Dette paper er publiceret på www.hum.au.dk/filosofi/wittgenstein. Det indgik på konferencen: "Sprog, verden, livsform arrangeret af The Wittgenstein Network, 4. og 5. april 2003. Offentliggørelse må ikke finde sted uden forfatterens tilladelse. Private Language and Mind-Body Dualism Jesper Kallestrup This paper argues for the conditional claim that some conceivability arguments against physicalism rely on a notion of qualia that entails the conceptual possibility of a private language in Wittgenstein s sense. I. Physicalism and Qualia Physicalism says that all the facts, including all the phenomenal facts, are metaphysically necessitated by the physical facts. If physicalism is true, it is metaphysically necessary that if two worlds are physically identical, they are identical in every other respect, i.e. no two worlds can differ in some respects without differing in their physical respects: (P) Physicalism is true at a possible world w iff any world which is a minimal physical duplicate of w is a duplicate of w simpliciter, where a 'minimal physical duplicate of w' means a world which is physically identical to w and does not contain anything else. Without this stop-clause, the conceptual possibility of causally inert non-physical substances would be ruled out. 1 It follows that if physicalism is true at the actual world, there is no metaphysically possible world, which is (minimally) physically identical to the actual world, but different in some other respect. Note that (P) does not claim that physicalism is necessarily true. The thesis is true at our world only if it meets the conditions specified on the right-hand side, so if our world does not, the thesis is false as a matter of fact. Note also that we need to define physicalism in terms of metaphysical necessity, because everyone can, and should, allow for law-like correlations between the mental and the physical, and so can, and should, accept supervenience with nomological necessity. 1 Cf. Jackson [1994]. 1
Many philosophers have believed, and still believe, that physicalism is false. A repeated line of argument is the so-called conceivability argument, which has the following general shape: a certain scenario is conceivable, and whatever is conceivable is metaphysically possible, so this scenario is metaphysically possible. But if this scenario is metaphysicaly possible, physicalism is false, so physicalism is false. In the present context, the most famous conceivability argument is due to Descartes (Meditation VI, 9]: I know that all which I clearly and distinctly conceive can be produced by God exactly as I conceive it although I certainly do possess a body with which I am very closely conjoined; nevertheless, because, on the one hand, I have a clear and distinct idea of myself, in as far as I am only a thinking and unextended thing, and as, on the other hand, I possess a distinct idea of body, in as far as it is only an extended and unthinking thing, it is certain that I am entirely and truly distinct from my body, and may exist without it. So, it is conceivable that I exist without my body given that my body and I have distinct essential properties, and whatever is conceivable is really possible by devine interference, so it is possible that I exist without my body. But if it is possible that I exist without my body, I am not strictly identical to my body, although I am closely connected to it. But one type of physicalism, namely reductive physicalism says that I am identical to my body (or brain), so this view is false. Descartes took this to be an argument for substance dualism, the view that each of us is composed of two distinct existing kinds of stuff. My body is an extended, dividable, non-thinking, transient, public substance that is subject to the laws of physics, whereas my soul is an unextended, individable, thinking, eternal, public substance that is not subject to the laws of physics. I am contingently identical to the former, but necessarily identical to the latter. Most modern proponents of anti-physicalism, however, do not believe in the existence of mental substances. They think that all substances have physical properties, hence that all substances are physical substances. 2 What they hold is merely that some physical substances have some mental properties that neither are identical to nor metaphysically supervene on physical properties. Call this view strong 2 A mental substance is presumably a substance with only mental properties. 2
property dualism 3. The kind of properties that the strong property dualists believe are the best candidates for failure of psycho-physical metaphysical supervenience are phenomenal properties. These are the properties that involve qualia: the characteristic qualitative feel associated with phenomenal states, the intrinsic character of experience, the how-it-is-like aspect of phenomenal consciousness. To have knowledge of phenomenal properties is, on this view, to have irreducible, propositional, phenomenal information. No amount of physical information will enable one to derive phenomenal information. There are two things possessing this information is not. 4 First, to know what-it-is-like is not to know what it resembles. I cannot know what a particular experience is like unless I have had it, but I can know what it resembles if I have had a qualitatively similar experience. Second, to know what it is like is not to know what experience one has. One can know what experience one has when one has an experience of seeing red without having had it, because there are ways of referring to it that involves no such