Only of a living human being and what resembles (behaves like) a living human being can one say: it has sensations; it sees; is blind; hears; is deaf; is conscious or unconscious.
WITTGENSTEIN
IT seems fair to say that there is no very general agreement on what exactly Wittgenstein has contributed to our understanding of the problem of other minds. Some will attribute this to the perplexing nature of Wittgenstein’s style, and perhaps there is some justice in this. On the other hand, it may be that the difficulties that we find in his style are partly the result of preconceptions that we bring to our reading of him. When it comes to the problem of other minds there is surely a readiness on our part to find the main lines of his position running along certain well-known paths. We expect to find some element of Cartesianism or some element of behaviourism in his position, for these seem to divide up the field without remainder. True, he may have disavowed certain consequences, such as the idea of a private language, that others thought they saw in these alternatives, but he cannot have rejected both in their entirety. Perhaps he struck a compromise by adopting elements of each. Against this way of reading Wittgenstein I will try to show that he did indeed reject both Cartesianism and behaviourism in their entirety. He rejects an element that these alternatives fundamentally share, namely, a certain way of saying what a human being is. In order to bring this element into the open, I will begin by reviewing those features of philosophical scepticism that give rise to the problems of other minds.
In his First Meditation Descartes makes clear the following features of philosophical scepticism: the sceptic is to set aside doubts about particular cases (‘Has the cat been put out?’, ‘Is the gun loaded?’) and instead is to search out grounds for calling in question an entire class of judgements. This is to be accomplished by undercutting in some way the ordinary sort of justifications we give for judgements of the class in question. Now scepticism thus understood has given rise to a set of demands that philosophers have usually tried to honour in the answers they have given to the sceptic. First, in answering the sceptic we are debarred from merely appealing to justifications of the ordinary sort (‘I looked’), for it is precisely these that he purports to have undercut. (This is what Moore seemed so often to disregard.) Secondly, if the sceptic is to be answered on his own terms and we are to progress from merely moral certainty to metaphysical certainty, as Descartes would have put it, we must begin from premisses that do not themselves have questionable presuppositions of any sort. We must find some way of grounding our ordinary judgements in what have been called ‘protocol statements’. (For simplicity of exposition I will retain this phrase, drawing on the etymological significance of ‘proto’.) Thirdly, this grounding of our ordinary judgements is to be accomplished by either (i) a justification of some extraordinary sort for making inferences from protocol statements, e.g., Descartes’ appeal to the veracity of God, or (ii) a construction (in letter if not in spirit) of our ordinary judgements out of protocol sentences by purely formal means. (I will call these the demands of scepticism.) Philosophers, as I said, have usually honoured these demands. There have been exceptions, such as Moore and Thomas Reid, but their responses to the sceptic have proved to be more puzzling than helpful. Accordingly, modern philosophy has been chiefly a contest for finding suitable ways of meeting the sceptic’s third demand. Thus, we have witnessed a succession of reductionists, on the one hand, and those they call metaphysicians, on the other. These are the lines, then, between which the skirmishes are carried on. Every so often a philosopher has tried to find middle ground, but the others call ‘Foul’ and the contest goes on with added subtleties.
This, in outline, is the background against which we read Wittgenstein. It will be well to review, then, the content of the sceptic’s demands as regards the problem of other minds. The first demand requires that we set aside our ordinary justifications for the statements we make about other people’s mental states, events, and processes, such as ‘I know she is worried; I’ve been talking to her’, ‘I could see he was in pain; he was grimacing and holding his elbow’, etc. (These must be excluded, if for no other reason, because ‘She told me’ and ‘He was grimacing’ seem to be, at least implicitly, statements of the sort the sceptic means to be calling in question.) The second demand is now the requirement that the protocol statements on which we ground any statements about other people’s mental (or ‘mental’) states, events, and processes are to be statements about human bodies. (Behaviourists sometimes talk about descriptions of ‘colourless movements’.) We might put this demand most graphically by saying that the protocol statements are to be free of any suggestion that the subjects to which they apply are essentially different from automata. The third demand is most commonly met either (i) by the argument from analogy, which is allowed to be less than what the sceptic will settle for but the best we can do if we are Cartesians, or (ii) by some form of behaviourism. Now let us ask where Wittgenstein is supposed to stand in response to the sceptic. There seem to be three interpretations: either Wittgenstein is trying to meet the third demand with his notion of criteria and is thus, despite his disclaimers, a subtle behaviourist; or he is carrying on, in a sophisticated way, Moore’s tradition of refusing to accede to the first demand and is thus what might be called an ‘ordinary language Cartesian’; or he is attempting to combine somehow these seemingly antithetical approaches and is thus perhaps the first crypto-Cartaviourist. What no one seems to have considered in all of this is what Wittgenstein has to say about the second demand and in particular the idea of ‘body’ or ‘bodily movements’ from which the whole problem begins. If in fact he advanced substantial considerations against this very root of the problem, he will have done something very different from anything suggested by current interpretations. It will be my claim in this essay that Wittgenstein struck at the root.
In order to make clear what such an approach to the problem would involve, it will be well to review the status of the second demand in the problem about the external world. There the demand is that we begin from protocol statements about sense-data or, more leniently, about appearances. I think it would now be widely conceded that the notion of sense-data is hopelessly confused and also that although we do understand and commonly make remarks about the appearances of things, these could not serve as the logical-epistemological foundation for our statements about such things as chocolate bars (‘It’s melted’) and footballs (‘It has a leak’). Some of the reasons for this can be stated briefly. First of all, it is obvious that children do not first master the language of appearances and then move on to construct or derive physical object statements. Moreover, there are good grounds for holding that there is a great deal in our physical object statements, e.g. words like ‘melted’ and ‘leak’, that could not occur in descriptions of appearances, and in any case learning the language of appearances logically presupposes a mastery of the language used in talking about physical objects. Indeed, the language of appearances is a highly sophisticated use of words. Who, after all, can easily describe hues and highlights and shadows and apparent convergence of lines and the like? And when do we take notice of such things? Children’s drawings do not suggest that they take much notice of appearances. For these and other reasons the idea that the language of appearances constitutes an epistemologically basic language has now been pretty well abandoned. One of the additional reasons for this is that we no longer find plausible those sceptical arguments, such as the argument from illusion, that seemed to create the need for—and to give us the very idea of—a protocol language. (No one thought there were sense-data before they found such arguments appealing.) I make particular mention of this point because it illustrates the essential connexion between sceptical arguments and the idea of a basic description or protocol language of the sort the sceptic demands. Thus, philosophers who would answer the sceptic on his own terms by meeting the third demand in some way share an assumption that is far more fundamental than the differences there may be between their opposing ways of meeting the third demand. In the problem of other minds this means that behaviourism and the argument from analogy are brothers under the skin: both rest upon the assumption that we are forced to recognize descriptions (or observations) of bodily movements as being epistemologically basic in our knowledge of other persons. Now it is just this assumption that Wittgenstein rejects. I refer especially to sections 281–7 of Philosophical Investigations, where he first introduces questions about bodies, souls, and human beings, and also to the way in which he follows this up in sections 288–316 with an attack on the idea of an inward or private identification of pain or thinking.1 What I want to bring out is the connexion between these two groups of passages. In order to do this, however, it will be necessary to begin by working back through the problem itself, for much of the published discussion of Wittgenstein’s views is simply the result of having got the problem of other minds badly out of focus. I will begin, then, by asking what this problem is.
