

different

(2) The same representational content can be represented by different phenomenal contents. (3) A phenomenal content Q can lack representational character. 6.2.1 One vehicle, different contents Many thought experiments and psychological phenomena have been used, or can be used, to argue that the same phenomenal content can have different representational contents. I will briefly present two of them. Inverted earth The “Inverted earth” thought experiment was developed by Block (1990). Roughly, it goes as follows: Let us accept that each colour quale Qi of the experience of a native subject S of the Earth represents a particular reflectance Ri. For example, a p-red (phenomenal red) quale represents the reflectance of a ripe tomato. Now, imagine that there is an “Inverted earth”, which only differs from the Earth by the fact that the reflectance of every object is inverted, and that the subject S is taken there after having being equipped with permanent reflectance-inverting 232

Block says: “I think that sensations—almost always—perhaps even always— have representational content. What's more, I think that it is often the phenomenal character itself that has the representational content. What I deny is that representational content is all there is to phenomenal character. I insist that phenomenal character outruns representational content. I call this view ‘phenomenism’. Phenomenists believe that phenomenal character outruns not only representational content but also the functional and the cognitive; hence they believe in qualia.” (2007, p. 533. Italics in the original).

187

lenses. For S, every object O’ on Inverted earth will look as its counterpart O looked on Earth. But note, first, that on Inverted earth each colour quale Qi of the experience of S tracks reflectance -Ri, i.e., the counterpart of Ri. And note, secondly, that when S has been living long enough on Inverted earth the function of each quale Qi presumably became to track the reflectance -Ri instead of Ri. Certainly, when S just arrived to Inverted earth a quale Qi used to misrepresent the reflectance -Ri. But if the representational content depends on the function of the representational device, and a given function can be acquired in virtue of the history of the device, the representational content of the colour experiences of S changed. So, it seems that an object O on Earth and its counterpart O’ on Inverted earth can appear to S with the same colour quale Qi, even though the reflectance that Qi represents on Earth (Ri) is different from the reflectance that Qi represents on Inverted earth (-Ri). This scenario contradicts representationalism: If representationalism is right, it is not possible for two sense experiences to share their phenomenal content and yet differ in their representational content. Recall that a representational content is determined by (i) the item that causes the representation and by (ii) the item that the representational system is supposed to represent given its function. What the Inverted earth scenario presumably shows is that changes in i can be counterbalanced by changes in ii in a way that preserves the phenomenal content unchanged. A brain in a vat A “brain in a vat” scenario can give another case of same phenomenal content with different representational content. Consider the totality of stimuli that the brain of an ordinary subject S receives throughout his life. Suppose that (i) the phenomenal characters of the experiences of S represent the objects and properties that caused them, and that (ii) these causal chains include as an intermediary link some particular stimuli on the brain of S. Now, suppose there is a replica “VB” of the brain of S, which instead of being embodied in a human organism is inside a vat,

188

and that is connected to a “virtual reality” machine. This machine exactly reproduces on VB the history of the particular stimuli that the brain of S received throughout his life.233 Presumably, given ii, the phenomenal characters of the experiences of the “subject” VB are qualitatively identical to the phenomenal characters of the experiences of S. However, given i, these phenomenal characters do not represent the same things. To be sure, it is not completely clear what is the representational content of VB’s experiences. It can be claimed that they represent the states of the virtual machine that caused them; or that they represent virtual objects; or that they do not represent anything since brains evolved to be embodied and not inside vats. But, in any case, it seems we have a situation where there is the same phenomenal content and yet differences in representational content. If brain in a vat scenarios are coherent and then, in principle, possible, then phenomenal content is internal. 6.2.2 One content, different vehicles There are many scenarios that are aimed at showing the possibility of experiences that differ in phenomenal content but share the representational content. I will discuss two of them. Synaesthetic experiences In an ordinary subject each type of stimuli T causes experiences in a specific sensory modality M. For instance, light causes a visual percept and vibrating air causes a sound percept. But in synaesthete subjects, stimuli of type T cause experiences not only in a modality M but also in another modality M’. For instance, the light reflected by a ripe tomato causes not only a red percept QV, but also a sound percept QH with a given timbre. 233

Suppose, furthermore, that during their respective lives the stimuli that VB received perfectly matched the stimuli that the brain of S received only by chance. Otherwise, it can be argued that the “distant causes” of the stimuli of VB are the same as the causes of the stimuli of the brain of S (I am indebted to M. Kistler for this remark). In fact, the purpose of including an ordinary brain in the example, which determines the stimuli that VB receives, is just to make the mental life of VB (presumably) coherent and realistic.

189

Accordingly, in the case of a synaesthete subject, it seems that different qualia belonging to different modalities can represent a single item. For instance, both the visual quale QV and the synchronous auditory quale QH can represent the reflectance R of a ripe tomato. If QH co-varies, systematically, with R, and thus becomes a reliable indicator of R, it can acquire the function of representing R.234 So, it seems that a synaesthetic experience can have the same representational content that an ordinary experience, even though the former differs in its phenomenal content from the latter. This possibility contradicts representationalism. Inverted spectra Inverted spectra scenarios, which I previously discussed,235 can be used against representationalism. The underlying intuition is that the association between each colour-quale and the reflectance it represents is (metaphysically) contingent: Phenomenal contents that differ by having the colour-qualia inverted (relative to each-other) can have the same representational content. For a representationalist inverted spectra scenarios are not coherent. For him, the representational content completely determines the phenomenal content, and thus for equal representational contents there are equal phenomenal contents. However, I do not see any patent incoherence in the possibility of inverted spectra. If I wear inverting lenses I will be able to notice, as soon as I try the lenses, that colours are inverted. Now, what prevents the existence of a creature S such that its spectrum is, and always was, inverted relative to mine? Certainly, when I try the inverting lenses the colour-qualia of my experience start to misrepresent reflectances. But for a creature S belonging to a species that has evolved with its spectrum inverted (relative to mine) this would not be the case. The colour-qualia of S would veridically represent 234

As an alternative it can be claimed that, while in the case of an ordinary subject only QV represents R, in the case of the synaesthete the pair formed by QV and QH represent R. For an elaboration of the argument from synaesthesia see Wager (1999). 235 See Chapter 2.

190

reflectances. So, it seems that the colour-qualia in my visual experiences and the ones of the experiences of S represent the same properties, and yet there is a difference in the corresponding phenomenal contents. This possibility contradicts representationalism. To be sure, it is controversial whether inverted spectra cases can be detected since, ex hypothesi, these qualitative differences do not entail any functional difference. But the possibility of inverted spectra does not depend on the availability of epistemic criteria to detect them. If physicalism is true, there is a physical difference that amounts to the fact that a spectrum E’ is inverted relative to a spectrum E, and this physical difference can (in principle) be observed in the absence of an adequate scientific theory that reveals the relative inversion among S and S’. It is important to note, first, that inverted spectra apparently must comply with some structural constraints that grant functional equivalence to the noninverted ones. It has been argued that the totality of the 3-dimensional spectrum has to be inverted; just to invert two colour ranges would not do.236 Secondly, note that even if inverted spectra were possible it is not obvious that “inverted qualia” can obtain in other modalities. Concerning vision and audition I take qualia inversions to be plausible possibilities. But regarding touch, taste, smell, and proprioception, it is not clear that this is the case.237 Let us briefly consider an example related to pain. Two subjects, S and S’, behave the same way in relation to items at high temperatures: They avoid being burnt. However, when S’ touches, e.g., a red hot charcoal he experiences a quale Q’ that is the inverse of the quale Q that S experiences in the same situation. While Q is a burnt pain-quale, Q’ is some pleasure quale. Is this a plausible 236 237

See Palmer (1999). Dokic (2000) argues that bodily qualia, e.g., pain-qualia, differ from the qualia involved in external perception, e.g., visual qualia, in their reflexivity. Probably, the reflectivity of some “modalities” of qualia coincides with the impossibility of them being inverted, since “the intentional object of bodily awareness, unlike that of external perception, constitutively depends on the subject’s experience.” (p. 77).

