Vous êtes sur la page 1sur 19

Mind Association

Sentience and Behaviour Author(s): Robert Kirk Reviewed work(s): Source: Mind, New Series, Vol. 83, No. 329 (Jan., 1974), pp. 43-60 Published by: Oxford University Press on behalf of the Mind Association Stable URL: http://www.jstor.org/stable/2252795 . Accessed: 14/03/2013 02:10
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Oxford University Press and Mind Association are collaborating with JSTOR to digitize, preserve and extend access to Mind.

http://www.jstor.org

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

Sentience and Behaviour


ROBERT KIRK

The sceptic's suggestion that others, despite their anatomical and behavioural resemblance to himself, might after all be insentientwithout sensory experiences of any kind-is familiar enough. But one of the assumptions on which his position rests is not, I believe, generally considered as seriously as it deserves to be. For the sceptic must assume that it is at least logically possible for there to be an organism indistinguishable from a normal human being in all anatomical, behavioural and other observable respects, yet insentient. And this assumption does not involve any obvious commitment to scepticism: to adopt it seems to be compatible with claiming to know that in fact there are no such organisms. But the question of its validity has immediate relevance to the mind-body problem, as well as to the philosophy of perception. If it can be shown to be valid, then certain widely held views about the analysis of reports of experience must be abandoned, and it is hard to see how Materialism can survive. But if it can be shown to be invalid, then it seems that Materialism is almost home and dry. Some philosophers have made use of the assumption I have described1; and perhaps many people will regard it as obviously valid. But there are objections to be met. Can it make sense to say that something indistinguishable from a man might still be different in the radical way imagined? Even if the description can be adequately clarified, doesn't it involve contradictions or conceptual absurdities? My aim is to show that it is indeed logically possible for there to be organisms answering to the description I have given (Zombies, for short). My starting point will be the story of Dan, who, it seems, turned into a Zombie by stages.
I

E.g. Thomas Nagel in 'Armstrong on the Mind', Philosophical Review, 79 394-403; Moreland Perkins in 'Sentience', Journal of Philosophy, 68 (1971), 329-337. Keith Campbell in Body and Mind, New York, I970, uses a rather similar assumption.
(1970),

43

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

44

R. KIRK:

One day Dan (who had been perfectly normal until that time) accidentally cut his hand and started to behave very strangely. True, his behaviour included features indistinguishable from the typical behaviour of a normal person suffering pain of the kind and intensity to be expected in the circumstances. He winced, exclaimed 'Ouch!', nursed his hand, answered 'Yes' to questions about whether it hurt, and so on. But in addition to, and as it were superimposed upon, these normal bits of behaviour was a second kind of behaviour wildly different from the typical behaviour of someone in pain. He expressed astonishment; and he uttered sentences and behaved in ways which, in more normal contexts, would have been taken to be protestations to the effect that he felt no pain whatever, that it was as if he were 'totally anaesthetized'. He appeared to be astonished by two things: first, by the fact (as he seemed to regard it) that he really felt no pain at all in spite of a fairly serious injury; and second, by the fact that he was nevertheless wincing, groaning, and uttering such sentences as 'It hurts like hell'. Thus his behaviour as a whole was bafflingly incoherent. (I am using the word 'behaviour' here and throughout in the sense of mere bodily movement, without commitment as to whether or not it consists of actions. Similarly for other words used in specifying Dan's behaviour, e.g. 'say', 'express'.) After a time-about as long as it would have taken for a normal person in that situation to have ceased feeling pain-Dan went on to describe his experiences (as it seemed) in the following terms. He continued to insist that he really had not felt any pain, and he said that so far as his groans, winces, exclamations, complaints and other 'pain-behaviour' were concerned, they had been a series of happenings over which he had had no control: he had felt 'like a puppet'; he had merely noticed their occurrence as one might notice one's finger involuntarily twitching. Indeed, he said that it seemed to him that the whole series of items constituting his 'pain-behaviour' had been a series of extraordinarily complicated involuntary twitches. Now obviously there can be no logicalL absurdity in this description of Dan's behaviour. A good actor could match it. So
I

Throughout I use 'logical' in a very liberal sense, so as to include what some might prefer to call 'conceptual'. I doubt if Quinean scepticism about the legitimacy of such notions seriously affects my main points.