II
Consider how we are to state the problem of other minds. We might ask: ‘Do other people have a mental life, as I do?’ But this clearly won’t do, for they are not people, surely, if they do not have thoughts, emotions, sensations, desires, and so on. After all, we do not mean to be asking in the ordinary way whether this or that person is in a coma or something of the sort. So we had better retreat to this formulation: ‘Are the things that I take to be people really people, that is, do they have thoughts and emotions and so on?’ But this, too, is unsatisfactory, for it is left unspecified what distinction we are being asked to make. If the question is whether they are people or not, we must ask: ‘People as opposed to what And here the answer is not at all clear. If I look at my son playing near by and ask, ‘What else might he be?’, no answer readily suggests itself. He is clearly not a statue, nor is he an animated doll of the sort we sometimes see looking very lifelike. He is my own child, my own flesh and blood.2
The problem of other minds seems to be in danger of foundering at the outset. It is clear, at least, that we cannot get the problem stated so long as we allow the concept human being (or person or child) to have its usual place. Somehow we must shunt it aside by setting some other concept over against it. Descartes sought to raise a doubt about the furniture of the world by supposing that he dreamed, and in this way he could talk not only of ships and shoes and beeswax but also of dreams of these. It is just such a move that is required if we are to launch the problem of other minds. But this move, too, ought to be found in the Meditations, for wasn’t it Descartes himself who launched the problem? Sum res cogitans. How did Descartes manage this?
He began with the following reminder about himself: ‘As though I were not a man who habitually sleeps at night and has the same impressions (or even wilder ones) in sleep as these [mad]men do when awakel’ The reminder is that he goes to sleep and dreams. But then he continues: ‘When I reflect more carefully on this, I am bewildered; and my very bewilderment confirms the idea of my being asleep.’ This provides Descartes with that challenge to his former opinions that he was looking for: he may be only dreaming that he sees and hears. It is the next sentence, however, that approaches our present problem: ‘Well, suppose I am dreaming, and these particulars, that I open my eyes, shake my head, put out my hand, are incorrect; suppose even that I have no such hand, no such body. . . .’3 Here we have the beginning of an answer to our question: with the supposition that he is dreaming Descartes sees a place to enter a wedge between himself and his body, a wedge that is driven further in the remaining Meditations. But there is a difficulty here. Descartes begins by reminding himself that he is ‘a man who habitually sleeps at night’ and dreams, and he adds that these dreams occur while ‘I am undressed and lying in bed’. This is Descartes’ beginning and the point at which we must grasp what he says. There is no difficulty, of course, in understanding at least a part of this. People go to bed, usually undressed; they sleep, calmly or restlessly, and they dream. Dreams, of course, are what people tell when they wake up or perhaps write in a diary or keep to themselves. So a dreamer here (and this includes Descartes) is a human being: he gets dressed and undressed, sleeps on a bed or pallet, tells dreams while eating breakfast, and so on. If this is what we are to understand by Descartes’ opening remark, we need not put up resistance yet. But then comes the wedge: ‘suppose even that I have no such hand, no such body’. Here we must call a halt. We were to think of Descartes as a man who, undressed and in bed, often dreams. It was only with that understanding that we were able to take his first step with him. Does this still stand? If so, what is this ‘body’ that he now supposes himself not to have? Can he, without this ‘body’, sleep, either calmly or restlessly, and dream? Or has Descartes unwittingly contradicted himself here? Has he appealed to the possibility of dreaming only to take back something that the very possibility of dreaming itself requires? This does appear to be the case. But if that is so, then we can go no further with him. Either we are to think of him—and he is to think of himself—as a man who, undressed and in bed, often dreams, and then we understand him, or he takes this back and wipes out all he has said. And this holds for the remaining Meditations, for everything that Descartes goes on to say in the Meditations is said under the supposition that he may be dreaming. Whatever sort of philosophical doubts this may raise, there is at least one thing certain; if he should ask himself ‘What am I?’, he can answer that he is a man who sleeps, undressed and in bed, and often dreams. To take back this beginning is to take back everything.