191

scenario? Moreover, given a pain-quale Q, is there a specific (pleasure) quale Q’ that is its inverted counterpart? Inverted spectra cases seem plausible because, presumably, the difference there is between two colour-qualia does not entail a difference between the functions they can accomplish. For representational purposes, the essential feature of colour vision is the contrast there is among different colours. It does not seem to matter which colour represents which reflectance, as far as different colours represent different reflectances, and some relations among colours mirror some relations among reflectances. But in the case of pain/pleasure it seems that the differences between two qualia entail differences in the functions that these qualia can accomplish. It seems that a pain-quale cannot accomplish the same function that a pleasure quale. A creature avoids painful sensations because they are painful, and looks for pleasure sensations because they are pleasant. The implausibility of pain/pleasure inversions favours the representationalist view. If phenomenal content is just representational content, a pain-quale is a representation of something to be avoided; the possibility of a pain-quale representing something desirable is precluded. However, I think that in order to defend the distinction between phenomenal and representational content it is not necessary for every modality to be “invertible”. If inverted spectra are possible, this is enough for the case against representationalism. What the implausibility of qualia inversions in some modalities challenges is epiphenomenalism about qualia and, in general, the idea that phenomenal properties cannot accomplish any causal role.238 I will come back to this question later. In sum, the inverted spectra argument seems to show that phenomenal content can be dissociated from representational content. However, it is dubious that inverted qualia are possible in every perceptual modality. So, the possibility of inverted spectra should not be taken to show that, in general, qualitative differences do not entail functional differences. 238

For instance Kim (2005) argues that qualitative properties cannot be “functionalised”, since they do not have causal powers.

192

6.2.3 Nonrepresentational phenomenal contents One of the most suggestive arguments for representationalism is the alleged transparency (or “diaphanousness”) of experience. In M. Tye’s words, transparency is “the thesis that, when we introspect, the qualities of which we are directly aware are qualities of external things, if they are qualities of anything at all.” (2009, p. xiii). The idea is that qualia are always experienced as properties of external objects. In fact, in order to attend to a quale, one does not attend to putative properties of one’s experience; instead, one attends to the objects presented in experience. Therefore, it seems that qualia are properties of the perceived objects, and not of the subjective experience. For instance, J. Campbell’s says when describing his “Relational View” of experience that “the qualitative character of the experience is constituted by the qualitative character of the scene perceived.” (2002, pp. 114-115. Italics added). If transparency obtains then, as representationalism maintains, every quale represents some item. To be sure, the acceptance of transparency does not force an identification of phenomenal content with representational content,239 but the case against representationalism is seriously weakened. If phenomenal content is different from representational content, why is it the case that—according to the representationalist—every phenomenal experience has a representational character? If the transparency thesis obtains the very essence of every quale is representational. In order to reject the argument from transparency it must be shown that there are phenomenal contents that do not have a representational character. I will briefly argue that this is the case since (1) there are some sensory qualia that do not represent anything and (2) some nonrepresentational mental states have phenomenal content. (1) Certainly, the proper function of perceptual states is to represent external items. So, under normal conditions, the sensory qualia in the phenomenal content of a mental state represent external items that are at the causal origin of them. Because of this, in order to experience, e.g., a p-red-quale, one has to look to a red object. However, under 239

Tye (2009) rejects a version of representationalism that he calls “strong intentionalism” while retaining the thesis of transparency.

193

abnormal conditions, a mental state M can have a sensory quale Qi that (i) was not caused by an external object that produced a stimulus through a sense receptor and (ii) does not present any aspect of the world to be somehow.240 Accordingly, a mental state M can have a quale Qi that does not represent or misrepresent anything. This seems to be the case, for instance, with the (visual) experience of phosphenes or the (auditory) experience of tinnitus. Phosphenes are present in the visual field, but they do not appear as objects or as properties of objects. They do not seem to be at any particular distance; they do not seem to occupy a position in the tridimensional visual scenario. They are present in the visual field, but they do not appear as components of the visual scene (by contrast with, e.g., a table one perceives while experiencing phosphenes). A similar situation occurs when one closes the eyes. There is a visual experience, a phenomenal content, but there are no objects in some visual scene. Indeed, one can attend to the qualities of this visual experience, even though there is no real or hallucinated “object” to attend to.241 Similarly, under tinnitus there is a sound that is experienced but that does not appear as coming from any direction. And one can attend to the sound, even though there is no source of sounds to attend to. The sound fills the totality of the auditory experience, its background, and thus is not and cannot be experienced as coming from any object. (2) If the proper function of perceptual states is to represent external items, and only perceptual states involve qualia, then the proper function of every type of qualia arguably is to represent some item. This 240

In the case of hallucinations (and brains in a vat and the like) it can be claimed that, even though there is no external object, the corresponding mental states do represent the world as being somehow. In fact, transparency still obtains: Qualia appear as properties of the hallucinated “objects”. Accordingly, hallucinations would be cases of misrepresentation, and not cases of phenomenal content without representational content. 241 Block says about “phosphene-experience”: “Can you attend to those sensations? I believe I can. Even if you can't attend to them, are you aware of them? According to the representationalist, all awareness of those sensations could only consist in awareness of the colored moving expanses that are represented by them. My view is that one can be aware of something more. Again, I don't think this sort of consideration can change anyone's mind, but I hope it will have an impact on the noncommitted.” (2007, p. 546-547).

194

result, clearly, is consonant with representationalism. However, it can be argued that there are mental states other than perceptual states that also involve qualia. Some nonperceptual states seem to have phenomenal contents that are likely to have nonrepresentational functions. G. Rey says: “Many have noted that states like that of elation, depression, anxiety, pleasure, orgasm seem to be just overall states of oneself, and not features of presented objects.” (1998, p. 441. Italics in the original). In fact, what could the experience of a sudden sharp emotion represent? Obviously, emotions have a cause and, arguably, a biological function. But it is clear that they do not represent any property of an external object, and it is very doubtful that they do represent some internal state. Suppose one is depressed. What could the emotion of depression represent? Does it represent a brain state? Certainly not: The experience of depression does not represent the brain as being somehow. Does it represent the state of being depressed? Certainly not: The emotional state is the state of being depressed. Moreover, note that when a state “R” of a system S has the function of representing R, it is always possible for S to misrepresent R: The representation “R” might be caused by R’, and not by R. Now, is it possible for the emotion of being depressed to misrepresent the state of being depressed? Certainly not; once again, the emotion of being depressed is the state of being depressed. In sum, for the e-physicalism view qualia are physical properties. Some qualia have the function of representing something, and when they do so they convey a representational content. But some qualia do not seem to represent anything because there is no nomological dependence between them and any property of an external object, or because they do not have the function of representing anything. If in fact there are nonrepresentational qualia, representationalism is to be rejected. Note that in disciplines like music, gastronomy, visual arts, perfumery, etc., the perceptual systems are customarily exploited for nonrepresentational purposes. The aim is to produce certain qualia, not to represent something with the qualia. Through sense stimuli, an experience with the desired phenomenal content is intended. To be sure,

195

this possibility of using perceptual mechanisms for nonrepresentational purposes does not contradict representationalism, since the phenomenal content still represents what it is supposed to represent given its proper function. But this possibility suggests that there is a dissociation of phenomenal content and representational content. If phenomenal content is identical to representational content, there is no point in bringing about some specific phenomenal content unless there is the aim of exploiting a representational content. 6.3 Phenomenal character and physical states For e-physicalism consciousness is an internal physical property of the corresponding experiencer. Now, I claimed, firstly, that the instantiation of this property in a subject S does not cause the fact that there is something it is like to be S. For S to instantiate the consciousness property is for S to be such that there is something it is like to be S. Secondly, I suggested conceiving the consciousness property as taking different “values” in the same sense that the temperature property can take different values. To each one of the possible values of the consciousness property would corresponds a specific what-it-is-likeness, in the same sense that to each one of the possible values of the temperature property corresponds a specific temperature. Accordingly, I will say that: (C8) The fact that a (conscious) subject S instantiates at t a physical property X, namely, the c-consciousness property, with a specific value Xi, is identical to the fact that there is at t a specific what-it-is-like-ness of being S. Statement C8 is, no doubt, quite problematic. We have the intuition that qualia and, thereby, phenomenal characters, cannot be something physical. But the idea of a causal entailment from the instantiation of a specific physical property to the occurrence of a specific phenomenal character is even more problematic. If phenomenal contents are physical properties, at least there is no place for contingency (except in our intuitions). D. Papineau (2002) argues that, even though conscious mental

