Introduction: Taking Consciousness Seriously. 1. Two Concepts of Mind I. FOUNDATIONS

Notes on David Chalmers The Conscious Mind (Oxford: Oxford UP, 1996) by Andrew Bailey, Philosophy Department, University of Guelph (abailey@uoguelph.ca) Introduction: Taking Consciousness Seriously... 2 I. Foundations... 2 1. Two Concepts of Mind... 2 2. Supervenience and Explanation... 3 a) Supervenience... 3 b) Reductive Explanation... 4 c) Logical Possibility and Necessity: Chalmers Two-Dimensionalism... 5 d) The Logical Supervenience of Almost Everything on the Physical... 7 II. The Irreducibility of Consciousness... 9 3. Can Consciousness be Reductively Explained?... 9 Argument 1: The logical possibility of zombies... 9 Argument 2: The inverted spectrum... 9 Argument 3: From epistemic asymmetry... 9 Argument 4: The knowledge argument... 9 Argument 5: From the absence of analysis...10 4. Naturalistic Dualism...11 a) An Argument Against Physicalism...11 b) Objections to the Argument...11 c) Other Arguments for Dualism...14 d) Is This Epiphenomenalism?...16 e) The Logical Geography of the Issues...18 IV. Applications...20 9. Strong Artificial Intelligence...20

2 Introduction: Taking Consciousness Seriously Three constraints: i) Take consciousness seriously (i.e. don t redefine the hard problem away). ii) Take science seriously (i.e. keep theories compatible with contemporary science). iii) Take consciousness to be a natural phenomenon, falling under the sway of natural laws (xiii) (i.e. assume there is some correct scientific theory of consciousness to be found). [T]he problem of consciousness may be a scientific problem that requires philosophical methods of understanding before we can get off the ground. Everything I say here is compatible with the results of contemporary science; our picture of the natural world is broadened, not overturned (xiv). I. FOUNDATIONS 1. Two Concepts of Mind Consciousness is the subjective quality of experience qualitative feels, or qualia. E.g. sensory experiences, pain and other bodily sensations, mental imagery, conscious thought, emotions, the sense of self. Ultimately one would like a theory of consciousness to do at least the following: it should give the conditions under which physical processes give rise to consciousness, and for those processes that give rise to consciousness, it should specify just what sort of experience is associated. And we would like the theory to explain how it arises, so that the emergence of consciousness seems intelligible rather than magical. In the end, we would like the theory to enable us to see consciousness as an integral part of the natural world (5). Chalmers begins by making a fundamental conceptual distinction between the phenomenal and the psychological concepts of mind : a) [T]he phenomenal concept of mind is the concept of mind as conscious experience, and of a mental state as a consciously experienced mental state (11). b) [T]he psychological concept of mind is the concept of mind as the causal or explanatory basis for behavior. A state is mental in this sense if it plays the right sort of causal role in the production of behavior, or at least plays an appropriate role in the explanation of behavior (11). Chalmers claims that both these notions of mind are correct (within its own domain), that they are complementary, and that they are (close to) exhaustive: neither is alone the whole story about mentality, but together they form a complete picture. (There are two sources of evidence which are responsible for generating the problems of consciousness, Chalmers says: third-person observations, which ultimately come down to behavior (21), and first-person observations which come down to experience (21). Once the psychological and phenomenal (and relational) facts about a person are fixed, then all the mental facts are also fixed there seems to be nothing mental that can be independently varied (21).) These two different conceptual frameworks should not be conflated: for example, Descartes (arguably) thought that all mental concepts are phenomenal, while modern functionalists think that all mental concepts are psychological for Chalmers, both views are equally wrong. We might find out that one of these conceptual frameworks can be reduced to, or assimilated into, the other but we cannot stipulate this at the start of the investigation. Most everyday mental concepts e.g. pain, perception, emotion, belief, learning have both phenomenal and psychological components. This (according to Chalmers) is because phenomenal and psychological properties tend to co-occur, and so for everyday purposes it is not important to carefully distinguish between them. However, for scientific purposes we need to make the distinction (which, Chalmers argues, is a perfectly coherent conceptual distinction despite the fact that we usually pick out phenomenal properties by their associated psychological property). In fact, the notion of consciousness itself has both phenomenal and psychological senses: Chalmers urges us to take care to focus on the phenomenal sense rather than the various psychological senses to explain consciousness by merely explaining psychological consciousness (which

3 Chalmers calls awareness) is a sleight of hand which fails to address the real problem. The division of mental properties into phenomenal and psychological properties has the effect of dividing the mind-body problem into two: an easy part and a hard part (24). According to Chalmers, the psychological aspects of the mind pose no deep philosophical questions (just technical puzzles ); the phenomenal aspects of the mind remain baffling. We have a good idea of how the physical gives rise to the psychological; [w]hat remains ill understood is the link between the psychological mind and the phenomenal mind (25). It is this psychological-phenomenal link that Chalmers sets out to explore as much a mind-mind problem as a mind-body problem. 