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

SENTIENCE

AND BEHAVIOUR

45

let us imagine that we have actually found Dan behaving as described; and that from that time on, whenever he is in a 'painsituation' (i.e. a situation in which one would expect a normal person to suffer pain) his behaviour corresponds, mutatis mutandis, with that described above (although there is less apparent surprise, and more resignation, as time passes). Could we make any sense of these phenomena? II Dan's behaviour would be intelligible, I suggest, on the hypothesis that he has simply ceased to feel pain in pain-situations, despite the occurrence of behavioural features of types normally associated with pain. It is a natural enough suggestion, given the bizarre data. But I need to make very plain what a sound hypothesis it is, since this point will carry a lot of weight later in my argument. The hypothesis that Dan felt no pain would explain at least three puzzling features of his behaviour. Firstly, it would explain (what appeared to be) his expressions of astonishment at not feeling pain in pain-situations. On this hypothesis they would be genuine expressions of astonishment, not mere behavioural simulacra. Secondly, it would explain his seeming expressions of astonishment at his pain-behaviour. These too would be genuine expressions of astonishment. Thirdly, it would explain his description of the way in which his apparent pain-behaviour was produced as 'quite involuntary', 'something that happened' to him, 'like a series of very complicated twitches'. For on the present hypothesis, what appeared to be his pain-behaviour would be nothing of the sort, so that its occurrence could be expected to fit such descriptions. Moreover, although on this hypothesis these items of behaviour were not produced in response to any feelings of pain, they need not be totally inexplicable. For the workings of Dan's central nervous system might by themselves be sufficient to cause the physical movements in question: the contortions of the facial muscles involved in wincing; the workings of the organs of speech, and so on. Contrast with this hypothesis any hypothesis on which Dan still feels pain in pain-situations. How would it be possible, on such a hypothesis, to account for (what seem to be) his expressions of astonishment? Well, perhaps these items of behaviour are the basic, inexplicable datum, corresponding to the basic assumption

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

46

R. KIRK:

of the first hypothesis, to the effect that Dan feels no pains. But notice how many questions this hypothesis raises, compared with the first. On the first hypothesis it is certainly mysterious that Dan feels no pain in pain-situations. But an explanatory hypothesis cannot be expected to explain everything. And on this one, as I said, Dan's pain-behaviour need not be wholly uncaused: it may reasonably be supposed to be caused by his neural states, which in turn are caused by damage to bodily tissues. On the rival type of hypothesis, though, the expressions of astonishment come out of the blue; and apart from the mystery of their spontaneity there is the additional mystery of why they are expressions of astonishment. For not only is there nothing to cause any kind of astonishment: the actual situation is, ex hypothesi, one (viz. Dan's genuinely being in pain) in which the expression of astonishment at not being in pain is hardly intelligible. A further difficulty for any hypothesis on which Dan really feels pain in pain-situations is his own subsequent descriptions (as they seem to be) of his experiences. Not only does he continue to insist that he has felt no pain: he also describes his pain-behaviour as 'quite involuntary'. If he really felt pain, such remarks would be inexplicable. Of course, it would be possible to introduce further hypotheses to the effect that these utterances themselves are merely a set of complicated involuntary twitches. But if they are, how is it that Dan fails to comment on this fact? He could hardly avoid noticing them. And this difficulty also arises in the case of the expressions of astonishment. For it is strange that Dan, who does feel pain on the hypothesis we are now considering, should not notice the fact that such expressions are out of place, and evince some surprise. I conclude that if we were faced by Dan or someone behaving like Dan, it would be not merely intelligible, but very reasonable, indeed justifiable, to say that he had ceased to feel pain, despite his apparent pain-behaviour. III Dan continues for six months in the abnormal state I have described, and then suddenly starts to show a further set of behavioural peculiarities. This time they concern his sense of smell. Asked to smell some roses, he sniffs, utters some such remark as 'They smell marvellous'-but then shows astonishment