So the wedge that Descartes would drive between himself and his body is never really driven. Or rather, no place is found for the wedge to enter. For it is not that we understand about Descartes and his body. We understand only about Descartes, that philosopher who habitually undressed and went to bed at night and whose dreams, by his own testimony, were sometimes wilder than the fantasies of madmen. But this is not to say that we understand only about his body. No, to say that we understand only about Descartes is to say neither more nor less than we mean, for no place has been found yet for the word ‘body’—at least not in the special sense (if it is a sense) that Descartes requires. This is a point we tend to forget. Descartes introduced a highly extraordinary use of the word ‘body’. He has to be understood to be using it always in the context of his distinction between himself and his body. So his use of the word is not at all like these: ‘His body was covered with mosquito bites’, ‘His body was found at the bottom of the cliff’, ‘He has a strong body but no brains’, and so on. In saying such things as these we do not use ‘body’ as the one side of a Cartesian distinction. We are not saying, for instance, ‘His body, but not his mind, was covered with mosquito bites’. That would be utter nonsense. If I say that someone’s body was covered with mosquito bites, I could also say ‘He was covered, etc.’ The word ‘body’ comes in here as part of the emphasis: not just his ankles and wrists, but his back and stomach, too. Again, in speaking of a corpse we can say either ‘His body was found, etc.’ or ‘He was found dead, etc.’ The word ‘body’ in the first of these is used to make the contrast between dead and alive. No special ontology need come in here. As for the third sentence in the above list, it might be found in a requested letter of recommendation, and from it we should take the warning that the man can do heavy work but should not be expected to go at his work with much intelligence. In these and in other ordinary cases our understanding of the word ‘body’ is tied to particular contexts to a variety of particular distinctions of the kind just illustrated, and none of these provides a place to drive a conceptual wedge between Descartes and his body. But once again, this should not lead us to conclude: Then Descartes was only a body. For what distinction would that be making? It was not, after all, a corpse that wrote the Meditations
There is a bit of a clue, in what Descartes says, to how he may have failed to realize that he was introducing an extraordinary use of the word ‘body’. He says: ‘suppose even that I have no hand, no such body’, and thus it looks as though he supposed that ‘hand’ and ‘body’ are words of the same sort or that ‘hand’ and ‘body’ are related as ‘shirt’ and ‘clothing’ are, so that one could work up from supposing that you had no hand, no foot, etc. to supposing that you had no body, as you might work up from supposing you had no shirt, no coat, etc. to supposing that you had no clothing. This would perhaps be encouraged by the fact that we do use both the expressions ‘my whole body’ and ‘my whole suit of clothing’, and we also say ‘He lost a hand’ as well as ‘He lost a shirt’. But the parallel fails just where it is crucial for Descartes. I could understand, given a certain context, a man’s believing that he had no right hand, that he was a one-armed man, but I can make no sense of a man’s believing that he has no body, that he has never had a body. I might have occasion to worry about a child being born with no hands, but there is no occasion to worry about a child being born with no body. And this is not merely because bodies are required for birth. Bodies are not born; they are stillborn. What are born are babies, human beings.
We may summarize our results as follows: Descartes’ use of the word ‘body’ presupposes that he has driven his wedge, that he has provided the right sort of contrast between ‘I’ and ‘my body’, but on the other hand there seems to be no place for his wedge to be driven unless his use of the word ‘body’ is itself presupposed, and these requirements are incompatible. (If anyone should wonder about the locution ‘my body’ and ask what the body belongs to if not the mind, he need only remind himself of the locution ‘my mind’.)4
We began by trying to formulate the problem of other minds and encountered a difficulty in discovering what could be contrasted with a human being in such a way as to allow the problem to arise. In turning to Descartes we hoped to find the required contrast in his use of the word ‘body’, but it now appears that this has only further exposed the difficulty. Is the case hopeless, then? To answer this it is necessary to take notice of a reply that might be made to the foregoing arguments. The reply is this. Names of mental states, events, and processes, including the word ‘dreaming’, get their meaning from private ostensive definitions. For this reason it does not follow that if Descartes begins from the reminder that he dreams, he must allow ever after that he is a man who goes to bed and sleeps. To speak of dreaming carries no such implication, for the state we call ‘dreaming’ is something known to us by means of introspection or inner sense, and introspection discloses nothing of a bodily nature.5
It is to this account of words like ‘dream’ and ‘pain’ and ‘thinking’ that we must ultimately trace the problem of other minds. Putting the matter in unabashed metaphor, it is the idea that since the inner sense that reveals our mental states does not discover anything bodily, it must be possible to conceptually skim off a mental side of our nature leaving a physical remainder called ‘the body’.6 It is this idea of a physical remainder, the ‘senseless body’ which some mind may ‘have’, that gives us our problem, for it is a consequence of this idea that when we look at another person all that we really see is something that, in itself, is no more an appropriate subject of pain or thought than a stone is. Philosophers have puzzled over the question ‘Why shouldn’t we regard a complicated automaton as we do a person’, but what ought to puzzle us is the question ‘If all that we see of other “people” are “senseless bodies”, how could we have got as far as connecting the concepts of thought and sensation with them at all?’ The argument from analogy should not impress us here unless, as Wittgenstein saw (283), we are willing to go a step further and allow that perhaps stones have pains and machines think. For if I identify pain and thinking inwardly, if I do not learn these concepts in learning a common language, then my concepts pain and thinking are not essentially related to living human beings (in the ordinary sense), and so my body might turn to stone or into a pillar of salt while my pain continues. But in that case I should allow that the pebbles I walk on may be in pain, too. It would be gratuitous to restrict the concept to human beings and to what more or less resemble (behave like) them.7 But the real difficulty here is not to account for our restricting our concept of pain to human beings but to account for how we extend it beyond our own case. For on the supposition we are here considering (that I learn what ‘pain’ means from my own pains) it will not do to say that I extend the concept to others by simply supposing that they sometimes have the same thing that I have so often had, since this explanation presupposes the very use of words, namely, ‘same sensation’, that we ought to be explaining (350–2). But this means that I could never get as far as using the argument from analogy or anything like it. I could not even understand the question whether there are other beings that feel what I call ‘pain’. The only recourse here is to admit that I cannot extend the use of ‘pain’ from myself to others and to hold to a strict logical behaviourism: when I say that other people are in pain, I am merely speaking of the movements of those senseless bodies that I see. What this would fail to account for, of course, is my pity or concern for them. Since I can no more think of my children as suffering, in the sense that applies to me, than I could think this of a stone, my pity for them should strike me as a logical incongruity. It’s as if I were to fall passionately in love with a fleck of dust.
It should now be possible to get an understanding of Wittgenstein’s discussion of the problem of other minds. In particular, we can see the connexion between those passages (281–7) in which he first introduces questions about bodies, souls, and human beings and the next group of passages (288–316) in which he attacks the idea of an inward or private identification of mental states, events, and processes. The essential point is that if there is confusion in the idea of an inward, private, identification, then there is also confusion in the idea of conceptually skimming off a mental side of our nature, leaving a physical remainder called ‘the body’. The philosophical idea of a ‘senseless body’ must be dropped. But in that case we must also reject the idea that when we look at another person we see only a ‘body’, i.e., something which is no more a possible subject of pain or thinking than a stone would be. And finally, in rejecting that idea, we eliminate the only grounds of scepticism with regard to other ‘minds’ and in this way eliminate, too, the only source of the plausibility of behaviourism. In short, by rejecting the idea of a private identification, we get back our ordinary concept of a living human being. In place of ‘colourless bodily movements’ we now have human actions and reactions; we are back in the world of people running from danger, telling us their woes, nursing painful bruises, grimacing, frowning in disapproval, and so on. Thus, Wittgenstein’s primary contribution to the problem of other minds was his attack on the idea of a private language, of a private identification of mental states and processes. Although he has a number of other important things to say about such words as ‘pain’ and ‘thinking’, these cannot be understood apart from a grasp of his primary contribution. His reminders about such words will be of no use until we have eliminated the philosophical notion of ‘body’ and have brought human beings back into our discussions. I will return later on to consider some of these reminders and in particular to consider the objection that may now have occurred to the reader, that it must be a question-begging move to make the concept human beings primary in any account of mental predicates.