196

states have a physical nature, we can think about them in two different ways: (i) Physically, i.e., as physical states or (ii) phenomenally, i.e., as states such that there is something it is like to be in them. The referent is the same in both cases, but is conceptualised in different forms. A similar idea is defended by J. Ramos (2001). He adopts a mind-brain identity theory in a monist framework but argues, firstly, that every mental phenomenon has three aspects: The physical, the psychological, and a further one that is social.242 Secondly, he claims that it is essential of a mind that it has these three aspects. Certainly, we can think about a single item under different “modes of presentation”, and each one of these modes can present a different aspect of the (single) referent. But if one holds that conscious mental states have a physical nature, the physical mode of presentation is privileged over the other(s) one(s). It is the physical conception of conscious mental states that reveals their ontological nature. Thus, one would expect that the physical way of thinking about conscious mental states could capture everything that can be known about them and, in particular, the fact that there is something it is like to be in them. There should be no essential property of conscious mental states that can only be captured by the phenomenal (or psychological) way of thinking. This is the reason why C8 proposes to conceive phenomenal contents as some type of physical property. This proposal, prima facie, can be taken as implausible or as a category mistake. But, if physicalism is right, I think there is no other option for each token of a quale than to be identical to some physical property. The physicalist has to bite the bullet. Previously243 I suggested that the intuition that qualia cannot be physical properties might follow, at least in part, from a misconception of the subjectivity of experience. We might take this subjectivity as ontological, in contrast with the objectivity of the physical world. But 242

The physical aspect concerns the physical substratum of every mind; the psychological aspect is the mental life as experienced from the first-person point of view; and the social aspect is the mind as experienced from the point of view of other minds, that is, from a social and cultural environment where behavioural concepts are rooted. This social aspect is essential since “only through the internalization of social practices can a mind be brought to life” (Ramos 2001, p. 21. Translated from Spanish). 243 See Chapter 3.

197

the subjective character of experience can be explained as the result of an epistemological condition. Therefore, the possibility is open for qualia to be objective properties and, thereby, physical properties. Here, I want to suggest that another source for the intuition that phenomenal contents cannot be physical properties might be a tendency to reify physical properties. Think about an electron. On one hand, it has a mass, a spin and an electric charge. On the other hand, it is supposed to be simple, that is, not composed of further entities. So, if one conceives properties as some sort of entities, the concept of an electron is incoherent. But if there is a clear distinction between the substance and the properties of the substance, there is no incoherency: The electron instantiates a mass, a spin, and a charge, but the electron is not a mass, is not a spin, is not a charge, or some kind of mixture of them. Correspondingly, to say that a physical system S instantiates the property P, does not mean that S includes among its constituents P-ness. If one says, on one hand, that a phenomenal content P is a physical property and, on the other hand, that P consists of a red-quale, this does not mean that the physical property P is red, nor that S has a red constituent. A red-quale is a physical property, not a physical entity. A red-quale is not something red. Certainly, I do not pretend to posses a solution for the questions of the relation there is between a substance and its attributes, and of the ontological nature of the attributes. This is a very tricky and difficult issue. I just want to make the following suggestion: To be a realist about consciousness and, in particular, about qualia, does not involve a commitment to the existence of some sort of “redness” substance. The what-it-is-like-ness of being S can include a red-quale, and this redquale can be a physical property, without S having any “redness” constituent. Why, when a subject S instantiates a physical property—the cconsciousness property, there is something it is like to be S? This is the case because in nature, among other properties, there is the consciousness property, and for S to instantiate the consciousness property is for S to be such that there is something it is like to be S. I claimed that some physical systems, in virtue of their structure and the dynamics they undergo, instantiate the consciousness property. The

198

question of what structures and dynamics are required for consciousness can be answered by specifying them. But the question of why a system complying with these structures and these dynamics is such that there is something it is like to be it, cannot be answered in a noncircular way. 6.3.1 Phenomenal character and biological functions For e-physicalism phenomenal characters and contents are physical properties, which were (and still are) subject to natural selection. They acquired a function through evolution in virtue of their physical condition, their causal powers, and the effect that the exercise of these powers had for the fitness of the corresponding creature. Let us consider an example. Presumably, through evolutionary history the items that cause burns had a negative impact for a creature’s chances of survival. Because of this, the what-it-is-like-ness of feeling burn pain is such as to cause a behaviour that prevents or stops this sensation: (i) The proper function of the system that produces tactile sensations is to represent high temperatures in a painful manner. (ii) The proper function of the burn-pain-quale is to cause avoidance-of-hot-items behaviour. Consider two subjects, S1 and S2. The (masochist) subject S1 deliberately and periodically burns herself, while the (ordinary) subject S2 systematically avoids to be burnt. Suppose that when S1 and S2 touch a red-hot charcoal, they experience the qualia Q1 and Q2 respectively. There are two interesting options concerning the relation between Q1 and Q2: (1) Q1 and Q2 represent the same property (a high temperature), but Q1 represents it in a pleasurable way while Q2 represents it in a painful way. (2) The quale Q1 is of the same type that the quale Q2. Both represent the same property (a high temperature) in the same painful way. If 1 is the case S1 is not a masochist after all. Since S1 does not experience a pain burn quale when touching the charcoal, an avoidanceof-hot-items behaviour is not caused. The problem comes from the

199

system that produces tactile sensations. It represents a high temperature in a pleasurable way, as if being burnt was something (historically) positive for survival. If 2 is the case the system of S1 that produces tactile sensations correctly represents the high temperature: It represents it in a painful manner. The problem is that the pain burn quale Q1 does not performs its function, since S1 looks after being burnt instead of avoiding it. In both cases, the subject S1 is not fit from the evolutionary point of view. She behaves against her interest of survival and, thereby, of reproduction. In scenario 1, the system that produces tactile sensations represents a property of the world in an inappropriate way. In scenario 2, her tactile experience does not cause the appropriate behaviour. Accordingly, natural selection will favour subject S2 over S1. In general terms, and very roughly, the evolutionary history of phenomenal contents could be something like this: Our world is such that some physical systems, in virtue of their structure and the dynamics they undergo, instantiate the cconsciousness property. Now, this property can take a huge number of different “values”: There are many possibilities for the what-it-is-like-ness of being some entity. At some point in the evolutionary history of life, some creature happened to instantiate the c-consciousness property. There was a what-it-is-like-ness of being it; it had mental states with phenomenal contents. Since qualia are physical properties, firstly, some properties of the environment and of the body of the creature were able to cause some qualia to obtain. Which qualia obtained was a function of the physical characteristics of the stimuli. Secondly, the occurrence of qualia, in turn, had effects on properties of the environment and of the body of the creature. Which were these effects was a function of the physical characteristics of the cause, i.e., of the particular qualia. In short, qualia were causal intermediaries in some interactions between the organism and the world. Qualia were susceptible of playing causal roles. Each interaction between an organism and the environment,

200

and in particular the ones mediated by qualia, had an impact for the biological interests of the creature. So, through selection pressures, qualia that increased the fitness of the creature were selected. More and more sophisticated representational systems, with more diverse possible qualia, and with more cognitive functions played by qualia, appeared. The phenomenal contents became rich and reliable in the performance of their functions. At some point in the evolution of these cognitive systems there was a “decoupling” between the environment and the qualia. The systems acquired the capacity to internally induce some qualia by simulating, to a certain extent, their cause. And, by inducing some qualia, the systems also acquired the capacity to trigger, to a certain extent, some of their effects. There was the possibility of remembering and imagining phenomenal experiences. 6.3.2 Phenomenal space   What I label “phenomenal space” is the abstract space generated by the different “values” that the consciousness property can take. Each phenomenal character (and each phenomenal content) corresponds to a precise position in phenomenal space, and each one of these positions corresponds to a structure of qualia. For illustrative purposes, I will advance a simple and rudimentary model of phenomenal space using examples from human phenomenology. The idea is only to show in what sense one instance of a physical property could determine a whole phenomenal character. Recall that the consciousness property is—I argued—an emergent property instantiated in complex physical systems. Therefore, the emergence base of the consciousness property can occupy different states that account for the different “values” that it can take. Firstly, phenomenal characters include qualia in different “modes”. For instance, there are visual and auditory contents. Which are the different modes, and if they can be sharply differentiated, is a controversial question.244 For some theorists every quale belongs to one of the sensory modalities; for others, states like emotions, or cognitive processes like metacognitive activities,245 also involve distinctive qualia. 244 245

See, e.g., Spence & Deroy (2012). See, e.g., Proust (2007).