experience, e.g. the experience one has when, under normal circumstances, one looks at fire engines. Strong property dualists also deploy our purported rational access to real modalities to prove physicalism wrong: it is conceivable on ideal reflection that there be worlds physically identical to our world in which all qualia are inverted or absent, but what is thus conceivable is metaphysically possible, so there are metaphysically possible absent qualia worlds and zombie worlds that are physically identical to the actual world. Thus Chalmers [1996, p. 97] has an argument why there are conceptually possible zombie-worlds that rests on multiple realisability of functional organisation: given that it is conceptually coherent that the group-mind set-up or my silicon isomorph could lack conscious experience, it follows that my zombie twin is an equally coherent possibility. For it is clear that there no more of a conceptual entailment from biochemistry to consciousness than there is from silicon or from a group of homunculi. If the silicon isomorph without conscious experience is conceivable, we need only 3 See Chalmers [1996]. Strong property dualism contrasts with weak property dualism which accepts metaphysical supervenience but not type-type identity. 4 Cf. Lewis [1999]. Instead of the hypothesis of phenomenal information, Lewis advocates the well-known ability hypothesis according to which knowing what an experience is like, amounts to no more that possessing abilities to imagine, recognise and remember it. It is knowing-how, not knowing-that. 3
substitute neurons for silicon in the conception while leaving functional organisation constant, and we have a zombie twin. Nothing in this conception could force experience into the conception; these implementational differences are simply not the sort of thing that could be conceptually relevant to experience. So, consciousness fails to logically supervene on the physical. In this passage, Chalmers argues from a possibility same functional organisation but different physical constitution and lack of phenomenal properties - (P) allows for to a possibility same functional organisation and same physical constitution but lack of phenomenal properties - (P) prohibits. 5 The question I would like to address in the following is: do these conceivability arguments presuppose a notion of the privacy of qualia that is at odds with Wittgenstein s argument against the possibility of a private language? Before I attempt to answer that question, we need to scrutinize the shape, cogency and scope of this argument. II. The Argument against Private Language Wittgenstein s considerations about the possibility of a private language can be found in his Philosophical Investigations 6 243-71, and especially in 256-60. Before that goes the rule-following considerations, that the ability to master a language is like the ability to follow rules. Some commentators, notably Kripke [1982, p. 68] have argued that the impossibility of private languages follows straightforwardly from these more general conclusions about linguistic understanding. As Kripke sees it, the sceptical paradox of 201 is a profound scepticism about rule-following: no facts about me or my past makes it the case that by plus I mean addition and not some bizarre bent concept. The bad news is that we should be irrealists with respect to discourse about meaning. The good news, if you like, is that Wittgenstein proposed a Humean solution to his own paradox: although statements about meaning do not have truth-conditions, and so are not fact- 5 Another type of argument infers an ontological conclusion from epistemic premises: (P) is false, because it is possible that someone should come to know all the physical facts about colour- vision, and yet not know all the facts about colour-vision; when black-andwhite Mary is released from her room, she comes to know a new fact about the world, namely what colour experiences are like; Cf. Jackson [1986], and Stoljar [2000] for discussion. 6 All section references in the following are to this work. 4
stating, they have assertibility-conditions pertaining to community agreements. And if, Kripke s thought goes, meaning is constituted by the best judgements of a potential or actual speech community, the notion of private meaning is senseless. To what extent Kripke is right, I shall not discuss, but there is no doubt that the rule-following considerations are closely connected to the private language argument. I will focus on the question whether the mentioned sections that specifically deal with private language pose a cogent argument against its possibility. What is a private language? It is a language that necessarily only one individual, namely its originator, can understand. The necessity is important. The argument against private language better not rule out the possibility of a language that as a matter of fact only one individual understands. 