2. Supervenience and Explanation a) Supervenience Chalmers arguments in The Conscious Mind are built, to a large degree, around the crucial notion of supervenience. Supervenience is a metaphysical notion which can be used to formulate relations of dependence between two different domains: in particular, it can be used to develop a theory of the precise way in which the phenomenal depends on the physical/psychological. B-properties supervene on A-properties if no two possible situations are identical with respect to their A- properties while differing in their B-properties. The key phrase to be cashed out here is possible situations : Local supervenience takes the situations to be individuals that is, it says that any two possible individuals that instantiate the same A-properties instantiate the same B-properties (33 34). Global supervenience takes situations to be entire universes (or worlds ) that is, it says that there are no two possible worlds identical with respect to their A-properties but differing with respect to their B-properties (34). Logical supervenience takes the possibility in question to be conceptual/logical possibility that is, it says that no two logically possible situations are identical with respect to their A-properties but distinct with respect to their B-properties (35). Something is logically possible, roughly, iff it is conceivable (says Chalmers). Natural supervenience takes the possibility in question to be nomic/natural possibility that is, it says that any two naturally possible situations with the same A-properties have the same B-properties (36). Something is naturally possible, roughly, iff it is consistent with the laws of nature that hold in the actual world. Local supervenience entails global supervenience, but not vice versa. Logical supervenience entails natural supervenience, but not vice versa. If B-properties supervene logically on A-properties, then once God (hypothetically) creates a world with certain A-facts, the B-facts come along for free as an automatic consequence. If B-properties merely supervene naturally on A-properties, however, then after making sure of the A-facts, God has to do more work in order to make sure of the B-facts: he has to make sure there is a law relating the A-facts and the B- facts (38). (There is a technical problem with the notion of logical supervenience that needs to be addressed: we want to say that, for example, biological properties are globally logically supervenient on (micro-)physical properties even though it is logically possible that there could be a world just like ours but containing extra entities such as ectoplasmic wombats with biological properties. Chalmers solution is to index the supervenience relation to our world, and to restrict supervenience claims to positive facts, those that cannot be negated

4 simply by enlarging a world (40). E.g. for any logically possible world W that is A-indiscernible from our world, then the [positive] B-facts true of our world are true of W (40).) It is important to notice the ontological significance of the difference between logical and (merely) natural supervenience: logically supervenient B-facts may be different facts [than the A-facts] but they are not further facts (41); by contrast, merely naturally supervenient B-facts are new facts about the world, over and above the A-facts. Hence, materialism/physicalism is true if all the positive facts about the world are globally logically supervenient on the physical facts (41), but false otherwise i.e., physicalism is true if all the positive facts about our world are logically entailed by the physical facts. b) Reductive Explanation A second notion which is very important for Chalmers framework is the notion of reductive explanation: that is, of completely explaining a higher-level phenomenon (such as life or reproduction) entirely in terms of simpler entities or processes. Chalmers argues that such reductive explanations are possible only if we have available a functional analysis of the phenomenon to be explained they operate precisely by showing how the lower-level processes perform the relevant functional tasks. For example, heat can be reductively explained by treating it as a causal-role concept and showing how roughly the motion of molecules plays the relevant causal role. Thus there are two projects in reductive explanation of a phenomenon. There is first a project of explication, where we clarify just what it is that needs to be explained, by means of analysis. Second, there is a project of explanation, where we see how that analysis comes to be satisfied by low-level facts. The first project is conceptual, and the second is empirical (51). The paradigm of reductive explanation via functional analysis works beautifully in most areas of cognitive science, at least in principle (48). E.g., we can give a cognitive model that shows how a certain system might support the right kind of causal processes for, for example, learning. However, [w]hatever functional account of human cognition we give, there is a further question: Why is this kind of functioning accompanied by consciousness? (47). Once we give an adequate reductive explanation of a functional property, it is logically impossible that something could instantiate the lower-level property but not the functional one; however, for any reductive explanation of a functional property, it is (apparently) always logically possible to ask whether it is accompanied by consciousness. That is, there is an explanatory gap. Hence, something is reductively explainable in terms of lower-level properties if and only if it is logically supervenient on those properties. Reductive explanation entails logical supervenience, because otherwise there would always be further unanswered questions. Logical supervenience entails reductive explanation, because, although this supervenience relation may not be an illuminating explanation, it will at least be a mystery-removing explanation: [i]t does this by reducing the bruteness and arbitrariness of the phenomenon in question to the bruteness and arbitrariness of lower-level processes. [I]t at least eliminates any sense that there is something extra going on (49). (In fact, Chalmers argues, mystery-removing explanations are often also as it happens illuminating, because micro-physical explanations generally have the virtues of autonomy and simplicity.) (A few things to note about reductive explanation: a) Reductive explanation need not involve the reduction of a higher-level entity or theory with a lowerlevel: for example, it is compatible with multiple realizability, as long as each instance of the higherlevel phenomenon has a lower-level explanation. b) Reductive explanation is not the only kind of explanation for example, there are also historical, high-level or teleological explanations. c) Practical reductive explanations need not go all the way to the microphysical level and in principle there might be higher-level reductions (e.g. from the C-level to the B-level) without a further reduction to the bottom level even being available. d) Reductive explanations will often be global rather than local.)