201

I will not argue for any particular view; let the variable i represent the different modes of qualia there happen to be. Now, to each one of these modes corresponds a dimension xi in phenomenal space. So, if every quale belongs to one of five sensory modalities, phenomenal space has five dimensions. Secondly, in each mode of qualia there are different types of qualia. For instance, vision includes colour, edges, regions, textures, etc. Accordingly, to each dimension xi in phenomenal space can be associated a subspace xij. For example, if i =1 corresponds to the visual modality, the variable j can correspond to the different types of content (colour, edge, etc.). Thirdly, it seems that, in turn, the values that can take each type of content (belonging to a single modality) generate a further subspace. For instance, the different colours form a tridimensional space. So, each subdimension xij can in turn contain further dimensions. For example, if i = 1 corresponds the visual modality, and j = 3 to the colour constituent, k = 2 can represent a particular dimension of colour. Certainly, to completely specify a phenomenal character, there must be added further dimensions, as well as constraints on the possible values that simultaneous qualia can take in each dimension. These constraints would account for structural properties like the fact that a region must be delimited by an edge. Finally, it is clear that phenomenal characters change over time; experiences appear, evolve, and disappear. Accordingly, I will say that the temporal evolution of a phenomenal character (and of a phenomenal content) corresponds to “a trajectory” in phenomenal space. Here, the important point is that for every difference between phenomenal characters there must be a difference in the “value” of the consciousness property. And of course, a plausible model of human phenomenal space will be much more complex than what I sketched. For instance, it would have to take into account the role of attention, the capacity of perceptual memory, the contextual effects, the putative permeability of phenomenal contents in the case of ambiguous figures, and the spatial distribution and temporal specificities of the qualia involved, e.g., in “positioned scenarios” (Peacocke 1992). Note that for the e-physicalism view, when phenomenal content

202

conveys representational content, the representation is analogue. There are different proposals about how to draw the digital/analogue distinction, and about the distinctive characteristics of a digital representation versus an analogue one.246 I propose to draw the distinction in the following manner: In the case of an analogue representation, the physical properties of the vehicle of representation and the physical properties of the represented object are isomorphic to some extend or in some respect. In other words, when the state “R” of a system S represents R, there is a (quasi) isomorphism between some of the physical properties of “R” and some of the physical properties of R. The shape of the grooves on a vinyl record gives one example. There is an isomorphism between these shapes and the “form” of the corresponding sound waves produced through the phonograph. The colours and spatial relations among the pigments of a photo paper give another example. These properties are nomologically related with the spatial relations and the reflectances of the objects represented by the photo paper. By contrast, in a digital representation the vehicle of representation belongs to a higher-order level that the physical level—it supervenes on physical items. Even though symbols are realised by some sign or “mark”, they are not identified with the physical object that constitutes the mark; each symbol is “multirealisable”. As a result, in order for some marks Mi to realise the representation “R” of R no isomorphism is required between the physical properties of Mi and R. Now, if qualia—the constitutive ingredients of phenomenal characters and contents—are physical properties, and each quale corresponds to a specific value of a physical property, phenomenal characters and contents are to be individuated at the physical level and not at a higher-order level. Therefore, phenomenal contents, when they convey a representational content, they do in an analogue or nonconceptual format.247 This idea, i.e., that phenomenal content 246

See, e.g., Lewis (1971); Goodman (1976); Dretske (1981); and Trenholme (1994). 247 When it concerns the content of mental representations, the distinction symbolic/analogue is usually equated to the distinction conceptual/nonconceptual.

203

represents (when it does) in a nonconceptual way, has been defended by many philosophers especially in the last three decades.248 6.4 The knowledge argument F. Jackson’s “knowledge argument” (1982) starts with the following scenario: Mary, the vision scientist, has been living her whole life in a white-and-black room; she never had the experience of seeing a colour. However, Mary knows “everything in completed physics, chemistry, and neurophysiology, and all there is to know about the causal and relational facts consequent upon all this, including of course functional roles.” (p. 291). One day, for the first time, Mary goes out of her room and, for instance, sees a ripe tomato and then has the experience of seeing red for the first time. The question is: Does Mary learn something new? “It seems just obvious that she will learn something about the world and our visual experience of it. But then it is inescapable that her previous knowledge was incomplete. But she had all the physical information. Ergo there is more to have than that, and Physicalism is false.” (Jackson 1982, p. 130. Italics in the original). There are two main ways to defend physicalism from Jackson’s conclusion. One is to claim that, in spite of our intuitions, Mary does not learn anything new. The other, to argue that physicalism is compatible with the idea that Mary learns something new.249 I will outline the most influential proposal in each direction, and follow the second one. 6.4.1 Mary does not acquire new knowledge A realist about qualia can defend physicalism by claiming that, in fact, 248 249

See, e.g., Gunther (2003). Perry (2001) exploits the perspectival character of consciousness to defend “antecedent physicalism” from Jackson's knowledge argument. Roughly, Perry claims that phenomenal contents are only subjectively knowable because they are a kind of “reflective contents”. He writes: “[T]here is no reason to think that the difficulties involved are special problems for physicalism. The culprit, I believe, is a certain doctrine about knowledge, one that crept into the problem with the remark in setting up the problem. As I put it: ‘Mary learns everything there is to know about the physical world, for after all, what is known can be written down and she can read it.’ This remark relies on what I call the ‘subject matter assumption’.” (p. 102. Italics in the original). The “subject matter assumption” says: “The content of a belief is simply whatever is believed about whatever the belief is about.” (p. 113).

204

Mary does not learn anything she did not know before. In this direction, the most interesting approach is the “phenomenal concepts” one.250 The idea is that Mary acquires a new way of thinking about something that she already knew, namely, the colour red. Mary does not learn about the existence of a new item; her knowledge about red was complete before she had the experience of seeing red. What Mary acquires is a new mode of presentation of red, namely, a “phenomenal concept”, which can only be acquired by experience. Tye (2009) characterises phenomenal concepts as follows: [P]henomenal concepts are supposedly the concepts we exercise when (but not only when) we notice or become aware of the phenomenal character of our experiences and feelings through introspection. Our experiences have phenomenal character whether or not we attend them. But when we notice how an experience feels, what it is like, in doing so we are bringing it beneath a phenomenal concept. (p. 56) First, I wonder if there are motivations for the thesis of the existence of these phenomenal concepts apart from the import it has for the knowledge argument. Secondly, I wonder how phenomenal concepts are supposed to be individuated. There are two options: (i) Each quale we experience is brought beneath a phenomenal concept, or (ii) several qualia are brought beneath a single phenomenal concept. Both options seem problematic. Option i is implausible. Given that we experience a huge amount of different qualia, we would have a huge amount of different phenomenal concepts. For instance, we can discriminate millions of different colours in a visual scene. In doing so, do we exercise millions of different concepts of colours? Now, it can be claimed that phenomenal concepts are demonstrative indexicals with the form “this colour”. Accordingly, it would be possible to master infinitely many phenomenal concepts by mastering their character. But phenomenal concepts are supposed to be exercised not only when we notice the 250

See, e.g., Tye (1995), Papineau (2002) and Levin (2009).

205

phenomenal character of an experience, but also when we recognise the phenomenal character of an experience, and for this recognition to be possible, a preexisting infinitely rich comparison class is, presumably, required. Option ii requires an account of how the extension of a phenomenal concept is delimited. But first, given the fine grain of phenomenal contents,251 any delimitation is likely to be arbitrary. Second, if there are less phenomenal concepts than qualia we can experience, while having an experience it is possible to exercise less phenomenal concepts than qualia we experience and presumably recognise. Thus, it is not clear what the utility of the phenomenal concepts is. Certainly, we do have concepts of colours, sounds, emotions, and so on, which we exercise in ordinary language when we talk about our experiences. But the phenomenal concepts proposed to deal with the knowledge argument are not supposed to be these public concepts.252 If they were, they would not be that special and they would not capture the what-it-is-like-ness of being in a conscious mental state. Indeed, according to Tye (2009) the available accounts of phenomenal concepts are flawed precisely because they have not clearly distinguished phenomenal concepts from ordinary concepts that do not have anything to do with qualia. That is why, for Tye, the phenomenal concepts approach has failed in providing a way out of the knowledge argument. He says that zombies can “express the very same concepts as we do with their ‘what it is like’ and other phenomenal vocabulary […] It follows that the concepts we express in our ‘what it is like’ (and other phenomenal vocabulary)—the concepts we apply introspectively to what it is like for us—do not require for their possession that we have undergone the relevant experiences.” (2009; pp. 68-69). 6.4.2 The ability hypothesis   D. Lewis (1983; 1988) defended physicalism from the knowledge 251

Peacocke (1992) argues that the phenomenal contents of perceptual experience are “fine grained”. Based on this, he claims that the representational content they convey is nonconceptual. 252 Phenomenal concepts seem to have a private character. If this is the case, they are susceptible of Wittgenstein's critique of private languages.