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

SENTIENCE

AND

BEHAVIOUR

47

and distress. He appears to be astonished by two things. Firstly by the fact that, although he has hitherto enjoyed a keen sense of smell and is not suffering from a cold, he cannot detect the scent of the roses at all. And secondly by the fact that, despite his not having been aware of any scent, his facial muscles have just formed an expression of enjoyment, and his vocal organs have produced a remark that would have been appropriate only if he had really detected, and enjoyed, the scent of roses. The facial expression and the movements of his organs of speech were all (so he seems to think) involuntary, just as in the case of his production of sentences 'expressing' or 'describing' pains in pain-situations. If all this were really happening we should expect Dan to show distress, especially in view of his continuing abnormalities in pain-situations. And so he does, and with increased intensity when the same sort of thing is repeated in all subsequent situations where a normal person would be aware of some smell. Clearly there is no logical absurdity in this description. Nor, for reasons similar to those already given in connection with the case of Dan's behaviour in pain-situations, is there any logical absurdity in the hypothesis that Dan has been unable to detect smells since the time in question. Indeed, for the same kind of reasons as were given for the earlier case, it would be a reasonable hypothesis: it would explain more than any hypothesis according to which Dan still detected smells. The story continues. After each of the next few periods of six months an additional peculiarity, analogous to those already described, suddenly appears in Dan's behaviour, affecting, or seeming to affect, one sense after another. Similar arguments to the foregoing make it a reasonable hypothesis that Dan is progressively being deprived of sensory experiences of the various kinds in question. The reasonableness of this hypothesis is reinforced by the fact that Dan himself, after the first couple of sensory losses (or apparent losses), begins gloomily to speculate (as it seems) about which sense will be the next to go, and whether he will eventually lose all varieties of sensory experience while (macabre prospect) his body generates an exact simulacrum of normal conscious behaviour. For it fits in well with the hypothesis that the hypothesis should itself be adopted by Dan; and the fact that Dan behaves, within the limits shortly to be described, as if he has adopted it is itself hard to explain unless the hypothesis is true.

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

48

R. KIRK:

Let us suppose, then, that Dan has now reached the stage where, according to the hypothesis that I am assuming we have adopted, he is left with the enjoyment of only two senses: sight and hearing. Most of his verbal and non-verbal behaviour (still, of course, in the neutral sense) is appropriate to his circumstances and to the plans, purposes and intentions that a person in those circumstances might reasonably be expected to have. But there are gross abnormalities. At times, and with what looks like growing desperation, Dan insists that he has lost all his senses but sight and hearing, and dreads the time when one or both of them will go too. He insists on the involuntary nature of all or most of that behaviour the appropriateness of which for normal people would depend on their having full enjoyment of the senses which, on our hypothesis, he now lacks. For instance, he sometimes hears himself 'complaining' of giddiness (i.e. uttering appropriate sentences) even though, as he insists, and as our hypothesis has it, he really has no bodily sensations at all. He expresses surprise when he hears such remarks issuing from his mouth. Thus, by his account, quite lengthy sequences of his own behaviour have become inaccessible to him: they occur automatically, without his noticing or being able to attend to them. Moreover, even when he is able to attend to some parts of his behaviour by sight or hearing, he is powerless to inhibit them. He hears himself 'describing' the flavour of an apple, say, even when (as he insists) he is most keenly aware of the absence of any corresponding experience. It is only in the gaps between such involuntary pseudo-linguistic performances (which is what they are on our hypothesis) and then only with great effort, that he is able to interpose his own account-which, of course, contradicts what his apparently automatic utterances would normally be taken to have expressed. The effort he apparently has to put into forcing out his own descriptions of his situation has been increasing with each stage of his deterioration. Loss of sight, as it seems to be, comes after the next interval of six months. Dan (or the individual I am still calling Dan) wakes one morning and with a prodigious effort (as it seems) manages to groan: 'I've gone blind: can't see a thing.' Yet he prepares to go to work as usual, with only a few effortful complaints about having gone blind. When he says 'Goodbye' to his wife, or at any rate produces that utterance, as if it were a normal morning, he