7 Moreover, a private language is not to be found amongst natural languages. It is a philosophical invention, a by-product of some misguided philosophical theories, and the philosopher s task is, as always following Wittgenstein, to treat the illness by therapy, to get peace by getting the problem to disappear; 133, 255. The problem with a private language is not just that they do not exist; it is that they could not possibly exist. The reason for this is simply that a language that is necessarily unintelligible to anyone but its originator is unintelligible to her as well, and a necessarily unintelligible language is not a language at all. Needless to say, the subject matter of a private language is a private subject matter, properties and objects that necessarily only one individual can come to have knowledge, or justified beliefs, about. So, no concrete objects and properties in spacetime are private in the requisite sense, because they are, actually or possibly, within the epistemic scope of more than one individual. Nor would abstract or fictional objects and properties be private, since nothing in principle bars two individuals from coming to know about them. The best candidates for privacy are mental objects and properties when misconceived by certain philosophical or pre-philosophical theories. This is especially true for phenomenal properties. It is, if not a common-sense conception, then part of what has come to be known as Cartesianism, that the how-it-is-like aspect of sensations and 7 The discussion here has centred on the Robinson Crusoe example: in his case, the distinction between is right and seems right is not actually used, someone might have made use of it. See for instance Blackburn [1984] and Wright [1986, p. 235]. 5
the like is such that only the individuals who have the sensations could possibly form a justified opinion about their occurrence and character. Or take Russell s logical atomism according to which it is necessary and sufficient for understanding a logically proper name for a sense datum that one is acquainted with that particular. Russell admits that such a sense datum language would be private to the speaker in that names in it would never mean the same to the speaker and the hearer. 8 This is not a point about ownership. Clearly if I have a token sensation of a particular type, it is possible that you had another token sensation of the same type, but not that you had the very same token sensation that I had. That should not be in dispute. The question is whether there is anything about my token sensation that you could not possibly come to know, or have justified beliefs about. If so, that token sensation is private to me. And the thought is that if there could be private sensations in this sense, there could also be a private language that describes or at least reports these sensations So, Wittgenstein asks us to imagine counterfactual circumstances in which a private language might be thought possible: the private linguist S has a sensation of pain that is not tied up with any typical natural pain-expressions such as wincing and escapebehaviour; 256. It is a private particular that S has exclusive epistemic access to, that is, no other individual could possibly form a justified view about that particular. The thought is that S introduces a name P for his sensation of pain by a private ostensive definition: (POD) Let P designate this (while focusing his attention inwardly on his sensation). That way, S has mentally associated P with his private sensation. It is true that S could not convey the alleged meaning of P to anyone else given that the sensation itself exhausts the meaning of P, and in the circumstances envisaged only S could have epistemic access to that sensation. We assume that P expresses the property it picks out, that its meaning is entirely constituted by its referent. 9 And knowledge of that referent requires what only S can have, namely acquaintance with it. So, if the subject matter of S s language is private, and its meaning is entirely due to that subject matter, then the 8 Cf. Russell [1918]. Interestingly enough, he also concedes that logically proper names would hardly ever mean the same on two distinct occasions to the speaker himself. 9 Nothing hangs on this assumption: if the meaning of P is only partially determined by the private sensation itself, then we can just introduce a new term Q for that part of the semantic content of P that is determined by that sensation. 6
meaning of his language is also private. 10 Suppose we grant all that. The question is whether S himself could be said to understand P, and related terms introduced in a similar way into his would-be private language. Wittgenstein s first point in 257 is to question the significance and purpose of (POD). More precisely, he alludes to the fact that ostensive definitions, if they are to succeed in introducing meaningful new terms into a language, must rely on a great deal of stage-setting in the language. This point goes back to 28-36 where Wittgenstein criticises the empiricist idea that the meaning of a term consists in a mental correlation with an experience or mental image in isolation from the meanings of any other terms. Depending on how the scenario is described, S must presuppose that pain already has a fixed meaning, for otherwise, S could not introduce a new term for an old type of sensation. Or if not for pain, then for sensation. If the latter did not already have a determinate meaning, S could not introduce a new term for a new type of sensation. Finally, if nothing else, S must assume that the demonstrative this, has the already established meaning that it designates the contextually salient object. This point is repeated in 261 and in 267-8 sensation and something are also words of the common language. In response, we could assume that S has introduced enough terms by ostensive definition for his private sensations; maybe non-sensational terms have a public use, but that is all. If the possibility of a private language is fuelled by substance dualism, it need not be assumed that all terms have private meanings, only mental state terms, so Wittgenstein better not make this assumption against him. In any case, it is not clear why anyone should believe that a private language for all terms is possible; an extreme sense data language according to which terms for tables, chairs and the rest refer to private mental images would lead to this belief, but its ontology should discourage everyone 10 As Wright [manuscript] points out, this assumes that mutual understanding requires reason to think that mutual understanding obtains; otherwise it might just be that as a matter of lucky fact S* does understand S s language, namely if they associate the same symbols with the same private sensations; maybe S* is a physical dublicate of S. In the absence of this assumption, a private language is better defined as one that necessarily no two people can have reason to think they share. 7