206

argument by claiming that Mary acquires a new knowledge, but that this knowledge is not factual. There is nothing that is the case that Mary did not know before leaving her room. The knowledge that Mary acquired is not propositional but a know-how: After seeing red for the first time, Mary has some abilities that she lacked before: She can imagine, remember and recognise the red colour or the experience of seeing red. Let us call these abilities the “Lewis abilities”. Different criticisms have been raised against Lewis’ proposal; I will briefly outline two objections. First, Lewis abilities should be grounded somewhere. If Mary has the ability to imagine (or remember or recognise) seeing red, presumably she acquired a representation of the what-it-is-like-ness of seeing red. It is this representation what accounts for Lewis’ abilities, and not the other way around.253 Indeed, it has been argued that Lewis abilities are neither necessary nor sufficient to know what it is like to see red.254 Second, it seems that it is not true that Mary, after seeing a colour for the first time, acquired the ability to recognise it. When she left her room and saw a ripe tomato, it was a specific shade of red, say Ri, what caused her experience. Thus, it is Ri what, according to the abilities hypothesis, she is now able to recognise. However, it seems that if Mary is confronted afterwards with some red object with a colour R*, similar but different from Ri, she might not be able to tell, on the base of her experience, if Ri is equal or not to R*. The discussion about the abilities hypothesis and phenomenal concepts involves many different questions, like the distinction between “know-that” and “know-how”, theories of concepts, representationalism, introspection, self-knowledge, and demonstratives. Thus, a comprehensive discussion goes beyond my present purposes and, in any case, I will briefly argue that even though Mary learns something new about the world, this does not show that physicalism is false.

253

Gertler (1999) argues that it is the knowledge of the what-it-is-like-ness what accounts for Lewis abilities, and not vice-versa. 254 On this see Conee (1994) and Alter (1998). For a discussion and defence of Lewis’ “ability hypothesis” see Nemirow (2009).

207

6.4.3 The limits of scientific knowledge Physicalism is the thesis that every entity that populates the world has a physical nature, not that everything that can be known about natural phenomena can be captured in scientific language.255 If before leaving her room Mary knows everything there is to know about vision, then she knows what it is like to experience any possible visual experience. For physicalism consciousness is a natural phenomenon, and recall that (D) an entity is phenomenally conscious if and only if there is something it is like to be that entity. Therefore, if Mary knows everything about every natural phenomenon, she knows everything about consciousness and, in particular, she knows what it is like to be in any possible conscious state. But if before leaving her room what Mary knows is everything that can be captured by scientific theories, then she does not know what it is like to see red. The argument is as follows: (R) Scientific theories cannot capture the intrinsic (or “internal”) character of any property; they can only capture the relations an intrinsic property holds with other properties. [premise] Previously,256 I argued that, (C3) The c-consciousness property has an internal character. [premise] Now, in this chapter I claimed that, (C8) The fact that a (conscious) subject S instantiates at t a physical property X, namely, the c-consciousness property, with a specific value Xi, is identical to the fact that there is at t a specific what-it-is-like-ness of being S. [premise] Then, from (R), (C3) and (C8) it follows that: 255 256

Flanagan (1992) and Alter (1998) argue in this direction. See Chapter 3.

208

(i) Scientific theories cannot capture the what-it-is-like-ness of being a conscious subject. [conclusion] In fact, it seems that, (ii) For Mary to have a complete knowledge about a given conscious mental state it is required for her to have instantiated this mental state. Premise (R) was proposed by B. Russell (within the framework of logical atomism): A piece of matter is a logical structure composed of events; the causal laws of the events concerned, and the abstract logical properties of their spatiotemporal relations, are more or less known, but their intrinsic character is not known. Percepts fit into the same causal scheme as physical events, and are not known to have any intrinsic character that physical events cannot have, since we do not know of any intrinsic character that could be incompatible with the logical properties that physics assigns to physical events. There is therefore no ground for supposing that percepts cannot be physical events, or for supposing that they are never compresent with other events. (1927, p. 384) Before her release, Mary knew which physical property Q is identical to a red-quale. In particular, she knew what the emergence base of consciousness is, and what phenomenal content emerges from each state of this emergence base. Thus, by observing the brain state of a subject S, she was able to tell if S was experiencing a red-quale. Moreover, she was able to predict in what brain state she would be if she ever experienced a red-quale. But Mary also knew that in order to know what it is like to be in a conscious mental state M it is required to have been in the mental state M. If a phenomenal character has a representational character, it has— I claimed—a nonconceptual content. And if a phenomenal character does not have a representational character, the question about the

209

conceptual or nonconceptual nature of its content cannot even be raised. In both cases, phenomenal characters cannot be captured in propositional terms. Scientific terms can refer to a physical property Qi that is identical to a red-quale (and to each item in its emergence base), but knowledge of what property the term refers to does not amount to knowledge of what it is like to instantiate this property. This knowledge is not propositional, and thus cannot be conveyed by scientific theories. E. Conee (1994) and Tye (2009) argue in this direction: Mary acquires a new knowledge that is not propositional. They use Russell’s (1912b) distinction between knowledge by description and knowledge by acquaintance: Before leaving her room, all that Mary knew about colours was knowledge by description. But knowledge of what it is like to see red can only be acquired by acquaintance, and “knowledge by acquaintance of an entity is a kind of nonconceptual, nonpropositional thing knowledge.” (Tye 2009, p. 136). Therefore, only when Mary had the experience of seeing red, she acquired knowledge of the what-it-islike-ness of seeing red. In Tye’s words: [W]hen Mary sees something red for the first time, she comes to be acquainted with red and thus with the phenomenal character of the experience of red. She does not come to be acquainted with the experience of red. Thus, what she knows when she sees something red is not a brain state, even if the experience of red is a brain state. (2009, p. 135. Italics in the original) In sum, by learning (true and complete) scientific theories Mary cannot acquire knowledge of what it is like to see red. By contrast, when Mary is seeing the ripe tomato she happens to be in a specific brain state: She instantiates a conscious mental state with the red-quale in its phenomenal content. Now, given her (human) cognitive abilities she can (1) “register” the relevant properties Pi of this brain state and, later, (2) “reproduce” properties that approximately match the Pi.257 Thereby, she 257

The difference in the vividness of the experience between seeing and imagining (or remembering) a quale, and the impossibility to recognise, e.g., a precise shade of red, can result from the fact that we are not able to induce exactly the same properties Pi.

210

learns what it is like to see red and she can remember, imagine, and recognise the red-quale. On the basis of experimental research that relates the brain states of human subjects with their behaviour and, in particular, with the reports they can provide of their experience, science can develop—and is developing—theories that relate these brain states with these reports. In fact, given that human brains are very similar to each other, it is reasonable to suppose that we have very similar experiences when we see a ripe tomato or when we feel burn-pain. And given the public character of language, it is also reasonable to suppose that we will describe these experiences in similar terms. Therefore, experimental research can probably manage to identify the emergence base of the consciousness property in humans—what is called “the neural correlate”—and to relate the different states of this base with different phenomenal characters and contents. Nonetheless, these theories will never be able to capture the what-it-is-like-ness of being in a given mental state. Only by instantiating this mental state a subject can come to know the corresponding what-it-is-like-ness. If one acknowledges that the consciousness property is an internal property, and that some mental states have qualities, it seems clear that the what-it-is-like-ness of being in some conscious mental state transcends the limits of science. But physicalism does not require everything that can be known about natural phenomena to be expressible in scientific language. 6.5 Conclusion For e-physicalism phenomenal characters are physical properties of the experiencer: Each “value” that the c-consciousness property can take in “phenomenal space” is a specific phenomenal character, and corresponds to a structure of qualia. Certainly, the thesis that qualia are physical properties is very controversial. The aim of this chapter was to make it plausible. First, I argued that phenomenal content is different from representational content. It can have the function of representing, and in that case the representational content it conveys is nonconceptual. Secondly, I suggested that the consciousness property has biological

211

functions that result from natural selection. Thirdly, I addressed Jackson’s “knowledge argument”. As the argument shows, the what-itis-like-ness of having a given experience can only be known by having the experience. However, this does not show physicalism to be false. Physicalism is compatible with the idea that not everything that can be known about natural phenomena can be captured in scientific theories. In particular, these cannot capture nonpropositional nonconceptual knowledge.