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

SENTIENCE

AND BEHAVIOUR

49

struggles to exclaim 'No! I can't see!' But he goes on to drive to work as usual. Once there, he makes a tremendous effort and shouts: 'But I can't see anything! I want to go home!' He is taken to a mental hospital. After six months, plus another week or so, he is pronounced cured, and returns home, where his wife gratefully accepts the tale the doctors tell her. However, those of his friends who have been closely and intelligently following the course of Dan's troubles fear that it might be more reasonable (though perhaps not more natural), in view of his history, to suppose that instead of having been cured at the mental hospital, what happened was that his last sensory link with the world-his sense of hearing-had, after the usual six months' interval, been cut. They fear that in his case it really is 'silent and dark within' -he has turned into a Zombie. But as time passes, their hypothesis gradually comes to seem absurd. After all, Dan now seems to be normal, and they can hardly avoid treating him as if he were. So they come to accept that he is normal. But even if it is psychologically very difficult, even impossible, in the face of behaviour indistinguishable from normal human behaviour, to believe that the individual who exhibits it is really insentient, this could still in fact be true: what it is psychologically impossible to believe need not therefore be false. I think that Dan's history so far makes it a tenable hypothesis that he has become, or been superseded by,2 a Zombie. At least it seems to make this suggestion intelligible. However, a different suggestion might be thought more reasonable in the circumstances: that what came back from the mental hospital (let us call it D) was neither a Zombie, nor Dan, but a different person, who had been taking over Dan's senses one by one. I shall now take the story a little further, to a point where the hypothesis that D was in fact a Zombie becomes not only tenable, but justifiable. (I shall need to use this conclusion at a later stage.) In general, D behaves for six months in ways indistinguishable from the ways in which his friends would have expected Dan to behave. For example he (or it) answers to the name Dan. However,
I 2

A phrase used by Iris Murdoch in rejecting analytical Behaviourism, The Sovereignty of Good (London, I970), p. 13. It might be argued that Dan, being a person, could not become or turn into a Zombie, which no doubt is not a person. Cf. David Wiggins, Identity and Spatio-Temporal Continuity (Oxford, I967), passim. But the issue is irrelevant to my argument, since my concern is only to show the logical possibility of Zombies, and that we could have good reasons for supposing that we had been confronted by one.

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

50

R. KIRK:

when six monthshaveelapsed,D occasionally produces,apparently with greateffort,inarticulate noises which seem to expressdelight. After a time there are sentencessuch as 'I can hearagain! Perhaps I'll get back my other senses.' After anothersix months: 'Marvellous! I can see again as well as hear! I wish the next couple of yearswould pass quicklyso that I could be myself again.I seem to have been asleep,but I vaguelyrememberthe horribleexperience of losing all my senses, one afterthe other.' I need not fill in much more detail,but one thing is important:at no time does D produce anythinglike expressionsof distressor apprehension. Now according to the hypothesis we are examining, D is a personwho has takenover from, or superseded,Dan: when D has taken over completely,Dan is, if not non-existent, then without any form of sensory experience,while D has all his senses about him. But then D's productionof the sentence 'I can see again as well as hear'would markthe second stage of his being taken over by a different individual (Dan, on a reasonablehypothesis): it would markthe loss of sight by the person D is supposed to be. But a normal person, suddenly deprived of his sight, would be expectedto show some distress,and, as my story has it, D shows nothing of the sort. Thus there is a significant asymmetrybetween what happens at the onset of the successive stages of D's take-overby (presumably) Dan, and what happenedat the onset of each of Dan's originalsensorylosses: Dan showed distress,but D shows none. This fits in well with the hypothesisthat what came back from the mental hospital was a Zombie, but not with the hypothesisthat it was a person.For a Zombie,despiteappearances, has no senses to lose, and thereforecannotbe distressedat the loss of a sense. It would be possible to fill in furtherdetails, but I think I have said enoughto makeit clearthat the hypothesisthat Dan had been supersededby another person would be less plausible than the hypothesis that he had turned into, or been superseded by, a Zombie; and that faced by the sorts of behaviour that I have ascribedto Dan, we should be justified in supposing that, for a time, we had been dealingwith a Zombie.

IV
Now so far I have said nothing about the contentsof Dan's skull. Let us suppose that, to anyone familiar with the workings of