At the beginning of this essay I remarked that our difficulty in understanding Wittgenstein’s contribution to the problem of other minds might be the result of preconceptions that we bring to our reading of him. These preconceptions should now be clear. The philosophical ideas of ‘body’ and ‘bodily movement’ have simply become unquestioned notions; they set for us what we take to be the problem of other minds. The problem, as we understand it, is that of grounding or justifying our ascriptions of mental states and processes on observations of bodily movements. Thus, when Wittgenstein speaks of ‘behaviour’, we inevitably read into this our own concession to the sceptic; we think of Wittgenstein as trying to solve the same problem that others have tried to solve with behaviourism or the argument from analogy, only it is not clear what his own solution comes to. In struggling with this people have seen that he allows no place for the argument from analogy or at least that he makes no appeal to such an argument, and this has given rise to suggestions that, despite all he says to the contrary, Wittgenstein settled for some form of behaviourism. In defence of this suggestion interpreters have fastened on his concept of criteria and have argued that he puts forth a ‘criteriological’ theory of meaning that could not amount to anything but a subtle form of behaviourism. I believe that I have already given sufficient reasons for dismissing this interpretation, but because the misunderstandings about the rôle of criteria have run so deep, I will digress from my main topic to say something about this.
The view that I want to discuss is that Wittgenstein came to talk about criteria as a means of solving the problem of other minds, where that problem is understood in the usual way as a problem of our knowledge of other minds. This interpretation has been set forth in detail in a recent article by C. S. Chihara and J. A. Fodor,8 and it will be useful to look carefully at what they say. They first give us their account of Wittgenstein’s aims in the following passage:
Among the philosophical problems Wittgenstein attempted to dissolve is the ‘problem of other minds’. One aspect of this hoary problem is the question: What justification, if any, can be given for the claim that one can tell, on the basis of someone’s behavior, that he is in a certain mental state? To this question, the sceptic answers: No good justification at all.
Just what this claim is that needs justification is not explained to us at this point, but apparently it is to be thought of as a claim made in answer to the sceptic. Of course, whether it is a claim that one should make will depend upon what it is that the sceptic denies and why he denies it. Is he thinking (quite correctly) that the movements of a ‘senseless body’ would provide no logical foothold for words like ‘pain’ and ‘thinking’ and then supposing (quite incorrectly) that all we see of other ‘persons’ are such ‘bodily movements’? Apparently Chihara and Fodor are so oblivious to there being any difficulty about this philosophical idea of body that they do not even see it as the challengeable point in the sceptic’s position, which on their view seems to arise out of nothing. They characterize scepticism as follows:
It is assumed as a premiss that there are no logical or conceptual relations between propositions about mental states and propositions about behavior in virtue of which propositions asserting that a person behaves in a certain way provide support, grounds, or justification for ascribing the mental states to that person. From this the sceptic deduces that he has no compelling reason for supposing that any person other than himself is ever truly said to feel pains, draw inferences, have motives, etc.9
Now because these authors do not see that scepticism arises out of a challengeable notion of ‘body’, they can only understand Wittgenstein to be meeting the sceptic head on. ‘Wittgenstein’s way of dealing with the sceptic,’ they tell us, ‘is to attack his premiss by trying to show that there do exist conceptual relations between statements about behavior and statements about mental events, processes, and states.’10 We are also told that on Wittgenstein’s view there is a ‘logical connexion… between pain behaviour and pain’ and that ‘Wittgenstein used the term “criterion” to mark this special connexion’.11 The position thus described is then further characterized in the following way: ‘To hold that the sceptical premiss is false is ipso facto to commit oneself to some version of logical behaviourism where by “logical behaviourism” we mean the doctrine that there are logical or conceptual relations of the sort denied by the sceptical premiss.’12 In a footnote to this sentence Chihara and Fodor offer the following, highly significant, justification for using the word ‘behaviourism’ in the classification of Wittgenstein’s philosophical position:
… insofar as C. L. Hull can be classified as a behaviorist, there does seem to be grounds for our classification. Hull’s view, as we understand it, is that mental predicates are in no sense ‘eliminable’ in favor of behavioral predicates, but that it is a condition upon their coherent employment that they be severally related to behavioral predicates and that some of these relations be logical rather than empirical—a view that is strikingly similar to the one we attribute to Wittgenstein. Cf. C. L. Hull, Principles of behavior (New York, 1943).
This comparison of Wittgenstein with Hull is significant for the reason that it shows most clearly what Chihara and Fodor understand by the word ‘behaviour’, which they use in describing the positions of both men. The authors are surely aware that in Principles of Behavior what is thought of as ‘behaviour’ is what Hull calls (in contradistinction to purposive action) ‘colourless movements’.13 Hull, in other words, is engaged in an attempt to respond to the sceptic’s demands on the sceptic’s own terms. So one point that becomes clear in this comparison of Wittgenstein with Hull is that these authors think of Wittgenstein as using his notion of criteria to connect mental states with ‘bodies’. A second point that begins to emerge here will be seen if one recalls that it is Hull’s stated aim to begin from ‘colourless movements and mere receptor impulses as such’ and build up (or ‘deduce’) such concepts as purposive action, intelligence, intention and other mental verbs and predicates. Now Chihara and Fodor see clearly enough that Wittgenstein did not think that there was a deductive relation to be found here, but it is also clear that they believe that he was engaged in the same programme as Hull and differing from him, perhaps, only in settling for a looser logical relation. So the second point that emerges from the comparison is that they think of Wittgenstein as having held an empiricist theory of ‘concept formation’, which they describe as the view that in learning mental predicates we are learning ‘criterial connexions which map these terms severally onto characteristic patterns of behaviour’.14 In section II of their essay they assert that Wittgenstein held ‘an operationalistic view of the meaning of certain sorts of predicates’, including such words as ‘pain’, ‘motive’, and ‘dream’. Since this is a highly implausible interpretation of Wittgenstein, given his general remarks about language, it is important to see that their sole reason for thinking that Wittgenstein held such a theory of language is their belief that he meant to be replying to scepticism on the sceptic’s own terms (‘bodily movements’) and that he introduced his concept of criteria in order to accomplish this. Wittgenstein’s criteria, on their view, are quite particular patterns of behaviour that are related, in some way, to the meaning of a word such as ‘pain’, where ‘behaviour’, once again, is thought of as being ‘bodily movements’.