213

CONCLUSION The metaphysics of phenomenal consciousness is, no doubt, a “hard problem” (Chalmers 1996). Any theory of consciousness requires sacrificing some deep conviction somewhere. On one hand, dualism is, at best, obscure. Although consciousness does not seem to have a physical nature—especially from the subjective point of view, to my knowledge no one has so far advanced a coherent, or plausible, or enlightening, dualist alternative. Physicalism appears as the best option by default. But on the other hand, it is not easy at all to fit the phenomenon of subjective experience into a physicalist frame. And this is so especially if one holds a popular view like “microphysicalism”, or if one believes that something that can only be known subjectively cannot be physical. In short, given our present knowledge and philosophical tradition, there is no simple solution for the problem of consciousness. If one follows—as I do—the physicalist line, two broad possibilities are open. The first one is to adopt eliminativism and claim that, despite appearances and intuitions, in the (physical) world there is nothing that corresponds to “subjective experience”, “phenomenal consciousness”, “qualia”, and similar categories. To be sure, there is a widespread conception of physical reality that leaves no place for conscious phenomena. “What there is” is taken to be no more than mind-less subatomic particles aggregated in different structures. As a result, eliminativism is purported to be the only scientifically respectable view. However, I consider this conception of physical reality to be too simplistic and even naïve. We live in a complex and weird physical world. First, our best theories propose the existence of “entities” that we can only capture through highly abstract mathematical equations. It is not possible to adequately represent them in a pictorial or intuitive way. Thus, to what extent we really “understand” the functioning of Mother Nature is an open question. Secondly, contemporary physics is known to be incomplete. There is a gap between the images of the microphysical world and cosmology, which does not hinge on the details. Apparently, something is deeply wrong with microphysicalism.

214

I take it that the reality of consciousness is (or should be) beyond doubt. Consciousness may be different from everything that has been proposed, or from anything that we can conceive, but in any case it is part of (a physical) reality. Thus, if consciousness does not fit in our picture of the physical world, it is this picture that has to be revised, and not the belief in the reality of consciousness. This idea leads to the second physicalist option: To elaborate a metaphysical view of physical reality that allows for the inclusion of subjective experience among natural phenomena. The view I labelled “e-physicalism” is an attempt in this direction. Firstly, I argued for a layered picture of physical reality: Some properties of complex physical systems do not supervene on, but emerge from, their fundamental constituents, and thus are ontologically novel and have original causal powers. Secondly, I tried to make at least plausible the idea that the phenomenal properties of (conscious) mental states are physical properties with these virtues. In fact, I think there is no other option for the physicalist who is a realist about consciousness than to accept that phenomenal contents, whatever they specifically are, are something physical. No doubt that physicalism about consciousness is not very intuitive; but (some form of) physicalism is to be accepted for its merits in other areas. E-physicalism is intended to be compatible with contemporary physics and neuroscience. But given that natural sciences are empirical endeavours, new results and theories could undermine it or, hopefully, provide further support. Similarly, new philosophical ideas to come will surely have interesting implications for e-physicalism. Indeed, relevant arguments I am unaware of, or ignored by mistake, in all probability have been advanced. Many philosophers have been producing a very rich literature on the problem of consciousness, in particular during the last three decades. Among the several problems I discussed or mentioned in this work there are some that, in particular, deserve further development. Firstly, the physicalist account of the unity of consciousness (Chapter 3). I associated a phenomenal character with one instance of the consciousness property, and being this property an emergent one I suggested that it could take a huge amount of different “values” in “phenomenal space” (Chapter 6). This proposal is still very abstract. The

215

elaboration of a thoroughly physicalist account of the unity of consciousness is an extremely difficult and conceptually tricky enterprise, requiring further analysis and research. Secondly, the discussion of emergence versus microphysicalism (Chapter 4). There are several physical phenomena that can be analysed in order to explore the virtues and flaws of each view. Moreover, some interesting research is being done in the area of complex dynamic systems that can shed light on the question of how new properties emerge (if they do). Thirdly, the account of phenomenal contents (Chapter 6). From ephysicalism it follows that qualia are physical properties, internal to the conscious entity. Now, this is a highly controversial view with some partisans and many detractors. But given the complexity of “the problem of qualia” and the variety of not only philosophical but also empirical arguments involved, I only outlined some of the aspects of this discussion and sketched some germinal ideas. A full elaboration of the question of qualia surpasses the scope of this work. Finally, it is worth recalling that I faced some traditional metaphysical problems on which there is no general agreement. For instance, the questions of: Realism vs. nominalism; scientific realism vs. scientific antirealism; the nature of the laws of nature; the nature of time; personal identity; physical identity; vagueness and the problem of the many. I had to embrace a given position concerning each of these questions without providing enough (or any) justifications. Given the scope and complexity of “the hard problem of consciousness”, I count on some charity from the reader.

217

REFERENCES Alexander, S. (1920). Space, Time, and Deity. London: Macmillan. Alter, T. (1998). A Limited Defense of the Knowledge Argument. Philosophical Studies, 90 (1): 35–56. Alter, T. Walter, S. (eds.) (2009). Phenomenal concepts and phenomenal knowledge. Oxford: Oxford University Press. Aspect, A. Grangier, P. Roger, G. (1981). Experimental Tests of Realistic Local Theories via Bell’s Theorem. Phys. Rev. Lett., 47: 460. Aspect, A. Grangier, P. Roger, G. (1982). Experimental realization of Einstein-Podolsky-Rosen-Bohm Gedanken Experiment: A new violation of Bell’s inequalities. Phys. Rev. Lett., 49: 91. Aspect, A. Delibard, J. Roger, G. (1982). Experimental tests of Bell’s inequalities using time-varing analysers. Phys. Rev. Lett., 49: 1804. Baars, B. J. (1997). In the Theatre of Consciousness. The workspace of the mind. Oxford: Oxford University Press. Baars, B. J. Banks, W. P. Newman, J. B. (eds.) (2003). Essential Sources in the Scientific Study of Consciousness. Cambridge, Mass.: MIT Press. Bayne, T. (2008). The unity of consciousness and the split brain syndrome. Journal of Philosophy, 105 (6): 277–300. Bayne, T. (2010). The unity of consciousness. Oxford: Oxford University Press. Bedau, M. A. Humphreys, P. (eds.) (2008). Emergence: Contemporary Readings in Philosophy and Science. Cambridge, Mass.: MIT Press. Bell, J. S. (1964). On the Einstein-Podolsky-Rosen paradox. Physics, 1: 195-200. Bell, J. S. (1987). Speakable and Unspeakable in Quantum Mechanics. Cambridge: Cambridge University Press. Berkeley, G. (1948–1957) [1710]. The Works of George Berkeley, Bishop of Cloyne. A.A. Luce and T.E. Jessop (eds.). London: Thomas Nelson and Sons. Blackmore, S. (2005). Conversations on Consciousness. Oxford:

218

Oxford University Press. Bohm, D. (1951). Quantum Theory. New York: Prentice Hall. Bohr, N. (1935). Can quantum-mechanical description of physical reality de considered complete? Phys. Rev., 48: 696 (1935). Block, N. (1990). Inverted Earth. Philosophical Perspectives, 4: 53-79. Block, N. (2007). Consciousness, function, and representation. Cambridge, Mass.: MIT Press. Block, N. (forthcoming). Anti-Reductionism Slaps Back. To appear in Philosophical Perspectives. Brentano, F. (1874). Psychology from an Empirical Standpoint. London: Routledge and Kegan Paul. Broad, C. D. (1925). The Mind and Its Place in Nature. London: Routledge & Kegan Paul. Burge, T. (1979). Individualism and the Mental. In French, P. Wettstein, H. Uehling, T. E. (eds.) (1979). Burge, T. (1986). Individualism and Psychology. Philosophical Review, 95: 3–45. Campbell, J. (2002). Reference and Consciousness. Oxford: Oxford University Press. Carnap, R. (1966). Philosophical Foundations of Physics. New York: Basic Books. Carruthers, P. (2005). Consciousness. Oxford: Oxford University Press. Chalmers, D. (1996). The conscious mind. Oxford: Oxford University Press. Chalmers, D. (ed.) (2002). Philosophy of Mind: Classical and Contemporary Readings. Oxford: Oxford University Press. Churchland, P. Hooker, C. (eds.) (1985). Images of Science. Chicago: The University of Chicago Press. Clauser, J. F. Horne, M. A. Shimony, A. Holt, R. A. (1969). Proposed experiment to test local hidden-variables theories. Phys. Rev. Lett., 23: 880-884 (1969). Clark A. Chalmers, D. (1998). The Extended Mind. Analysis,

219

58:10-23. Coleman, S. (2006). Being realistic. In Strawson, G. et al. (2006). Conee, E. (1994). Phenomenal Knowledge. Australasian Journal of Philosophy, 72: 136–150. Crick, F. H. (1994). The Astonishing Hypothesis: The Scientific Search for the Soul. New York: Touchstone. Davidson, D. (1980). Essays on Actions and Events. Oxford: Clarendon Press. Davidson, D. (2001). Subjective, intersubjective, objective. Oxford: Oxford University Press. Dennett, D. (1987). The intentional stance. Cambridge, Mass.: MIT Press. Dennett, D. (1991). Consciousness explained. Boston: Little, Brown and Co. Dennett, D. (2005). Sweet Dreams. Cambridge, Mass.: MIT Press. Descartes, R. (1983) [1641]. Oeuvres De Descartes, 11 vols. Edited by Charles Adam and Paul Tannery, Paris: Librairie Philosophique J. Vrin, 1983. Dokic, J. (2000). Qui a peur des qualia corporels? Philosophiques. Le matérialisme contemporain, vol. 27, No 1: 77-98. Dretske, F. (1981). Knowledge and the Flow of Information. Oxford: Blackwell. Dretske, F. (1995). Naturalizing the Mind. Cambridge, Mass.: MIT Press. Duhem, P. (1906). La théorie physique; son objet, sa structure. Paris: Chevalier & Rivière éditeurs. Edelman, G. Tononi, G. (2000). A Universe of Consciousness. New York: Basic Books. Einstein, A. Podolsky, B., Rosen, N. (1935). Can quantummechanical description of physical reality de considered complete? Phys. Rev., 47: 777-780. Einstein, A. (1948). Quanten-Mechanik und Wirklichkeit. Dialectica, 2: 320-324. Faye, J. Needham, P. Scheffler, U. Urchs, M. (eds.) (2005). Nature’s Principles. Dordrecht: Springer. Flanagan, O. (1992). Consciousness reconsidered. Cambridge,

220

Mass.: MIT Press. Fodor, J. (1974). Special sciences. Synthèse, 28: 97–115. Fodor, J. (1983). The Modularity of Mind. Cambridge, Mass.: MIT Press. Fodor, J. (1990). A theory of content and other essays. Cambridge, Mass.: MIT Press. Fodor, J. (2008). LOT 2. The Language of Thought Revisited. Oxford: Clarendon Press. Freedman S. J. Clauser, J. F. (1972). Experimental test of local hidden-variable theories. Phys. Rev. Lett., 28: 939-941. French, P. Wettstein, H. Uehling T. E. (eds.) (1979). Midwest Studies in Philosophy, IV. Minneapolis: University of Minnesota Press. Gertler, B. (1999). A Defense of the Knowledge Argument. Philosophical Studies, 93: 317–336. Gnassounou, B. Kistler, M. (eds.) (2005). Causes, Pouvoirs, dispositions en philosophie. Le retour des vertues dormitives. Paris: Presses Universitaires de France. Goff, P. (2006). Experiences Don’t Sum. In Strawson, G. et al. (2006). Goldman, A., (1993). Consciousness, folk-psychology, and cognitive science. Consciousness and Cognition, 2: 364–382. Goodman, N. (1976). Languages of Art: An Approach to a Theory of Symbols. Indianapolis: Hackett Publishing Company. Gray, R. (2003). Tye’s Representationalism: Feeling the Heat? Philosophical Studies, 115: 245-56. Greenberger, E. D. M. Horne, M. A. Shimony, A. Zeilinger, A. (1990). Bell’s theorem without inequalities. Am. J. Phys., 58: 11311143. Greenstein, G. Zajonic, A. (1997). The Quantum Challenge, Modern Research on the Foundations of Quantum Mechanics. London: Jones and Bartlett Publishers. Gunderson, K. (ed) (1975). Language, Mind and Knowledge. Minnesota Studies in the Philosophy of Science, VII. Mpls: University of Minnesota press. Gunther, H. Y. (ed.) (2003). Essays on Nonconceptual Content.

221

Cambridge, Mass.: MIT Press. Harman, G. (1990). The Intrinsic Quality of Experience. Philosophical Perspectives, 4: 31-52. Hempel, C.G. (1969). Reduction: Ontological and Linguistic Facets. Philosophy, Science, and Method: 179–199. Heyting, A., (1956). Intuitionism: An Introduction. Amsterdam: North-Holland Publishing. Hofstadter, D. (1979). Gödel, Escher, Bach: an Eternal Golden Braid. New York: Basic Books. Horgan, T. Tienson, J. (2002). The Intentionality of Phenomenology and the Phenomenology of Intentionality. In Chalmers, D. (2002). Humphrey, N. (2006). Seeing red. Cambridge Mass.: Harvard University Press. Humphreys, P. (1997). How Properties Emerge. In Bedau, M. A. Humphreys, P. (eds.) (2007). Jackendoff, R. (1987). Consciousness and The Computational Mind. Cambridge, Mass.: MIT Press. Jackson, F. (1982). Epiphenomenal Qualia. Philosophical Quarterly, 32: 127–136. Jackson, F. (1986). What Mary Didn’t Know. Journal of Philosophy, 83: 291–295 Jacob, P. (1997). What Minds Can Do: Intentionality in a NonIntentional World. Cambridge: Cambridge University Press. James, W. (1890). Principles of Psychology, Vol. 1. London: Macmillan. Jeannerod, M. Jacob, P. (2005). Visual cognition: a new look at the two-visual systems model. Neuropsychologia, 43: 301-312. Kim, J. (1993). Supervenience and Mind. Cambridge: Cambridge University Press. Kim, J. (1998). Mind in a Physical World. Cambridge, Mass.: MIT Press. Kim, J. (2005). Physicalism, or something near enough. Princeton: Princeton University Press. Kistler, M. (2002). The Causal Criterion of Reality and the Necessity of Laws of Nature. Metaphysica, vol. 3, No. 1: 57-86.

222

Kistler, M. (2005). Necessary Laws. In Faye, J. Needham, P. Scheffler, U. Urchs, M. (eds.) (2005). Kistler, M. (2006a). La causalité comme transfert et dépendance nomique. In Philosophie, 89: 53-77. Kistler, M (2006b). Les causes des actions. In Le temps philosophique, Numéro spécial : L’action, 12: 141-175. Kistler, M. (2007). The Causal Efficacy of Macroscopic Dispositional Properties. In Kistler, M. Gnassounou, B. (eds.) (2007). Kistler, M. Gnassounou, B. (eds.) (2007). Dispositions and Causal Powers. Aldershot: Ashgate. Kistler, M. (Forthcoming.). La cognition entre réduction et émergence - Étude sur les niveaux de réalité. Kochen, S. Specker, E. P. (1967). The problem of hidden variables in quantum mechanics. J. Math. Mechanics, 17: 59-87. Kripke, S. (1972). Naming and necessity. Cambridge, Mass.: Harvard University Press. Levin, J. (2009). What is a phenomenal concept? In Alter, T. Walter, S. (eds.) (2009). Levine, J. (1983). Materialism and qualia: The explanatory gap. Pacific Philosophical Quarterly, 64: 354-361 Levine, J. (1988). Absent and Inverted Qualia Revisited. Mind and Language, 3: 271-87. Lewis, D. (1971). Analog and Digital. Noûs, 5: 321–327. Lewis, D. (1983). Postscript to “Mad Pain and Martian Pain”. Philosophical Papers, vol. 1. Oxford: Oxford University Press. Lewis, D. (1988). What Experience Teaches. Proceedings of the Russellian Society, University of Sydney, 13: 29–57. Libet, B. (2004). Mind and Time. Cambridge, Mass.: Harvard University Press. Locke, J. (1975) [1689]. An Essay Concerning Human Understanding. P. H. Nidditch (ed.). New York: Oxford University Press. Lycan, W. (1996). Consciousness and experience. Cambridge, Mass.: MIT Press. Macdonald, G. & Papineau, D. (2006). Teleosemantics. Oxford: Clarendon press.