from endorsing such a language. If privacy has any plausibility, it must be in the mental, or more precisely in the phenomenal, realm. It is presumably for this reason that Wittgenstein pursues another line in 258. The point is here that if (POD) is to fulfil its purpose, it must establish a connection between the sensation and P that has normative future consequences. Ignoring complications to do with ambiguity, vagueness and context-sensitivity, the meaning of a referential term must be a property that determines whether given present or future objects are correctly picked out by that term. It is not enough that S attends to his sensation while writing P down in his diary. If S s mental act is to have the required semantic consequences, it cannot just be an occurrent event in his consciousness; by (POD) S must form a semantic intention to comply with a rule that governs the use of P. He must undertake in the future to conform to a regularity in the use of P sanctioned by that rule: P is used correctly iff all and only sensations of the same type as the initial sensation are called P ; 261-3. If the reference of P is fixed by associating P with a token of a particular type of sensation, P is used incorrectly if then subsequently used to pick out tokens of a distinct type of sensations. But, and here is the problem, there is no criterion for correctness that could guide S s use of P : whatever is going to seem right to [S] is right. And that only means that here we can t talk about right ; 258. There is no distinction between S having the impression that P is used correctly, and S using P correctly. This follows from the supposed private nature of the referent of P. It has no connection with his sayings and doings, so whether it occurs or not at some later time is entirely a question about whether S judges that it occurs or not at that time. But S s assertion that P reoccurred is a judgement about some state of affairs that are independent of the judgement, so they may obtain without being so judged, and they may be judged to obtain without doing so. S s assertion is, as any other, correct iff the state of affairs asserted obtains. If P occurs, but it seems to S as if P does not occur, S will judge that P does not occur, and if P does not occur, but it seems to S as if P does occur, S will judge that P occurs. But on the present view, there is no space for ways S s judgements about the reoccurrence of the initial sensation could go wrong. In the case of judgements about past perceptions of external objects, there is the possibility that S misremember or misclassify 8
the right perception, even when the conditions for observation and S s other cognitive faculties are not defective. Not so in the case of S s private sensations. See also 259-60. Or in the language of rule-following: if the meaning of P consists in a rule for its correct application, S s understanding of P consists in his grasping this rule. But, as it goes in 202, to think one is obeying a rule is not to obey a rule. Hence it is not possible to obey a rule privately : otherwise thinking one was obeying a rule would be the same as obeying it. But with S, there is no distinction between what the rule for P requires, and what it seems to S that the rule requires, and if there is no such distinction, there is no rule, hence no meaning for P has been bestowed on it by (POD). So far so good. The next step is modus tollens: if the sensation that P is set up to designate is private, and P s meaning is entirely due to that sensation, then the meaning of P is also private. But there is no private meaning, no private rule-following, so the supposed referent of P is not private. So, when S marked P for the first time, he did not make a note of anything; 260. There just are no private mental phenomena. We can sum up the argument against private languages in this way: (1) If phenomenal state P is private to S, and the meaning of P is exhausted by P, only S can in principle know the meaning of P. (2) To know the meaning of P is to know when P is correctly used and when P seems to be, but is not, correctly used. (3) So, if in principle only S can know the meaning of P, only S can in principle draw the distinction between a correct and an incorrect use of P. (4) If in principle only S can draw the distinction between a correct and an incorrect use of P, S cannot draw that distinction at all since it involves reference to in principle publicly accessible circumstances a practice, form of life, agreement in judgements. (5) So, it cannot be that in principle only s knows the meaning of P. (6) So, P is not a private state. III. Two Objections Let us have a brief look at two objections to the foregoing argument. The first objection is that it relies on an unacceptable form of verificationism. This objection in fact splits up 9