I want now to show that this interpretation is wrong on both counts. That Wittgenstein did not introduce his concept of criteria in order to answer the sceptic will have to be shown by making clear the quite different problem for which he did introduce it. I will come to this presently. As for the other part of the interpretation, we have already seen reason to reject the idea that Wittgenstein meant to answer the sceptic on his own terms and so thought of behaviour as ‘bodily movements’. Indeed, it is by rejecting this very notion of ‘body’ and ‘bodily movement’ that Wittgenstein undercuts the whole problem. It is true that he sometimes speaks of certain behaviour as being a criterion for something, but we must see whether what he counts as behaviour is anything that the sceptic would be prepared to allow. In his discussion of ‘saying something to oneself in the imagination’ Wittgenstein speaks of criteria in the following passage: ‘Our criterion for someone’s saying something to himself is what he tells us and the rest of his behaviour; and we only say that someone speaks to himself if, in the ordinary sense of the words, he can speak And we do not say it of a parrot; nor of a gramophone’ (344). Here what Wittgenstein calls ‘our criterion’ includes someone telling us something, and it is clear from the contrast made with parrots and gramophones that he is thinking of this as a human action and not merely certain sounds emanating from some ‘senseless body’ (whether the shape be human or otherwise). Also what the person tells us might be: ‘When I saw him enter the room, I said to myself…,’ and surely the sceptic does not want to concede at the outset that ‘persons’ (or ‘bodies’) see. Moreover, Wittgenstein’s specification here includes that the person in question speaks a language, and this means that he does such things as answering questions, giving orders, telling jokes, giving directions, asking advice, stating his business, giving lectures, confessing his ignorance, complaining of aches and pains, and so on. In short, before something can be understood as a criterion for a man’s saying something to himself, there must already be known about him all of the sorts of things that the sceptic means to be calling in question. So by ‘behaviour’ Wittgenstein cannot have meant anything like what the sceptic must mean. But in that case reference to criteria cannot be relevant to scepticism. This point will be even clearer if we notice that in some cases Wittgenstein even counts sensations and thoughts as criteria. In one passage (160) he presents two cases of which he asks: ‘Should we here allow his sensations to count as a criterion for his reading or not reading?’ In the first of these cases it is clear that the man’s sensations would be irrelevant, but in the second case this is not so. Elsewhere Wittgenstein gives an example in which a man’s having ‘thought of the formula’ is both his and our justification (criterion) for saying that he knew how to continue the expansion of a series. Then this explanation is added: ‘The words “Now I know how to go on” were correctly used when he thought of the formula; that is, given such circumstances as that he had learnt algebra, had used such formulae before’ (179). Now if thoughts and sensations are among the things that Wittgenstein will count as criteria, then not only are criteria not at all the sort of thing that it would be relevant to mention in replying to the sceptic but also criteria are not in every case behaviour, even in the ordinary sense of the word. There seem to be no grounds here for thinking of Wittgenstein as a behaviourist.
It should by now be clear that it must have been some quite different problem from the problem of other minds that led Wittgenstein to introduce his concept of criteria. And in fact it is not difficult to discover what that problem is. Since much of the misunderstanding has arisen from that passage in The Blue Book (pp. 24–5) in which Wittgenstein first introduced the pair of terms ‘symptom’ and ‘criterion’, it will be well to begin by looking there. He tells us that he is introducing these terms ‘in order to avoid certain elementary confusions’, and from the context of the few pages immediately preceding and following this remark it is clear that these are confusions that arise out of asking ‘What is expecting?’ or ‘What is knowledge?’ or ‘What is time?’ and so on while hoping to ‘find some common element’ in all of the applications of the general term. Here is Wittgenstein’s account of the difficulty:
We said that it was a way of examining the grammar (the use) of the word ‘to know’ to ask ourselves what, in the particular case we are examining, we should call ‘getting to know’. There is a temptation to think that this question is only vaguely relevant, if relevant at all, to the question: ‘what is the meaning of the word “to know”?’ We seem to be on a side-track when we ask the question ‘What is it like in this case “to get to know”?’ But this question really is a question concerning the grammar of ‘to know’. …
When, several paragraphs later, Wittgenstein explains the words ‘criterion’ and ‘symptom’ it is clear that he has two purposes in mind. The first is to use this pair of terms to characterize the idea that there must be a common element in all cases of the application of a general term; the second is to characterize his own objection to that idea. The first of these comes to this: we have the idea that there is a ‘law in the way a word is used’ (p. 27), some single test for deciding the applicability of the general term to particular cases. We can call this ‘the defining criterion’. Now if there were such a defining criterion, then whatever else might be true of the various particular cases will be only indirectly relevant to deciding whether the general term is applicable. Such a piece of evidence can be called ‘a symptom’. Now the philosophical idea that there must always be a common element in virtue of which a general term applies can be described as the idea that for such words as ‘expecting’ and ‘knowledge’ there should be a defining criterion. It is because we have this idea that we are tempted to say, when particular cases are brought up for consideration, that the varying details of those cases are irrelevant to the question ‘What is expecting?’ We think the details, since they are not common to all cases, are only symptoms and so have no bearing on the grammar of ‘to expect’.—And now I think we can see the rationale for introducing this pair of terms. First of all, the claim that in the various particular cases of expecting we can discover only symptoms is self-contradictory, for how could we set out to search through these cases for that common element which we complain of not finding if we did not know that they were cases of expecting. If there must be a defining criterion, and we admit to not having discovered it (but only ‘symptoms’), then we should also allow that we do not even know whether we have been considering cases of expecting. But this is absurd. Accordingly, since we did know that these were cases of expecting, it must be that the idea of a defining criterion is a piece of confusion.15 Our demand that we find the ‘common element’ is
a very one-sided way of looking at language. In practice we very rarely use language as such a calculus … We [in philosophizing] are unable clearly to circumscribe the concepts we use; not because we don’t know their real definition, but because there is no real ‘definition’ to them. To suppose there must be would be like supposing that whenever children play with a ball they play a game according to strict rules’ (p. 25).