5 c) Logical Possibility and Necessity: Chalmers Two-Dimensionalism Logical necessity is a third key notion for Chalmers, and it is essential that we accept his version of this notion for his subsequent arguments to work. The two basic (non-formal) ways to cash out logical necessity are: i) True in all possible worlds. ii) True in virtue of meaning conceptually true. The general notion of non-formal conceptual truth has been under attack since the second half of the last century, and Chalmers sets out to defend (his usage of) it. Attack 1: Most concepts do not have definitions giving necessary and sufficient conditions (52). But this is not a problem, Chalmers argues: one set of facts can entail another set without there being a clean definition of the latter notions in terms of the former. Given that A-facts can entail B-facts without a definition of B-facts in terms of A-facts, the notion of logical supervenience is unaffected by the absence of definitions (54). Attack 2: [P]urported conceptual truths are always subject to revision in the face of sufficient empirical evidence (55). But, Chalmers claims, this is not true of (global) supervenience relations: The facts specified in the antecedent of this conditional effectively include all relevant empirical factors. Empirical evidence could show us that the antecedent of the conditional is false, but not that the conditional is false (55). Attack 3: [T]here is a large class of necessarily true statements whose truth is not knowable a priori. At various points in this book, I use a priori methods to gain insight into necessity; this is the sort of thing that [this claim] is often taken to challenge (56). Chalmers attempts to meet this challenge by describing his theory of two-dimensionalism. A concept determines a function from possible worlds to referents: this is called the intension of this concept (and Chalmers equates this with the concept s meaning, though this is not essential to his account). When a particular possible world is specified (e.g. the actual world), this function determines an extension for the concept: that is, simply, it determines what things the concept is true of in that particular world. Chalmers argues that concepts in fact have, not just one, but two intensions: Primary intension: a function from worlds to extensions reflecting the way that actual-world reference is fixed (57). That is, for a given possible world, the primary intension of a concept picks out what the reference of the concept would be if the world turned out to be actual: if we had lived in a world where water was XYZ, water would refer to XYZ; but since in our world water is H 2 O, water refers to H 2 O. Chalmers puts this by saying that the primary intension of water picks out the watery stuff in the actual world, whatever that stuff turns out to be. Secondary intension: a function from counterfactual worlds to extensions, given that the actual world of the thinker is already fixed. That is, given that water has turned out to be H 2 O in the actual world, the secondary intension of water picks out H 2 O in every counterfactual world. [T]he secondary intension is determined by first evaluating the primary intension at the actual world, and then rigidifying this evaluation so that the same sort of thing is picked out in all possible worlds (59). A good way to understand this is to say that primary intensions pick out the reference of a concept in a world when it is considered as actual that is, when it is considered as a candidate for the actual world of the thinker whereas the secondary intension picks out the referent of a concept in a world when it is considered as counterfactual, given that the actual world of the thinker is already fixed (60). The key idea in all this for Chalmers is that: a) The primary intension of a concept, unlike the secondary intension, is independent of empirical [a posteriori] factors (57). We can characterise the primary intension of a concept a priori, by considering

6 our semantic intuitions under various counterfactual circumstances. b) It is the primary intension of a concept that is most central for my purposes: for a concept of a natural phenomenon, it is the primary intension that captures what needs explaining (57). (Chalmers also suggests that it is the primary content of my thoughts which governs their cognitive and rational relations (65).) This is the essence of his answer to Kripke. Chalmers semantic two-dimensionalism gives rise to a pair of notions of conceptual necessity: 1-necessity: truth evaluated in a possible world according to the primary intension of a concept. Chalmers also borrows the term deep necessity for this, since it is unaffected by a posteriori considerations. 2-necessity: truth evaluated in a possible world according to the secondary intension of a concept. Chalmers also borrows the term superficial necessity for this. A key idea for him here is that [t]hese two varieties of possibility and necessity apply always to statements. There is only one relevant kind of possibility of worlds; the two approaches differ on how the truth of a statement is evaluated in a world (63). For Chalmers, the notion of a logically possible world is something of a primitive (66) roughly, a world that (a) God might have created. Two-dimensionalism also entails that sentences have two associated propositions: a primary proposition determined by the truth conditions associated with the primary intension; and a secondary proposition determined by the truth conditions associated with the secondary intension. From this it follows that [a] statement is necessarily true in the first (a priori) sense if the associated primary proposition holds in all centered possible worlds (that is, if the statement would turn out to express a truth in any context of utterance). A statement is necessarily true in the a posteriori sense if the associated secondary proposition holds in all possible worlds (that is, if the statement as uttered in the actual world is true in all counterfactual worlds) (64). What is the relation between conceivability and possibility? In effect, there are two varieties of conceivability, which we might call 1-conceivability and 2-conceivability, depending on whether we evaluate a statement in a conceivable world according to the primary or the secondary intentions of the terms involved (67). For Chalmers, 1-conceivability implies 1-possibility and 2-conceivability implies 2-possibility (67). (1- conceivability, of course, does not entail 2-possibility.) Chalmers is careful to specify that a statement is conceivable (or conceivably true) if it is true in some conceivable world (67) that is, it requires two things, the conceivability of a relevant world and the truth of the statement in that world. One must take care to properly describe the world that one is conceiving, or things may seem conceivable which aren t. It follows from all this that the oft-cited distinction between logical and metaphysical possibility stemming from the Kripkean cases on which it is held to be logically possible but not metaphysically possible that water is XYZ is not a distinction at the level of worlds, but at most a distinction at the level of statements. A statement is logically possible in this sense if it is true in some world when evaluated according to primary intensions; a statement is metaphysically possible if it is true in some world when evaluated according to secondary intensions. The relevant space of worlds is the same in both cases. Most importantly, none of the cases we have seen give reason to believe that any conceivable worlds are impossible. So there seems no reason to deny that conceivability of a world implies possibility (67 68). Hence conceivability implies logical possibility, according to Chalmers. Does logical possibility imply conceivability? Chalmers says yes, as long as we understand conceivability as conceivability-in-principle perhaps conceivability by a superbeing (68). Modality is not epistemically inaccessible: the [logical] possibility of a statement is a function of the intensions involved and the space of possible worlds, both of which are epistemically accessible in principle, and neither of which is dependent on a posteriori facts. The class of 1-necessary truths corresponds directly

7 to the class of a priori truths (68 69). As one might expect, the two-dimensional framework produces two versions of logical supervenience: supervenience according to primary intensions (i.e. by 1-necessity), supervenience according to primary intensions (i.e. by 2-necessity). The latter is often called metaphysical supervenience, but Chalmers has argued that it is really a species of logical supervenience. Chalmers argues that the primary version of supervenience is the more central kind, especially when considering questions of explanation: it is the primary intension of a concept that determines whether or not an explanation is satisfactory (69). From all of this, it follows that B-properties are logically supervenient on A-properties if for any actual situation X, the A-facts about X entail the B-facts about X, and furthermore this implication from A-facts to B-facts will be a priori (70) since it will depend only on the intensions of the concepts involved (plus knowledge of the A-facts) i.e., one might say (though Chalmers does not), it will be an analytic truth. There are therefore at least three avenues to establishing claims of logical supervenience: these involve conceivability, epistemology, and analysis. To establish that B-properties logically supervene on A-properties, we can (1) argue that instantiation of A-properties without instantiation of the B-properties is inconceivable; (2) argue that someone in possession of the A-facts could come to know the B-facts ; or (3) analyze the intensions of the B-properties in sufficient detail that it becomes clear that B-facts follow from A-facts in virtue of these intensions alone (70). d) The Logical Supervenience of Almost Everything on the Physical All high-level facts about natural phenomena are logically supervenient on the totality of physical facts, Chalmers argues, except those about consciousness. That is, physicalism is true in all (natural) domains except for the mental. Chalmers argues for this in three ways: It is in principle inconceivable that a world could be identical to ours in every microphysical detail but different biologically / architecturally / astronomically / behaviourally / chemically / economically / sociologically etc. etc. When high-level facts fail to logically supervene on the physical facts, there are deep epistemological problems: it would mean that the externally observable facts which, Chalmers apparently takes it, are fixed by the physical could be just as they are, but the high-level facts different, and hence that no data could suffice to tell us which of the various possible, say, biologies is the actual one. Since there is no deep epistemic problem of biology (etc.), biological facts must logically supervene on the physical. (Chalmers notes that there are some deep epistemic problems to do with physical regularities e.g. causation but he sidesteps these by including a specification of these regularities in the supervenience base.) Furthermore, Chalmers says it is plausible to think that a sufficiently intelligent super-being should be able to simply read off high level facts from the microphysical facts. A proper analysis of the intensions concepts of high level properties will show them to be satisfied by fixing the microphysical facts in a particular way. This is because most high-level intensions can be characterized either structurally or functionally (or both). For example, Chalmers claims that life or economic prosperity are functional properties. Chalmers then goes on to consider some problem cases: Consciousness-dependent properties Intentionality Moral and aesthetic properties Names Indexicals Negative facts

8 Physical laws and causation None of these, he argues, presents any reason to think that any natural phenomenon but consciousness fails to supervene on the physical. The position we are left with is that facts about the world are exhausted by (1) particular physical facts, (2) facts about conscious experience, (3) laws of nature, (4) a second-order That s all fact, and perhaps (5) an indexical fact about my location. Modulo conscious experience and indexicality, it seems that all positive facts are logically supervenient on the physical. It is plausible that every supervenience relation of a highlevel property upon the physical is ultimately either (1) a logical supervenience relation of either the primary or the secondary variety, or (2) a contingent natural supervenience relation. If neither of these holds for some apparent supervenience relation, then we have good reason to think that there are no objective high-level facts of the kind in question (87 88).