into two separate parts. First, one could attack the conditional that if P is a private sensation, then P has a private meaning. Not because P has a public meaning, but because P has no meaning at all. The thought is that one can make no inference from semantic premises to ontological conclusions. 11 That is, one can give up on private meaning, but hold on to private sensations. 12 This response must provide an account of what it is that stops S from introducing a symbol for his private sensations if indeed he has any. Second, one could attack the conditional that if (POD) confers a determinate meaning on P, S must be able to make the distinction between correct and incorrect use of P. One way of sustaining this claim is to draw a wedge between meaning and understanding, where the latter is taken to consist in grasp or knowledge of meaning. What meaning an expression has is an objective matter that possibly goes beyond any means we have for coming to know about them. In our case, when S introduced P by (POD), he somehow created facts about its correct use that goes beyond his own future, informed opinions about that use. That is, when S formed the initial semantic intention with respect to use of P, he then lay down what subsequently should count as correct and incorrect use of P independently of his own later views about this issue. A proper defence of verificationism with respect to ontology or semantics is, if possible at all, far beyond the scope of this paper. Needless to say, Wittgenstein did not endorse this platonic or verification-transcendent conception of meaning, i.e. that there is a fact of the matter whether S s use of P on a particular occasion accords with 11 For a developed view see Devitt [1984]. 12 One might follow McDowell [1998, pp. 294-5] in thinking that the private language argument shows that there is no pre-conceptual given that infants and animals have in common with us. Suppose I have a wound that causes pain. It is true that even if I had not had any concepts I would still have been in pain, but not true that the pain I actually have is identical to the pain I would have had had I not had any concepts. So, there is no highest common factor between my pain and an infants/animals pain. This account at best has it that animals and us have different kinds of pains, at worst implies that animals have no pains. For what does their pain consists in? Not in a phenomenon that is only there if conceptualised, because animals are incapable of conceptualisations, and not in a preconceptual subject-matter for conceptualisation, because no such thing exists at all. Moreover, the account entails that pains for creatures capable of conceptualisation, are essentially conceptual phenomenon. So, the pains I had when I was an infant were distinct from the pains I have when I am an adult! It is not just that I can articulate them better, they change in their nature. But then what makes them both pains? 10
his original intention, but a fact that nobody has any guarantee they will ever be able to disclose. According to Wittgenstein the meaning of an expression is constituted by the use made of it by those who understand the expression. This is witnessed by 43,139 which may be seen as a kind of verification principle. So, when S introduced P into his language, he did not run through all future possible applications of P such that whenever a token-sensation occurs in the future, there is a determinate answer as to whether it is of the same type as the original sensation dubbed P. The second objection is that the argument begs the question against the private linguist S. The conditional that if (POD) confers a determinate meaning on P, S must be able to make the distinction between correct and incorrect use of P is safe enough. The problem is that on the Cartesian conception of the mental, a suitable endowed subject S is not only incorrigible, but also infallible, and all-knowing with respect to his occurrent sensational states. S introduced the term P by (POD) to refer to a particular token of a type of private phenomenal state. Given that S has an infallible ability to recognise and re-identify tokens of the same type of state in the future, there is no problem about S having a private practice with respect to P. On this view, there is a distinction between what seems right to S and what is right, that is, the obtaining of the relevant state of affairs is distinct from S s judgement that they obtain. It is just that whenever they obtain, S judges that they obtain, and whenever they do not, S does not judge that they obtain. But the argument assumes, or so the thought goes, that there are cases where S mistakenly judges that P occurs and where P occurs with S judging so, and so makes an assumption the Cartesian rejects. 13 The point is good, but it does not get the Cartesian very far. There are two reasons for this. The first is that it is doubtful whether it so much as makes sense to be infallible with respect to a private subject matter. In 265-6 Wittgenstein imagines, by analogy, an individual with photographic memory, checking when a train leaves by bringing forward a memory image of the relevant page in the timetable. Now, this may work, but why? Because we have an independent test for the correctness of the memory image. We can 13 The same objection would presumably apply on the Russellian account according to which knowledge by acquaintance, but not knowledge by description, is infallible. Modern strong property dualists, however, make no such epistemological commitments. 11