Not only do we not ‘use language according to strict rules—it hasn’t been taught us by means of strict rules, either’. But if the idea of a ‘defining criterion’ is a bogus notion, then in considering the details of particular cases we are not considering merely ‘symptoms’. These details, which vary from one case of expecting or knowing to another, can show us something about the grammar of the words. That is, it is useful in philosophy to ask ourselves what in this case and in that case would enable us to recognize (would justify us in saying) that someone was expecting a visitor. In one case it might be that the man is pacing up and down the room, occasionally looking at his watch, and at the same time talking about how good it will be to see his old friend again. In another case it might be that, although he is not thinking about the expected visitor at all, he has the man’s name on his appointment calendar and has laid out certain items in preparation for the man’s visit. That is, if we were asked how we knew that this person expected a visitor, these are the sorts of things we would mention. The relevant point, then, is that there is not some one thing, the same in all cases, that justifies our use of the word ‘expecting’, and if anything has a right to be called a ‘criterion’ it is (Wittgenstein seems to imply) such details of particular cases that we would find it relevant to take notice of in our everyday life.
It is worth remarking, perhaps, since the opposite has been so often supposed, that Wittgenstein nowhere suggests that a philosopher, by using this concept of criteria, can bring out everything that is of philosophical interest in the use of a word. It is only the misguided inclination to see Wittgenstein as a behaviourist that would lead one to think that he intended to offer an ‘operational analysis’ of the meaning of a word like ‘expecting’. The words ‘symptom’ and ‘criterion’ were introduced as a means of dealing with a certain philosophical problem, as a way of breaking the hold of a certain picture of the workings of language, namely, the idea that words are learned and used according to strict rules and that accordingly we can reject as irrelevant the varying details of particular cases when asking, for example, ‘What is expecting?’ In Philosophical Investigations he expresses this idea by saying that we think there is ‘nothing at all but symptoms’ (354). This is explained in an earlier passage as follows:
In case (162) the meaning of the word ‘to derive’ stood out clearly. But we told ourselves that this was only a quite special case of deriving; deriving in a quite special garb, which had to be stripped from it if we wanted to see the essence of deriving. So we stripped those particular coverings off; but then deriving itself disappeared. —In order to find the real artichoke, we divested it of its leaves. For certainly (162) was a special case of deriving; what is essential to deriving, however, was not hidden beneath the surface of this case, but this ‘surface’ was one case out of a family of cases of deriving.
And in the same way we also use the word ‘to read’ for a family of cases. And in different circumstances we apply different criteria for a person’s reading. (164)
One point to gather from this passage is that in so far as Wittgenstein uses the concept of criteria to oppose the notion of ‘the hidden’, this is not the notion that arises in the problem of other minds, the problem that grows out of Descartes’ metaphysical use of ‘body’ but rather that notion of the hidden that arises out of looking for a common element and finding none. I take it that Wittgenstein’s opposition to this notion of ‘the hidden’ does not make him a behaviourist.
Several misunderstandings can be cleared up by noticing how the word ‘criterion’ comes into Wittgenstein’s treatment of two related confusions that can grow out of the search for a common element in all cases of expecting or understanding or thinking. The first of these can be roughly indicated by considering the word ‘calculate’. We use this word to say what someone is doing when he is using paper and pencil to calculate how much material will be needed for a building job, but there are also cases in which we say that someone has calculated although nothing is written down or said aloud. The latter we sometimes speak of as ‘calculating in the head’. Now if we expect to find in cases of the two sorts a common element which will be what calculating really is, we will immediately want to say that the use of paper and pencil is no part of the essence, since this is not common to the two sorts of cases. We will want to say that the essence of calculating is some quite particular mental process to which the learning of arithmetic in the usual way, i.e., with written and spoken calculations, is inessential. (See 385 and 344.) Now it is cases of this sort that Wittgenstein has in mind when he says that ‘the fluctuation in grammar between criteria and symptoms makes it look as if there were nothing at all but symptoms’ (354). That is, because we learn the word ‘multiply’, for example, in cases in which someone, teaching us, works out problems on paper (in which case it is indifferent to us whether he uses his results for anything further on these occasions), and because later on, when we speak of a man’s calculating in his head, it is not indifferent whether he uses the result (here a criterion would be his setting his saw in a certain place or drawing a line or gathering up the right amount of material after taking some measurements and pausing to think), we have a ‘fluctuation’ of symptoms and criteria. We keep the same word, ‘calculate’, even though what was a criterion in the first sort of case drops out in the second. Noticing this, we may think that, because it is sometimes possible to calculate without writing or speaking, it would have been possible to calculate had we never had a written or spoken language (arithmetic). Moreover, this same idea will occur to us with respect to concepts other than calculating, for we have similar shifts in the following pairs of cases: ‘speaking(aloud)’ and ‘speaking to oneself in the imagination’ (344–8); the use of ‘expecting’ both for cases in which the expectant person is thinking and talking about what he expects and for cases in which he is not (572–8 3); and the use of ‘thinking’ both for cases in which the thinker is writing or talking and for cases in which he is neither writing nor talking (318–42). In each of these cases we may be tempted to suppose that the essence of speaking, expecting, or thinking is something that could be identified apart from the mastery of language, and as a result we get such questions as ‘Could a deaf mute who had learned no language still speak to himself in the imagination?’ (348), ‘Can a machine think?’ (359–360), and so on. Now it is important to notice exactly the way in which Wittgenstein treats this kind of idea. He does not do so simply by saying that there are certain criteria for thinking, etc. He does not, that is, argue in the manner one would expect him to if he thought of criteria in the way described by Chihara and Fodor. Instead, there are a number of quite different considerations brought to bear on the problem. One is simply the use of analogies (316, 365). Another is to call into question the premise ‘What sometimes happens might always happen’ by giving some clear counter examples (345: see also p. 227). Still another is to remind us that a misleading metaphor can lead us to think that something that is logically primary in a use of words is really inessential (354–6).