9 II. THE IRREDUCIBILITY OF CONSCIOUSNESS 3. Can Consciousness be Reductively Explained? Chalmers argues in this chapter that consciousness cannot be explained in physical terms since it does not logically supervene on the physical. Chalmers gives five arguments for this conclusion: two arguments from conceivability, two arguments from epistemology, and an argument from analysis. Argument 1: The logical possibility of zombies A zombie can be defined as a creature that is molecule for molecule identical to me, and identical in all the low-level properties postulated by a completed physics, but [that] lacks conscious experience entirely (94). Chalmers argues that zombies are conceivable, and from this it follows (given what he has laid out in the previous chapter) that zombies are logically possible and hence that consciousness does not logically supervene on the physical. Chalmers suggests that something is conceivable if it is conceptually coherent, by which he apparently means that the intensions of the concepts involved do not lead to any contradiction (96). Since, he claims, there is nothing conceptually contradictory about zombies, they are conceivable. He also argues indirectly that zombies are conceivable by suggesting that it is highly plausible that non-standard realizations of his functional organization might lack consciousness, and arguing that from this it follows that his zombie twin is an equally coherent possibility. In making moves of this sort, Chalmers assumes that a) biochemistry and b) physical complexity are conceptually irrelevant to consciousness: that is, adding in these sorts of details about the implementations of a functional architecture cannot introduce the logical, a priori entailment of consciousness. So the only route available to an opponent here is to claim that in describing the zombie world as a zombie world, we are misapplying the concepts, and that in fact there is a conceptual contradiction lurking in the description. Perhaps if we thought about it clearly enough we would realize that by imagining a physically identical world we are thereby automatically imagining a world in which there is conscious experience. But then the burden is on the opponent to give us some idea of where the contradiction might lie in the apparently quite coherent description. If no internal incoherence can be revealed, then there is a very strong case that the zombie world is logically possible (99). Argument 2: The inverted spectrum If it is conceivable that there could be a being physically identical to me but with inverted conscious experiences (e.g. where two of the axes of the three-dimensional colour space are switched) then it is possible for facts about consciousness to vary independently of the physical facts, and so consciousness does not logically supervene on the physical. Even if the human colour space is asymmetrical in such a way that it could not be inverted, we could imagine mapping them onto colours in a more extended colour space (e.g. red might be mapped onto a warm version of blue), or we could simply imagine creatures with a symmetrical colour space and then conceive of their colour experiences being inverted. Argument 3: From epistemic asymmetry Eliminativism about conscious experience is an unreasonable position only because of our own acquaintance with it. [T]here is an epistemic asymmetry in our knowledge of consciousness that is not present in our knowledge of other phenomena (102). Since, if consciousness were logically supervenient on the physical there would be no epistemic asymmetry, it follows (according to Chalmers) that consciousness is not logically supervenient. Argument 4: The knowledge argument Mary has been brought up in a black-and-white room and has never seen any colors except for black, white,

10 and shades of gray. She is nevertheless one of the world s leading neuroscientists, specializing in the neurophysiology of color vision. She knows everything there is to know about the neural processes involved in visual information processing, about the physics of optical processes, and about the physical makeup of objects in the environment. But she does not know what it is like to see red. No amount of reasoning from the physical facts alone will give her this knowledge (103). If this is a correct description of a possible situation, Chalmers argues, it follows that the facts about consciousness are not entailed by the physical facts, and so consciousness cannot logically supervene on the physical. That is, knowledge of what red is like is factual knowledge that is not entailed a priori by knowledge of the physical facts (104) that is enough, according to Chalmers, to show that consciousness cannot be reductively explained. Another way of making the same point is to note that knowing all the physical facts about another organism or system such as a bat or a complex computer will not tell us what the consciousness of that creature is like (if anything). Argument 5: From the absence of analysis For consciousness to be entailed by a set of physical facts, one would need some kind of analysis of the notion of consciousness the kind of analysis whose satisfaction physical facts could imply and there is no such analysis to be had (104). In particular, functional analyses of the concept of consciousness must all fail (according to Chalmers) since [a]lthough conscious states may play various causal roles, they are not defined by their causal roles. Rather, what makes them conscious is that they have a certain phenomenal feel, and this feel is not something that can be functionally defined away (105). Chalmers gives two further arguments against functional analysis of the concept of consciousness: 1) [A]ny functionally analyzed concept will have a degree of semantic indeterminacy (105), says Chalmers. But facts about the presence of conscious are determinate, and not subject to semantic stipulation. Hence consciousness cannot the functionally analyzed. 