check the timetable once more. But in the case of private sensations, there are no independent means by which S can inspect the correctness of the memory image of the original sensation. Whatever S brings to mind, it will not tell him how P is to be applied in future cases if there are no external methods of testing. 14 The second reason is that even if we grant infallibility with respect to S s sensations that is not enough to make his judgements about them infallible. When S judges that now P occurs again, he intends to use P with the same meaning as when he baptised the original sensation P. That in turn presupposes that he brings the sensation in question under the right concept. But then infallibility with respect to the sensation falls short of fallibility with respect to the judgements. And surely everyone must concede that our ability for conceptual classification is fallible, this holds for judgements based on perception as well as for judgements based on introspection. But in the case of a private language, it does not make sense to say that S looses track of which term goes with which sensation, for whichever sensation S judges goes with P is the sensation that goes with P. Nor does it make sense to say that S has an impartial grasp of the meaning of P since S is assumed to be infallible and all-knowing when it comes to the sensation that constitutes the entire meaning of P. 15 IV. Anti-Physicalism and Privacy Let us go back to the dualist arguments against physicalism. According to substance dualism and strong property dualism, there are law-like correlations between mental and physical properties. It may be that say being in pain and having C-fibre firing are nomologically co-extensive properties, or that there are some ceteris paribus laws 14 This reflects an earlier point about understanding in 138-84, namely that to grasp the meaning of a term never consists in having a mental picture before the mind; rather it is constituted by an ability to use that term in accordance with publicly accessible rules. 15 Blackburn [1984] argues that the private linguist S, if ingenious, can have a private practice with respect to his use of P, if he constructs a theory around his observations of the private sensations. On Blackburn s view, S is fallible yet incorrigible. Wright [1986, pp. 239-47] responds that this fallibilism fails for the same reason as the private linguist who invents no theory but merely reports his sensations, because when theory and observation conflict there is no best way to resolve the inconsistency. 12
pertaining to pain-behaviour and pain. But that is all. There are no metaphysical or conceptual connections between a term for the property of being in pain and terms for physical neurological or behaviourial properties. In worlds with different laws of physics, individuals are in pain, but have neither C-fibre firing nor display painbehaviour, and in other worlds with yet different laws of physics, individuals have C- fibre firing and display pain-behaviour, but have no pain. In fact, there is a subclass of the former worlds in which individuals have pains and yet display no typical behaviour and are in no typical brain-state. And what is true for pain is true for all phenomenal states. According to these forms of dualism, it is conceptually and metaphysically possible that individuals be in various phenomenal states without showing any typical verbal or nonverbal behaviour, and without being in any typical neurological states, or vice versa. Such possibilities were, as we saw, explicitly embraced in the conceivability arguments against physicalism. But this means that, if not in the actual world with its law-like mentalphysical correlations, then at least in such worlds, phenomenal states are private in the sense we have defined the term. No two individuals could ever acquire justified beliefs about their respective sensations, so no two individuals can ever have any reasons to think they both understand P in the same way, because any such reason would derive from circumstances that they could both at least in principle come to know. In this case, all they have to go on is their respective utterances of now P reoccurred, and so on. The term P picks out all and only token-sensations of the same type as the sensation that P was originally introduced by (POD) to designate. But what counts as the right sensation here except what the private linguist finds is the right sensation? There are no publicly accessible criteria for the correct application of P. There are neither behaviourial nor neurological criteria for having P. The important point is that if the foregoing argument is sound, it afflicts all those views about the mental that deny any conceptual connection between mental state terms and terms for publicly observable circumstances, e.g. terms for behaviour. The argument shows that, as Wright [pp. 214, 235] says, going public must make all the difference, that is, P must have sufficient public content to enable others, at least in favourable circumstances, to form a justified opinion about its correct use independent of S s opinion. There must be public criteria, e.g. criteria in man s behaviour, for linguistic 13