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

SENTIENCE

AND BEHAVIOUR

5I

normal human brains, Dan's neural activity at any time appears to be consistent with his physical activity at that time. This means that in the stages intermediate between normality and putative Zombiehood his neural activity as a whole, like his behaviour as a whole, is bafflingly incoherent. To a Materialist, indeed, it will even appear to infringe the principle of Sufficient Reason. Such neural activity as corresponds, for example, to Dan's protesting that, despite appearances, he is not in pain, will appear not to have been caused by any preceding neural or other physical activity at all (although a Dualist could say that it was caused by something non-physical, or not wholly physical). But when Dan has, putatively, reached the stage of complete Zombiehood, his neural activity will be indistinguishable from that of a normal human being. And here my position comes into conflict with certain philosophical views, according to which it would be impossible for something to be both indistinguishable from a man in all observable respects, and radically different in the way envisaged. One, which I shall discuss at a later stage (sec. VI), may be roughly characterized as verificationistic. The other takes the line that sentience follows logically from the fact that what I call Zombies are indistinguishable from living men. One version of this second approach would be based on analytical Behaviourism. But since the arguments I shall present against a rather different view will (if they are good) incidentally refute analytical Behaviourism, I shall not discuss the latter separately. The argument I want to examine goes as follows: (i) All human behaviour (in the neutral sense) is in principle fully explicable in physico-chemical terms; (2) The performance of certain causal functions connected with behaviour is a logically sufficient condition for the sentience of any organism in which they are performed; therefore (3) Contrary to my description of them, what I am calling Zombies, must, logically, be sentient. For the moment I shall ignore the possibility that (i) is false, and confine myself to a consideration of (2). (2) is entailed by the causal analysis of the concepts involved in sentience. According to David Lewis:
'.. . the definitive characteristic of any experience as such is its causal role. The definitive causal role of an experience is expressible by a finite set of conditions that specify its typical causes and its typical effects under various circumstances.

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

52

R. KIRK:

By analytic necessity these conditions are true of the experience and jointly distinctive of it.'" According to this analysis, then, if a process now going on in some individual's brain has the causal role ascribed to, say, pain, it follows logically that the individual is in pain. So if, as I am assuming, all human behaviour is explicable in physico-chemical terms, there are physical (presumably neural) processes in Zombies which have the causal role ascribed by this analysis to pain, from which it follows, if the analysis is correct, that Zombies feel pain; and similarly for other varieties of sentience. Now I think it is quite a good prima facie argument against the causal analysis that each of us can imagine that what happened to Dan should happen to himself. Of course we could hardly imagine ourselves actually being in Dan's final stage; but we are familiar with the idea of sleepwalking, and Dan's situation as a Zombie need not be very much unlike that of a sleep-walker. But my aim in this paper is to give a stronger justification than this kind of intuitive plausibility to my thesis that Zombies are logically possible, and I am therefore going to provide further arguments. V We are assuming that all human bodily movements can in principle be explained in physico-chemical terms. We may therefore regard the brain as a device which causes the body to move in certain ways, and modifies these movements in response to various kinds of stimulation of its sense organs. Thus the brain performs a certain (very complicated) causal role, or set of causal roles, connected with the movements of the body to which it belongs. Now there is clearly nothing logically objectionable in the suggestion that a thing different in many ways from a human brain might nevertheless perform exactly the same causal roles, with respect to the movements of a certain human body, as would normally be performed by a human brain. But, as I shall shortly explain, unobjectionable cases can be described in which it is very obvious that the mere performance of those causal roles would not entail sentience. Nor, as I shall also argue, are the differences between such logically possible cases and the actual human case
D. K. Lewis, 'An Argument for the Identity Theory', Journal of Philosophy, 63 (I966), I7-25. See pp. i9 f. Cf. D. M. Armstrong, A Materialist Theory of the Mind (London, I1968), passim.

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

SENTIENCE

AND

BEHAVIOUR

53

such that their removal would allow this logical gap to be filled. If this is correct, it will follow that in the normal human case, too, contrary to what is implied by the causal analysis, the brain's performance of the relevant causal roles does not entail sentience. There can be no logical objection to the following: A team of scientists removes the brain from a living human being, makes alternative arrangements for the circulation of blood, etc., and itself takes over the causal functions that are in question. (One way in which this might conceivably be done would be by arranging for all the impulses from afferent nerves to be registered by characteristic radio transmissions to themselves, while they transmitted impulses to the efferent nerves, causing the body to move about, eat, and speak in ways indistinguishable from those of a man.) The body, thus controllable, would be a kind of superpuppet, and the team of scientists (whom I will call the Brain Team) would jointly perform the role of puppet-master. Here, then, is a situation where the relevant causal roles are certainly being performed, but where, very obviously, that fact does not entail sentience: the behaviour of our super-puppet no more entails sentience than does the behaviour of the wood and cloth variety. Of course this case is very far removed indeed from the actual human situation, and it might be doubted whether the point I have just made has any bearing on our problem. I shall now try to make it plain that the differences do not affect the logical situation: removing them will not produce an entailment of the sort required. Two glaring differences are in the size and location of the Brain Team. But if (as is possible, though fantastic) the Brain Team were microscopically small, or indeed if the super-puppet were gigantically large, they might be ensconced within its skull, controlling the body rather as the flight crew controls an aircraft. Yet obviously these modifications of relative size and location would not produce the required entailment: it would not follow that the complex entity which consisted of the body with its controlling Brain Team was a sentient being. This consequence would no more follow from their being inside the body than it follows that a pantomime horse, for example, is a sentient being from the fact are nsidei it. that i-ts animatov% Is it, perhaps, the consciousness and purposiveness of the members of the Brain Team which blocks the entailment in the case I have described? No: all the causal analysis requires is that the