223

McGinn, C. (1977). Charity, interpretation, and belief. Journal of Philosophy, 74: 521–535. McGinn, C. (1989). Can we solve the mind-body problem? Mind, 98: 349-366. McLaughlin, B. (1992). The Rise and Fall of British Emergentism. In Bedau, M. A. Humphreys, P. (eds.) (2007). McLaughlin, B. (1995). Varieties of Supervenience. In Savellos, E. and Yalcin, U. (eds). (1995). McLaughlin, B. (1997). Emergence and Supervenience. In Bedau, M. A. Humphreys, P. (eds.) (2007). Mele, A. R. (2009). Effective Intentions: The Power of Conscious Will. Oxford: Oxford University Press. Mermin, N. D. (1985). Is the moon there when nobody looks? Physics Today, vol. 38: 38 (April 1985). Mill, J. S. (1843). System of Logic. London: Longmans, Green, Reader, and Dyer. Millikan (1993). White Queen Psychology and Other Essays for Alice. Cambridge, Mass.: MIT Press. Nagel, E. (1961). The Structure of Science. London: Routledge and Kegan Paul. Nagel, T. (1974). What is it like to be a bat? Philosophical Review, 83: 435-456. Nagel, T. (1979). Panpsychism. In Nagel, T. Mortal Questions. Cambridge: Cambridge University Press. Nagel, T. (1986). The view from nowhere. Oxford: Oxford University Press. Nemirow, L. (2009). So This Is What It’s Like: A Defense of the Ability Hypothesis. In Alter, T. Walter, S. (eds.) (2009). Palmer, S. (1999). Color, Consciousness, and the Isomorphism Constraint. Behavioral and Brain Sciences, 22: 923-43. Papineau, D. (1998). Teleosemantics and Indeterminacy. Australasian Journal of Philosophy 76: 1–14. Papineau, D. (2002). Thinking about Consciousness. Oxford: Oxford University Press. Papineau, D. (2011). What Exactly is the Explanatory Gap? In Philosophia, Vol 39 (1): 5-19.

224

Parfit, D. (1986). Reasons and Persons. Oxford: Oxford University Press. Peacocke, C. (1992). A Study of Concepts, Cambridge Mass.: MIT Press. Penrose, R. Isham, C. (eds.) (1986). Quantum Concepts in Space and Time. Oxford: Clarendon Press. Penrose, R. (1994). Shadows of the Mind: A Search for the Missing Science of Consciousness. Oxford: Oxford University Press. Perry, J. (1979). The Problem of the Essential Indexical. Noûs, 13: 3-21. Perry, J. (2001). Knowledge, Possibility, and Consciousness, Cambridge, Mass.: MIT Press. Price, H. Corry, R. (eds.) (2007). Causation, Physics, and the Constitution of Reality. Oxford: Oxford University Press. Proust, J. (2007). Metacognition and metarepresentation: is a selfdirected theory of mind a precondition for metacognition? Synthèse, 159 (2): 271-295 Putnam, H. (1975). The meaning of ‘meaning’. In Gunderson, K. (ed) (1975). Putnam, H. (1988). Representation and Reality. Cambridge, Mass.: MIT Press. Quine, W. V. (1951). Two Dogmas of Empiricism. Reprinted in Quine, W. V. (1980). Quine, W. V. (1980). From a logical point of view. Cambridge, Mass.: Harvard University Press. Ramos Arenas, J. (2001). Las facetas de lo mental. Ideas y valores 117 (diciembre de 2001), 21-36. Recanati, F. (1997). Can We Believe What We Do Not Understand? Mind and Language 12 (1): 84-100. Rey, G. (1998). A Narrow Representationalist Account of Qualitative Experience. In Tomberlin, J. E. (ed.) (1998). Rosenthal, D. (2005). Consciousness and mind. Oxford: Oxford University Press. Rowlands, M. (2001). Consciousness and higher-order thoughts. Mind and Language, 16: 290–310. Russell, B. (1912a). On the notion of cause. Proceedings of the

225

Aristotelian Society XIII (1912-1913): 1-26. Russell, B. (1912b). The Problems of Philosophy. Oxford: Oxford University Press. Russell, B. (1927). The Analysis of Matter. London: Kegan Paul, Trench, Trubner; New York: Harcourt, Brace. Salmon, W. (1984). Scientific Explanation and the Causal Structure of the World. Princeton: Princeton University Press. Savellos, E. and Yalcin, U. (eds.) (1995). Supervenience: New Essays. Cambridge: Cambridge University Press. Schwitzgebel, E. (2008). The unreliability of naive introspection. Philosophical Review, 117: 245–273. Searle, J. (1992). The Rediscovery of the Mind. Cambridge, Mass.: MIT Press. Searle, J. (1995). The construction of social reality. New York: The Free Press. Searle, J. (1997). The mystery of consciousness. New York: The New York Review of Books. Searle, J. (2002). Consciousness and language. Cambridge: Cambridge University Press. Shapiro, L. A. (2004). The mind incarnate. Cambridge, Mass.: MIT Press Shimony, A. (1986). Events and Processes in the Quantum World. In Penrose, R. and Isham, C. (eds.) (1986). Shoemaker, S. (1982). The inverted Spectrum. Journal of Philosophy 79: 357-381. Sklar, L. (1974). Space, Time, and Spacetime. Berkeley: University of California Press. Skrbina, D. (ed.) (2009). Mind that Abides. Amsterdam: John Benjamins Publishing Company. Spence, C. & Deroy, O. (2012) Crossmodal correspondences: Innate or learned? I-Perception, 3: 316-318. Stoljar, D. (2006). Ignorance and Imagination. The Epistemic Origin of the Problem of Consciousness. Oxford: Oxford University Press. Strawson, G. (2006). Realistic Monism. In Strawson, G. et al. (2006).

226

Strawson, G. (2009). On the Sesmet Theory of Subjectivity. In Skrbina, D. (ed.) (2009). Strawson, G. et al. (2006). Consciousness and its place in nature. Exeter: Imprint Academic. Stubenberg, L. (1998). Consciousness and Qualia. Amsterdam: John Benjamins Publishing Company. Tomberlin, J. E. (ed.) (1998). Language, Mind, and Ontology Philosophical Perspectives, Vol. 12. Atascadero, CA: Ridgeview Publishing. Trenholme, R. (1994). Analog Simulation. Philosophy of Science, 61 (1): 115-131. Turing, A. M. (1936). On Computable Numbers, with an Application to the Entscheidungsproblem. Proceedings of the London Mathematical Society, 42: 230–265. (1937). Turing, A. M. (1950). Computing machinery and intelligence. Mind, 59: 433-460. Tye, M. (1995). Ten Problems of Consciousness. Cambridge, Mass.: Bradford Books. Tye, M. (2009). Consciousness Revisited. Cambridge, Mass.: MIT Press. Van Fraassen, B. (1980). The Scientific Image. Oxford: Oxford University Press. Wager, A. (1999). The Extra Qualia Problem: Synaesthesia and Representationism. Philosophical Psychology 12: 263-81. Williams, B. (1973). Problems of the Self. Cambridge: Cambridge University Press. Wittgenstein, L. (1953). Philosophical Investigations. Anscombe, G.E.M. Rhees, R. (eds.). Anscombe, G.E.M. (trans.) (1953). Oxford: Blackwell.