The second and related confusion in connexion with which Wittgenstein uses the concept of criteria can be seen by comparing the following passages:
When we do philosophy, we should like to hypostatize feelings where there are none. They serve to explain our thoughts to us. (598)
… We are tempted to say: the one real criterion for anybody’s reading is the conscious act of reading, of reading the sounds off from the letter. (159)
In order to get clear about the meaning of the word ‘think’ we watch ourselves while we think; what we observe will be what the word means!—But this concept is not used like that. (316)
‘But you surely cannot deny that, for example, in remembering, an inner process takes place.’— … When one says ‘Still, an inner process does take place here’—one wants to go on: ‘After all, you see it’. And it is this inner process that one means by the word ‘remembering’. (305)
How should we counter someone who told us that with him understanding was an inner process?—How would we counter him if he said that with him knowing how to play chess was an inner process? —We should say that when we want to know if he can play chess we aren’t interested in anything that goes on inside him.—And if he replies that this is in fact just what we are interested in, that is, we are interested in whether he can play chess—then we shall have to draw his attention to the criteria which would demonstrate his capacity, and on the other hand to the criteria for the ‘inner states’, (p. 181)
What Wittgenstein is suggesting in these passages is that we tend to assimilate concepts of one sort to concepts of a quite different sort, and this comes about in the following way. Since a person can read or think or remember or understand something without, on the particular occasion, saying anything, we are tempted to exclude the mastery of language from consideration when asking what thinking is, what understanding is, etc. Accordingly, in our search for the essence of each of these we gravitate towards a kind of concept, namely, sensation, that can be applied to a subject that has no mastery of language. (Brutes and infants can have sensations.) Putting the matter in another way, we concentrate on the ‘silent’ cases of thinking or understanding or remembering and ‘look into ourselves’ for the essential element, i.e., take notice of feelings, images, words going through our head, and so on. In this way we come to assimilate concepts like thinking, understanding, remembering, and so on to sensation words. (See p. 231.) Now one of Wittgenstein’s ways of opposing this is to ask us to compare, for example, the criteria for understanding or knowing how to play chess and the criteria for mental states (pp. 59 and 181). What will come out of such an investigation are such things as the following. We might say to someone, ‘Raise your finger the moment the pain stops’, but not, ‘Raise your finger the moment you no longer know how to play chess’. At the very least there will be this difference: to be able to say whether the pain has stopped I need not try to do anything, whereas I might need to try reciting some rules of chess or try making a few moves in order to learn whether I could still play chess. (Contrast ‘in pain’ with ‘sore to the touch’.) Again I could confidently plan to deceive someone into thinking that I am in pain, i.e. plan to act the part without being in pain, but I could not, by acting the part of a chess player, deceive someone who knew the game into thinking that I could play chess—or at least I could not do so without the aid of a confederate giving me signals. Or again, I might be shown by someone that I only imagined that I understood a certain word, that the explanation I would have given of it or the use I would have made of it is actually confused, whereas I could not be shown by someone that my arthritic fingers do not really hurt, that I haven’t been in pain at all. Now a misunderstanding can arise about Wittgenstein’s use of such considerations as these. For it might look as though he meant to deny that thoughts and images are ever essential to playing chess, understanding, remembering, and so on. But this is not at all what he means to deny. He is saying rather that it requires certain surroundings in order for there to be such a thing as the thoughts of a chess player or for an image that I have to be the image of the expression on a certain man’s face. His point is succinctly illustrated in the following passage: ‘The words “Now I know how to go on” were correctly used when he thought of the formula: that is, given such circumstances as that he had learnt algebra, had used such formulae before’ (179). In such a case if the man had not thought of the formula, he might not have known how to go on with the series of numbers. But Wittgenstein’s point is that it is not from an inventory of the man’s present mental contents that either he or someone else can see that he knows some algebra or that he even knows how to count. Or rather, since his present mental contents include his thought of the formula, i.e., something algebraic, we had better say that what is ‘present’ in this sense is essentially connected with what has gone before. (See p. 174.) A thought, then, is nothing like a sensation.
IV
Let us return now to the problem of other minds. At the end of section II I suggested that Wittgenstein’s rejection of the Cartesian notion of ‘body’ and his insistence on making the concept human being primary might strike someone as being a question-begging move. One may want to know how we get that concept in the first place and what right Wittgenstein has to inject it into the discussion of the problem without providing a justification. This is the point I want now to speak to, and I will approach it through a brief discussion of the argument from analogy.
Mill, in his classic statement of this argument, says: ‘I must either believe them [other human beings] to be alive, or to be automatons’,16 and he explains how, by means of an analogical inference, he ‘concludes’ that they are alive and are like him in their sensations and emotions. Now what is odd about Mill’s argument is that it is difficult to see at what point the supposed analogical inference would have any work to do. At what point in my life is it supposed to have been an open question for me whether my friends and family are people or something else? Disregarding the puzzle about what else they might be, let us ask: When was I in need of the argument from analogy? When did I suffer from ignorance or doubt of the sort that that argument is supposed to remove? Is the analogical inference thought to assist us out of some state that we are in only in earliest childhood or out of some state of doubt that may beset a person in his adult years? The latter alternative surely cannot be the right one. There is not an undercurrent of uneasiness that runs through all my various encounters with other people tempting me to recoil from them in horror or suspicion. I do not, for instance, suffer queer feelings that my children may be altogether unlike me in some essential respect. When one of them comes crying to me with a bumped head or a bleeding foot, I do not gaze wonderingly at the child, thinking: What can be happening here in this thrashing, noisy thing? And when I speak with people I do not feel foolish in the thought that my remarks may be only activating circuits in them or something of the sort. Although there may be some such form of insanity, some kind of dissociation, in which the victim simply cannot find his feet with any other human being, this is not a condition from which very many of us suffer. And yet if it is thought that we all stand in need of the benefits supposedly conferred by the argument from analogy, then it seems that we are all being represented as suffering from this form of insanity. But even the proponents of the argument from analogy do not believe this, and I conclude, therefore, that the argument could have no relevance for us in our adult years. Perhaps, then, the analogical inference is to be thought of as having its place in our childhood and as being our means of coming to understand others as human beings in the first place. This, if I understand them, is something like the view set forth by Chihara and Fodor. They maintain that as children we encounter in others complicated syndromes of behaviour (bodily movements?), for which we want some explanation. For instance, we repeatedly encounter the ‘pain syndrome’ and are ‘in need of an explanation of the reliability and fruitfulness of this syndrome, an explanation which reference to the occurrence of pain supplies’.17 And this being so, they reason, the ‘application of ordinary language psychological terms on the basis of behaviour’ is to be thought of ‘as theoretical inferences to underlying mental occurrences’.18 This is offered as an account of how we come to use psychological terms in speaking of other people, and the authors dismiss the objection that their account saddles the child with too great an intellectual burden by appealing to the fact that we all learn a language with a complicated grammatical structure and do so, apparently by natural capacities the exercise of which involves ‘the use of an intricate system of linguistic rules of very considerable generality and complexity’, which system of rules is not explicitly taught to us.19 This is the account that they offer as an answer to the sceptic and as an alternative to behaviourism. What they are offering, then, is an account of how the child passes from an earlier stage when he sees his parents and others as mere ‘bodies’ moving about to seeing them as human beings.