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

54

R. KIRK:

relevant causal roles be performed. It imposes no conditions on exactly how they may be performed, e.g. it does not prohibit their being performed by conscious, purposive individuals. Moreover, if the causal analysis were to be revised by the addition of some such condition, the modification would be unwarrantable. For there is no logical absurdity in the suggestion that conscious, purposive individuals might be found playing vital roles within the bodies or brains of other conscious, purposive individuals-even within the reader's own head, despite his undoubted sentience. For similar reasons, the structure or chemical composition of whatever mechanism performs the causal roles in question cannot relevantly affect the logical situation. The causal analysis imposes no conditions on the structure or composition of the mechanism, and even if it did, their imposition would be unwarrantable: we cannot rule out a priori what kinds of structures could be those of a sentient being, so long as they performed the relevant functions. However, someone who objected to the claim that Zombies are possible might rest his position on a simple assertion of the oneway entailment, rather than insist on the full causal analysis. He might assert that the functioning of a brain indistinguishable from a normal human brain was a logically sufficient condition for sentience. But it is clear that if (as we are assuming) other kinds of structures could perform the same causal roles as a human brain, it is merely arbitraryto insist on this entailment. It is only because he already knows that human beings are sentient that he is inclined to say that having a normal human brain entails being sentient. Besides, the concepts of structure and chemical composition seem to have no relevant links with those of sentience. It appears, then, that neither the Brain Team's size, nor its location, nor its members' consciousness and purposiveness, nor its structure, nor its chemical composition can account for the fact that its performance of all the relevant causal functions of a normal human brain does not entail the sentience of that entity which it controls. If this is correct, then even when all these differences between the Brain Team and a normal human brain have been removed, the alleged entailment still cannot hold. But to remove all these differences would simply be to replace the Brain Team by something not importantly different from a normal human brain. I conclude that the performance of the relevant causal functions by a normal human brain does not entail the sentience of the person whose brain it is,

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

SENTIENCE

AND BEHAVIOUR

55

I have been assuming, of course, that all human bodily movements are in principle explicable in physico-chemical terms. If they -are not, then presumably physical mechanisms could not perform the relevant causal functions in beings observably indistinguishable from men. In that case Materialism would be false, or in deep trouble, and the possibility of Zombies would lose some of its interest. However, there would still be no contradiction in the description of Zombies: they would flout the principle of Sufficient Reason, no doubt, but that is presumably not a logical principle, nor even one whose occasional infringement would involve conceptual absurdity. That concludes my main argument against the causal analysis. If the argument is sound, then the causal analysis fails to provide a basis for rejecting the logical possibility of Zombies, because it fails to take account of some of the phenomena necessarily involved in sentience. Now among these, it is generally held, is the phenomenon of private detectability: the fact that the way in which the subject of a sensation or sensory experience is put into a position to describe it is quite different from the ways in which others might be put into a position to describe it. That the causal analysis fails to do justice to private detectability may also be seen, I think, by considering the following possibility. (This will be a second argument against the causal analysis.) Suppose that human beings never felt pain in pain situations, but that those sorts of stimuli which, as things are, typically cause pains in humans, caused something different. Instead of pains they cause, let us suppose, an uncontrollable trembling of all the limbs, together with an automatic tendency of the hands to touch whatever part of the body has been damaged or (in the case of internal damage) whatever part is closest to the site of the damage. And suppose that the degree of violence of the involuntary trembling varies with the degree of seriousness of the damage. Suppose further that when in pain-situations a person starts trembling in the way described, there is a suspension of the normal operation of kinaesthesis, so that he can discover the occurrence of the trembling only by observation or inference. This means that there is no way for him to detect the trembling that is not open to others. In other words, the peculiar trembling process typically caused by the stimuli which actually cause pain in us is not, we are to suppose, privately detectable. All this is clearly a logical possibility. Now if all this were the case it is obvious that for a person to be