understanding: 269. But in the envisaged circumstances, others have no independent way, not even in principle, of forming any well-founded views about the correctness of S s use of P. In this respect, the Cartesian substance dualist and the strong property dualist are in the same boat. Both are committed to the conceptual possibility of a private language, yet Wittgenstein s argument, if sound, rules out this possibility. What it shows is not just that as things stand, there are no private languages, it is that, no matter how things are, there could not be a private language. The fact that the Cartesian substance dualist believes, and the strong property dualist denies that we are infallible and all-knowing with respect to our occurrent sensational states, is, if a meaningful claim at all, inessential to the argument. Both must allow for conceptual fallibility, and that, as we have seen, suffices for its cogency. The only difference between the two is that the former is committed to the more extravagant claim that mental substances are private, whereas the latter is merely committed to the claim that some mental properties are private. According to strong property dualism, the meaning of terms for phenomenal properties derives exlusively from their felt phenomenal qualities, but there is no way a third party can attain a justified view about those qualities in the envisaged circumstances. The only way to form a conception of an intrinsic phenomenal property is by direct acquaintance, or alternatively, by a priori reasoning from concepts of intrinsic properties that one is directly acquainted with, e.g. Hume s example of the missing shade of phenonemenal blue. One can form an extrinsic conception by descriptions of an intrinsic property the property causally responsible for such-and-such but such a conception leaves open many possibilities as to the nature of the property. 16 In response the dualists could avail themselves of the argument from analogy. A key lesson from Wittgenstein is that I cannot acquire mental concepts from first observing my own case and then inferring how it applies in other cases. I could not gain knowledge of S s mental life by an inference of analogy. S is in pain iff S exhibits behaviour typical when I am in pain. As Wittgenstein points out in 302, this cannot be 16 Chalmers [1999] is explicit that phenomenal concepts are not causal concepts in the sense that the property a phenomenal concept picks out is not conceptually connected to any - actual or potential - causes or effects that property has. 14
the general model for knowledge of other minds. For what I have to do is not just to imagine having my pain in S s body, which I could do on the dualist model, but to imagine having S s pain in his body! This is no doubt to quick. Here s one response: the nomological correlations between pain and pain-behaviour function as reference-fixers. I learn pain by observation of the circumstances input stimuli and output response - in which others use that term. I associate pain with the qualitative feel I experience when in those circumstances, and infer justifiably given these correlations that others are in pain when they are also in those circumstances. But the pain-behaviour is not part of the meaning of pain, because there are, at least counterfactual, pains without painbehaviour and pain-behaviour without pain. I suspect this line ultimately fails, but it is certainly worth pursuing by the dualists. Contrast with ordinary ascriptions of intentions. Suppose S forms the intention of bringing it about that P, e.g. always to help old ladies to cross the street. It is granted that there is an epistemic asymmetry in ordinary psychological discourse between the way we know our own mental properties and the way we know others. In the normal run of things, I can know without inference, and without giving reasons, that I have just formed a particular intention. You can come to know about my intentions, but you can be asked to provide reasons for your knowledge claim that must derive from my sayings and doings. In short, I have a first-person authority with respect to the content of my own intentions. Nevertheless, there are publicly accessible circumstances under which it would be true to say that I did not have the intention I claimed to have; if, say, every time I was confronted with an old lady waiting to cross by the road, I deliberately, and without good reason, ignored her obvious signs to cross. Under such circumstances, it would be true to say of me that I did not after all have the particular intention I claimed I had. References Blackburn, S.: The Individual Strikes Back, Synthese, 58, 1984, pp. 281-301. Chalmers, D.: The Conscious Mind: In Search of a Fundamental Theory, New York, Oxford University Press, 1996. 15
: Materialism and the Metaphysics of Modality, Philosophy and Phenomenological Research, 59, pp. 473-96, 1999. Devitt, M.: Realism and Truth, Oxford: Blackwell, 1984. Jackson, F., 1986, What Mary Didn t Know, Journal of Philosophy, 83, pp. 291-5. : Armchair Metaphysics, in Philosophy in Mind, ed. J. O Leary Hawthorne and M. Michael, Kluwer, 1994. Kripke, S.: Wittgenstein on Rules and Private Language, Oxford: Blackwell, 1982. Lewis, D.: What Experiences Teaches in his Papers in Metaphysics and Epistemology, Cambridge University Press, Cambridge, pp. 262-90, 1999. McDowell, J.: One Strand in the Private Language Argument, in his Mind, Value & Reality, Harvard University Press, Cambridge: Mass., 1998. Russell, B.: The Philosophy of Logical Atomism, in The Collected Papers of Bertrand Russell, Vol. 8: The Philosophy of Logical Atomism and Other Essays 1914-19, London: George Allen and Urwin, 1986. Stoljar, D., 2000, Physicalism and the Necessary A Posteriori, Journal of Philosophy, 97/1, pp. 33-54. Wittgenstein, L.: Philosophical Investigations, Blackwell, Oxford, 1953. Wright, C.: Does Philosophical Investigations I. 258-60 suggest a cogent argument against private language?, in Subject, Thought and Context, eds. P. Pettit and J. McDowell, Oxford, pp. 209-66, 1986. 16