2) Any functionalist analysis will collapse the legitimate (according to Chalmers) conceptual distinction between consciousness and awareness. The prospects for a structural analysis of the concept of consciousness look even worse. Hence, according to Chalmers, [i]t seems that the concept of consciousness is irreducible, being characterizable only in terms of concepts that themselves irreducibly involve consciousness (106). The failure of consciousness to logically supervene on the physical tells us that no reductive explanation of consciousness can succeed (106). There is an unbridgeable explanatory gap between the physical and conscious experience the fact that consciousness accompanies a given physical process is a further fact, not explainable simply by telling the story about the physical facts (107). The best kind of explanation we can hope for in principle is a non-reductive one, e.g. one appealing to additional bridge principles. Chalmers then responds to five objections: Are we setting the standards too high? Couldn t a vitalist have said the same thing about life? Is conceivability a guide to possibility? Isn t this a collection of circular intuitions? Doesn t all explanation have to stop somewhere? Finally, Chalmers sets out to illustrate the failure of reductive explanation by giving a critique of a number of accounts of consciousness that have been proposed by researchers in various disciplines (111). a) Cognitive modelling (Baars, Dennett). The best that these models can do, according to Chalmers, is give a theory of awareness and show how it is correlated with consciousness. b) Neurobiological explanation (Crick and Koch, Edelman). Again, these theories at best establish

11 correlations with consciousness. c) The appeal to new physics (Penrose and Hameroff). Even new physics comes down to structure and dynamics (function). d) Evolutionary explanation. The process of natural selection cannot distinguish between me and my zombie twin (120). A key point, for Chalmers, is that it s not just that we don t have a reductive explanation of consciousness yet; he takes his arguments to prove that we could never have a reductive explanation. Any account given in purely physical terms will suffer from the same problem. It will ultimately be given in terms of the structure and dynamical properties of physical processes, and no matter how sophisticated such an account is, it will yield only more structure and dynamics. While this is enough to handle most natural phenomena, the problem of consciousness goes beyond any problem about the explanation of structure and function, so a new sort of explanation is needed (121). 4. Naturalistic Dualism a) An Argument Against Physicalism Chapter 3 argued that consciousness cannot be explained by physical theories (an epistemological claim); this chapter argues that consciousness is not itself physical (an ontological claim). The basic argument for this goes as follows: 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 (123) That is, the failure of the logical supervenience of consciousness on the physical is enough to establish the falsity of physicalism, and hence establishes a version of dualism. Which version of dualism? Chalmers calls it naturalistic dualism, and its key tenet is that, though consciousness fails to supervene logically on the physical, it supervenes naturally on the physical: consciousness arises from a physical basis, even though it is not entailed by that basis (125). Naturalistic dualism is a version of property (rather than substance) dualism: there are properties of individuals in this world the phenomenal properties that are ontologically independent of physical properties (125). All physical events, including human behaviour, can be explained entirely in physical terms. Chalmers denies that naturalistic dualism is antiscientific or supernatural. Physical theories derive higher-level phenomena from fundamental physical features and laws. Similarly, a theory of consciousness will either treat consciousness as a basic feature of the world (like space-time, spin, charge, etc.) or will postulate a class of non-physical, more fundamental properties protophenomenal properties from which consciousness can be derived. There will also have to be a new class of fundamental laws, but these laws will not compete with the closed system of physical laws instead they will be psychophysical supervenience laws. Once we have a fundamental theory of consciousness to accompany a fundamental theory in physics, we may truly have a theory of everything. On this view, the world still consists in a network of fundamental properties related by basic laws, and everything is to be ultimately explained in these terms. Further, nothing about this theory contradicts anything in physical theory; rather it supplements that theory. [T]o embrace dualism is not necessarily to embrace mystery (127 128). b) Objections to the Argument Objections can be made to either premise 2 or premise 3 of Chalmers argument for dualism. Chalmers claims to have dealt with objections to the second premise in Chapter 3. In this chapter he deals with

12 objections to the third premise, which moves from the failure of logical supervenience to the falsity of materialism (129). Objection 1: Though consciousness may be emergent, this is still compatible with the truth of physicalism. Chalmers distinguishes between two kinds of emergence, weak and strong: properties may be emergent in the weak sense in that they are not obvious consequences of low-level laws (129), and this does not conflict with physicalism. But Chalmers has argued that consciousness is emergent in the much stronger sense that it is not logically supervenient on the low-level (physical) facts. Objection 2: [C]onsciousness and the physical might be two aspects of the same thing [and hence] consciousness might in a sense be physical (130). Chalmers responds that, as long as the conscious aspect is not entailed by the physical aspect, then this duality of aspects gives us a kind of property dualism (130) even if the two aspects of reality can be subsumed under a grander monism, it will not be a monism of the physical alone. Objection 3: To hold that property A is merely naturally supervenient on (caused by) property B is not to adopt a dualism of properties for example, suggests Searle, being liquid is merely naturally supervenient on (caused by) being H 2 O, but no one is a dualist about liquidity. Chalmers responds that this is simply a false analogy: the microphysical features of water do not cause