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

56

R. KIRK:

in the trembling state brought on by tissue damage would be: disturbing and disruptive of normal activity; likely to provoke winces, groans, or even screams, depending on the degree of violence of the trembling; and likely to cause the subject to want to take all possible steps to remove the cause of the disturbance. It is also clear that the phenomenon I have described would serve very much the same life-preserving functions as in fact the occurrence of pains serves for us. It would tend to make us do whatever was possible to repair damaged tissues, and to avoid situations in which such damage was likely to occur. Thus the typical causes of the trembling-states would be identical with those of our own pain states; and their typical effects on our behaviour and dispositions would be, if not identical, then substantially similar to those of our pain-states. Yet despite these very close similarities between the causal roles of pain-states and these trembling-states, the trembling-states would not be privately detectable. I do not see how this could be so if the causal analysis had any validity. VI I come now to the second, epistemological, objection to the suggestion that Zombies are logically possible. It goes as follows: 'You are committed to allowing the logical possibility of two individuals, A and B, who cannot even in principle be distinguished on the basis of observation of their current physical structures and states, yet are radically different in a very important respect: A is a man; B is what you call a Zombie-insentient. But how can it make sense to assert the existence of a difference that cannot be detected by observation? Ex hypothesi, both A and B conform to all the criteria on the basis of which we ascribe sensations and other experiences to people. How can you avoid the conclusion that they must both have sensations and other experiences? For if on the basis of all possible tests one of them comes out as sentient (and A, since he is human, certainly will), then so will the other. You may perhaps have shown that no contradiction follows from the description of a Zombie if one attends only to the details of the description. But these more general considerations strongly suggest that Zombies are not, after all, genuinely conceivable: the notion turns out to be incoherent.' The starting-point for my reply is the story of Dan, Notice

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

SENTIENCE

AND

BEHAVIOUR

57

first that, because he was observably distinguishable from a normal human being in every stage but the final stage of putative Zombiehood, and because the hypothesis of gradual sensory loss was shown to be a reasonable one, the objection can at most apply only to the description of him as a Zombie: it has no force against the assertion that a man could fit Dan's description and progressively lose his senses in the way hypothesized.-But at this point a supplementary objection might be raised. The description of Dan's behaviour in the intermediate stages is so peculiar, it might be said, that our ordinary concepts of pain, hearing, sight, experience and the rest have no natural footholds, and what we say about him is largely a matter for decision rather than for 'reasonable hypotheses'. But I think this misrepresents the situation. Dan's behaviour is certainly bizarre. But, as I argued in sections II and III, the hypothesis that he has ceased to feel pain, to be able to smell things, etc., has real explanatory power. Once we have hit upon that hypothesis, it can be seen that our ordinary concepts are adequate to deal with Dan's case. Nor is there any reason to suppose that the sense in which Dan may be said to have 'ceased to feel pain' is any different from that in which perfectly normal people cease to feel pain when anaesthetized. The situation is different from those cases where, even after the most careful consideration of possible explanations, we are at a loss what to say; where no matter what hypothesis or description we adopt, we are left with a sense of conceptual strain. Take the example of lobotomized patients who say they still feel pain although it does not bother them. If we say that they do still feel pain, we are left uneasy by the oddity of a pain which does not bother the sufferer. But if, instead, we say they are not in pain, we are left with the oddity of a person who says he is in pain when he is not. But Dan's case is not relevantly analogous. In his case, once the explanatory hypothesis has been found, there is no residual sense of conceptual strain-although of course the hypothesized facts may strain credulity, and are genuinely mysterious. The main reason why there is no sense of conceptual strain, no suspicion of a change or extension or attenuation of the meanings of such words as 'pain', 'feeling', and so on, is that according to the hypothesis what appears to be pain-behaviour is no such thing: it is just a series of complicated involuntary movements. Dan is not really both avowing his sufferings, describing his headaches, and so on, and also denying that he feels any pains. All he is really doing is

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

58

R. KIRK:

the latter. So there is, on this hypothesis, no conceptually disturbing incongruity between his assumed experiences on the one hand, and what he says and does on the other. Hence Dan's case is indeed a matter for empirical hypothesis, not for arbitrary conceptual decision. But now, if this epistemological objection has no force against the proposition that someone behaving as Dan does really could lose one sense after another, is it strong enough to resist the proposition that he could become or be superseded by a Zombie? If he could lose all but one of his senses, while his behaviour assumed the patterns sketched in section III, why should he not also lose the last, while there remained something which behaved in ways indistinguishable from the normal? It would be in accordance with a prima facie reasonable hypothesis that this should be so. For the six-monthly alterations in Dan's patterns of behaviour, taken together with the subsequent apparent reversal of the process, point unambiguously in that direction. Indeed, Dan himself (on our hypothesis) formed the expectation that he would lose all his senses and become a Zombie. It would have been a comfort to him if someone could have proved that what he dreaded involved a conceptual absurdity. But does the objection we are considering provide the materials for such a proof? I shall argue that it does not. The objection lays great weight on the point that the difference between a Zombie and a man is not detectable by observation. Now it is true that my position is incompatible with the view that all actual differences must be publicly observable. But the example of Dan shows, I contend, that we could nonetheless be justified, on the basis of observations made at other times, in saying that a given individual was, or had been, a Zombie. If I am right, the view just described is mistaken. So one of the things I claim to have shown is that there are no good a priori arguments against the possibility that two things should differ in an important respect and be indistinguishable by means of observation of their current structures and states. Objections which merely assume the validity of some general epistemological view incompatible with my position therefore need to be supplemented with detailed explanations of why I have failed to show that sentience provides a counterexample. To the extent that my imaginary objector relies on verificationistic dogma, then, his point has already been met. But the objector also makes some use of the notion of a criterion, and it might be thought that here his point is valid. This, again,

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

SENTIENCE

AND BEHAVIOUR

59

seems to me to be a mistake. If the satisfaction of the criteria is held actually to entail sentience, we have already seen that there is no such entailment (sec. V). But if, as would be more usual, the satisfaction of the criteria is not held to exclude the logical possibility of insentience, then there is logical room for Zombies. To be sure, the satisfaction of the criteria gives the best possible justification, ceterisparibus, for believing that whatever description is in question does apply; but it leaves open the logical possibility that it does not apply. Naturally I agree that if a Zombie were to complete its life without any Dan-like episodes, then everyone would be justified in saying that it was a normal human being. But in such a case what we should all say (with every justification) would be false: unidentified Zombies could not truly be described as sentient, no matter what people said. My position is that in some circumstances other factors than what was currently observable could justify the hypothesis that what was indistinguishable from a man might be a Zombie. (This sort of possibility is, indeed, allowed by one use of 'criterion'.) The objector has overlooked the possibility that there are other ways of being justified in saying that two indistinguishable things are radically different than what can be gleaned from observation of their structures and states at the time in question. I conclude that the epistemological objection fails. If my assertion of the logical possibility of Zombies necessarily committed me to sceptical doubts, I should now have to take on a phalanx of further objections, including all the valid points that have been made against the intelligibility of the notion of a private language. But it is not as if the bare logical possibility of being wrong about some statement made it impossible to know that the statement was true. I may consistently claim to know that there are no Zombies even though I think it is logically possible that I am wrong about this, and I therefore see no reason to suppose that anti-private language arguments have any force against my position.

VII
[f Zombies are logically possible, then not only must analytical Behaviourism be false, but also any view (such as the Central State Materialism of Smart and Armstrong) according to which statements about sensory experience are analysable or translatable in 'topic-neutral' terms; and any view (such as those of

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

6o

R. KIRK: SENTIENCE

AND BEHAVIOUR

Feyerabend and Rorty) according to which Materialism could be validated by appropriate, if inconvenient, changes of linguistic convention or conceptual scheme. (For it is clear that Zombies would be radically different from the rest of us no matter how we talked or thought.) Indeed, it is hard to see how any intelligible version of Materialism could be reconciled with the logical possibility of Zombies, given that we are sentient. It was, after all, because the 'nomological danglers' seemed to be incompatible with the rationale of Materialism that the possibility of topicneutral analyses of sensation statements was originally mooted. But I shall not pursue that question here; nor shall I do more than assert that a recognition of the logical possibility of Zombies can help to clarify central isues in the philosophy of perception and the philosophy of action. Perhaps I have said enough to show that if it is valid, the idea of Zombies, fantastic as it is, has useful work to do.
UNIVERSITY OF NOTTINGHAM

This content downloaded on Thu, 14 Mar 2013 02:10:06 AM All use subject to JSTOR Terms and Conditions

Vous aimerez peut-être aussi