liquidity, they constitute it. Liquidity does logically and not merely naturally supervene on the low-level, physical facts about water. The notion of a nonliquid replica of a batch of liquid H 2 O is simply incoherent. Consciousness is ontologically novel in a much more significant way than liquidity (130). Objection 4: Chalmers argument resembles that of Descartes, and Descartes argument was flawed: just because one can imagine that A and B are not identical, it does not follow that A and B are not identical (130). Chalmers responds that [i]t is crucial that the argument as I have put it does not turn on questions of identity but of supervenience (131) he claims that the argument rephrased in terms of the supervenience of facts about consciousness on the totality of physical facts can escape Descartes s fallacy. The following objections take this objection further. Objection 5: Chalmers argument only establishes that zombie worlds are logically possible; however, to refute physicalism he needs to prove that zombie worlds are metaphysically possible, and [w]hereas conceptual coherence suffices for logical possibility, metaphysical possibility is more constrained. The point is also often made by suggesting that there is a difference between conceivability and true possibility (131). For example, Kripke s Naming and Necessity demonstrates the existence of necessary truths whose necessity is only knowable a posteriori; similarly, perhaps the facts about consciousness follow necessarily from the physical facts but this can only be established a posteriori hence there can be a conceptual gap without a metaphysical gap (131). Chalmers begins his response to this objection by appealing to the notion of logical possibility he defined in Chapter 2: Logical possibility comes down to the possible truth of a statement when evaluated according to the primary intensions involved. The primary intensions of water and H 2 O differ, so it is logically possible in this sense that water is not H 2 O. Metaphysical possibility comes down to the possible truth of a statement when evaluated according to the secondary intensions involved. The secondary intensions of water and H 2 O are the same, so it is metaphysically necessary that water is H 2 O (132). If Chalmers is right about this, then the objection turns into the following: even though the primary intensions of phenomenal concepts differ from those of any physical concepts, their secondary intensions may be the same (for all Chalmers has shown). If so, then phenomenal and physical/functional concepts may pick out the same properties a posteriori despite the a priori distinction (132). Chalmers answers this reformulated objection in the following way: whether or not the primary and secondary intensions coincide, the primary intension determines a perfectly good [center-relative] property of objects in possible worlds. The property of being watery stuff is a perfectly reasonable property, even though it is not the same as the property of being H 2 O. If we can show that there are possible worlds that are

13 physically identical to ours but in which the property introduced by the primary intension is lacking, then dualism will follow. This is just what has been done for consciousness. By analogy, if we could show that there were worlds physically identical to ours in which there was no watery stuff, we would have established dualism about water just as well as if we had established that there were worlds physically identical to ours in which there was no H 2 O. And importantly, the difference with respect to the primary intension can be established independently of a posteriori factors, so that considerations about a posteriori necessity are irrelevant (132 133). The most general way to make the point is to note that nothing about Kripke s a posteriori necessity renders any logically possible worlds impossible. It simply tells us that some of them are misdescribed, because we are applying terms according to their primary intensions rather than the more appropriate secondary intensions. So although there may be two kinds of possibility of statements there is only one relevant kind of possibility of worlds. It follows that if there is a conceivable world that is physically identical to ours but which lacks certain positive features of our world, then no considerations about the designation of terms such as consciousness can do anything to rule out the metaphysical possibility of the world (134). (Furthermore, Chalmers notes, for consciousness the primary and secondary intensions coincide: what it is to have consciousness in either sense is to have a phenomenal feel. Unlike water, there is no gap between the appearance of consciousness and the underlying reality. But this point is controversial, and he does not depend on it.) Objection 6: [T]o claim that the zombie world is physically identical to ours is to misdescribe it. Just as the XYZ world seems to contain water but does not, the zombie world seems physically identical while being physically different (134 135). This might be so if there were properties essential to the physical constitution of the world that are not accessible to physical investigation. In conceiving of a physically identical world, we are really only conceiving of a world that is identical from the standpoint of physical investigation, while differing in the inaccessible essential properties, which are also the properties that guarantee consciousness (135). E.g. something might not be an electron unless it not only played the extrinsic role of an electron but also had its intrinsic nature, and the intrinsic nature of matter might be phenomenal or protophenomenal. This objection turns in part on a purely conceptual question: do phenomenal predicates work in the way suggested above? Chalmers suggests that this is highly implausible. Furthermore, even if this objection holds the difference between this view and the property dualism that I have advocated is small. It remains the case that the world has phenomenal properties that are not fixed by the properties that physics reveals (136). Objection 7: It could be held that there is a modality of metaphysical possibility that is distinct from and more constrained than logical possibility, and that arises for reasons independent of the Kripkean considerations. We can call this hypothesized modality strong metaphysical necessity, as opposed to the weak metaphysical necessity introduced by the Kripkean framework. On this view, there are worlds that are entirely conceivable, even according to the strongest strictures on conceivability, but which are not possible at all (136 137). Chalmers short answer to this objection is that there is no reason to believe that such a modality exists,, and the constraints it imposes on the space of possible worlds would be brute and inexplicable (137). Indeed, if some worlds are logically possible but metaphysically impossible, it seems that we could never know it. By assumption, the information is not available a priori, and a posteriori information only tells us about our world. This can serve to locate our world in the space of possible worlds, but it is hard to see how it could give information about the extent of that space (137). Furthermore, strong metaphysical necessity saves materialism only at he cost of making it entirely mysterious how consciousness could be physical (138). On this view, it would still be the case that the existence of consciousness cannot be derived from physical knowledge, so that consciousness cannot be reductively explained. And it would remain the case that we would need certain primitive connecting principles to explain the supervenience of the phenomenal on the physical. The only difference between the views is that the relevant psychophysical principles are deemed to be brute laws of necessity rather than laws of nature (138). Objection 8: The apparent conceivability of zombie worlds arises from some sort of impaired rationality, so

14 that if we were only more intelligent we would see that the description of the world is not coherent at all (139). Perhaps some mathematical truths are necessarily true even though they are not knowable a priori by us, so by analogy this might be true of some truths about consciousness. Chalmers responds that this analogy fails because [i]n the mathematical case, our modal reasoning leaves the matter open; our conceivability intuitions do not tell us anything one way or the other. There may be some weak sense in which it is conceivable that the statements are false for example, they are false for all we know but this is not a sense that delivers a conceivable world where they fail. In the zombie case, by contrast, the matter is not left open: there seems to be a clearly conceivable world in which the implication is false (139). Nothing in the mathematical analogy shows us how to rule out apparently conceivable worlds. While it must be conceded that any philosophical argument could go wrong because of cognitive impairment, in the absence of any substantial reason to believe this, this sort of objection seems quite ad hoc (139 140). c) Other Arguments for Dualism Jackson s Knowledge Argument: As a direct argument against materialism Jackson s argument is often seen as vulnerable due to its use of the intensional notion of knowledge (140). Chalmers argues these attacks fail. i) Many opponents to the knowledge argument (Churchland, Horgan, Lycan, McMullen, Papineau, Teller, Tye) have argued that, although Mary does gain new knowledge, this knowledge does not correspond to a new fact she simply comes to know an old fact in a new way (under a new mode of presentation). Chalmers responds that [t]hese gaps arise precisely because of the difference between primary and secondary intensions. This objection therefore comes to precisely the same thing as the objection from the distinction between logical necessity and (Kripkean) metaphysical necessity discussed earlier, and the discussion there of primary and secondary intensions is sufficient to refute it (141). Suppose that a is G and b is G are the same fact, but that this cannot be seen a priori: this must be because a=b and the secondary intensions are the same but the primary intensions differ. But when Mary learns that a=b then there must be some truly novel fact that she gains knowledge of. In particular, she must come to know a new fact involving that mode of presentation (142). That is, Mary learns a new fact that connects the two modes of presentation. For example, suppose that a is dthat(p) and b is dthat(q); if Mary knows that a is G but doesn t know that b is G, then Mary cannot know that something is both P and Q. It is this latter, conjunctive fact which is the genuinely novel one. (After all, Chalmers suggests, [e]ven when interpreted according to secondary intensions, there will be a possible world in which a is [G] but in which nothing is both P and Q (142) presumably this is because, though a=b by (metaphysical) necessity, their modes of presentation are contingent.) For example, [i]f one knows that Hesperus is visible but not that Phosphorus is visible (because one does not know that Hesperus is Phosphorus), then one does not know that one object is both the brightest star in the morning sky and the brightest star in the evening sky. This is a separate fact that one lacks knowledge of entirely (141). ii) Brian Loar tries to advance beyond objection i). Loar recognizes that analogies with the usual examples cannot do the job for the materialist, as such analogies allow that physical and phenomenal notions have distinct primary intensions, and the antimaterialist can simply apply the argument to the property corresponding to the primary intension (142). He therefore argues that two predicates can share the same primary intension even when this sameness is not knowable a priori. But, Chalmers argues, this could be so [o]nly if the space of possible worlds is smaller than we would have thought a priori we must think that the intensions differ because there is a possible world in which the two predicates have different reference, but in fact there must be no such possible world. But this requires a notion of brute strong metaphysical necessity, and Chalmers has already argued against this (and Loar does not defend it explicitly). iii) One could argue that Mary knows all the physical facts but lacks knowledge about the hidden (proto)phenomenal essences of physical entities. But this more or less concedes the argument to Jackson.