Before commenting on this I should like to contrast it with the quite different account that Wittgenstein gives of the way in which we come to use a word such as ‘pain’ in speaking of other human beings. First of all, throughout Philosophical Investigations Wittgenstein opposes the idea that in learning language there is somehow imparted to us an understanding that logically compels us to use a given word in a certain way. In place of this Wittgenstein puts the suggestion that learning language depends on certain ‘normal learners’ reactions’ (143–5), that is, certain primitive responses that get encouraged and developed as one learns words. In one place (p. 224) he speaks of language-games arising as ‘something spontaneous’, and there is one passage (310) that suggests that our using words like ‘pain’ and ‘hurt’ in speaking of others begins with something ‘instinctive’. (See also his use of the word ‘attitude’ in 284 and on p. 178.) In Zettel these suggestions are made more explicit.
It is a help here to remember that it is a primitive reaction to tend, to treat, the part that hurts when someone else is in pain; and not merely when oneself is—and so to pay attention to other people’s pain-behaviour … (540)
But what is the word ‘primitive’ meant to say here? Presumably that this sort of behaviour is pre-linguistic that a language-game is based on it, that it is the prototype of a way of thinking and not the result of thought. (541)
… Being sure that someone else is in pain, doubting whether he is, and so on, are so many natural, instinctive, kinds of behaviour towards other human beings, and our language is merely an auxiliary to, and further extension of, this relation. Our language-game is an extension of primitive behaviour. (For our language-game is behaviour.) (Instinct.) (545)
What makes it possible for Wittgenstein to give this account, it seems to me, is that he, unlike the behaviourist or Chihara and Fodor, rejects the idea that what the child is initially confronted with are ‘bodies’ that he somehow comes to see as human beings.
Behind the account offered by Chihara and Fodor lies an unspoken assumption that it is somehow natural for us in our infancy to see human beings as ‘bodies’, mere things, and this sends these authors in search of an account of the way in which we pass from seeing them in that way to seeing them as living human beings. But why should we think that the child is initially ‘set’ for seeing people as things? Indeed, what does it mean to speak of seeing a person as a thing? Where would we say that this has happened? This is something that we would describe in this way that might happen to an adult. I might have an uncanny experience in which I see a number of people as automata when they all begin to perform some task at the sound of a bell. Or perhaps this experience could occur as I stand observing the rush hour crowds pushing their way, stony-faced, along sidewalks and into subway entrances. (The subway tunnel is a great maw devouring spent machinery.) What this would involve, however, is something that is not yet there in the child. In these uncanny experiences I imagine a particular surrounding for what I see; I expect in these people what I have learned to expect of machines. For instance, if another bell rings, they will all cease work simultaneously and remain motionless until some other signal is given. Or in the rush hour crowd I perhaps look at the people’s legs to discover the mechanical nature of their movements or stare into their faces, thinking ‘Unseeing eyes’. And then the spell is broken when someone speaks to me.—Now if this is seeing a person as a thing, then, as Wittgenstein says, ‘the substratum of this experience is the mastery of a technique’ (p. 208). That is, I must have learned considerable language in order to see human beings as machines. Therefore, this cannot be the account we should give of the human infant’s way of seeing his parents. But of course it would also be wrong to say that infants see their parents and others as human beings, that they come ‘set’ in this way. For what does it mean to speak of seeing something as a human being? We might say this of an adult who is watching an ingeniously made humanlike automaton and has an urge to go up to it and ask a question or something of the sort. But we speak of ‘seeing as’ here just because if the machine were to give a few clicks and stop, the spell would be broken; the person would not try to wake it up nor would he treat it as he would a corpse. So this is an account to give of someone who knows what a machine is, and we do not want to give that account of the child learning language.
Now Wittgenstein’s account escapes these difficulties, I believe, just because it does not saddle the child at the outset with that philosophical notion of ‘body’ which has called into existence both behaviourism and the argument from analogy. Behaviourism is left with the logical incongruity of the child’s laughing with and pitying those ‘bodies’ that it sees around it, while Chihara and Fodor trade this logical incongruity for the logical anachronism of getting the child over a philosophical hurdle before he is out of the cradle. Wittgenstein, by both rejecting the philosophical idea of ‘body’ and by tracing language back to primitive responses, can allow the child’s laughing with others and pitying them and so on without either the incongruity or the anachronism. And now I think we can see how Wittgenstein would reply to the charge that he has introduced the concept human beings without justification and so has begged the sceptic’s question. He speaks to this point in another connexion when he writes: ‘What we have rather to do is accept the everyday language-game, and note false accounts of the matter as false. The primitive language-game which children are taught needs no justification; attempts at justification need to be rejected’ (p. 200). What I have been trying to show in this essay is the way in which the problem of other minds arises from a false account of the matter, of our language-game. It is this false account that leads to the idea that we all hold some unsupported belief, and this idea, in turn, gives rise to the demand for a justification. Wittgenstein’s alternative to this is summed up in his remark: ‘My attitude towards him is an attitude towards a soul. I am not of the opinion that he has a soul’ (p. 178).
V
If a human being is thought of as consisting of a ‘senseless body’ plus mental entities (or plus a mind with mental states), the logical category of mental states will be seriously misrepresented. They cannot, on this view, be states of a living organism and must be thought of in an altogether different way. What this difference is must be made clear if we are to understand Wittgenstein’s rejection of the idea that sensations are private objects. The important point here is that if sensations are not thought of as being states of a living organism, then it will be impossible to think of them as having a natural expression in the behaviour of living organisms. Instead, we will think of sensations, as most philosophers since Descartes have, as being objects perceived by means of ‘inner sense’. Accordingly, words such as ‘pain’ and ‘dizziness’ will not be thought of as tied up with the natural expressions of sensation; they will be thought of as being names for objects that the speaker alone can perceive. Now this shift from ‘state of a living organism’ to ‘object of inner perception’ is the subject of Wittgenstein’s well-known beetle-in-the-box passage (293), which concludes with the remark that ‘if we construe the grammar of the expression of sensation on the model of “object and name” the object drops out of consideration as irrelevant’. This passage has been misunderstood by people who have failed to appreciate Wittgenstein’s insistence that sensations are states of living organisms; they have taken him to be saying that sensations do drop out of the language-game as irrelevant.20 Yet Wittgenstein explains his meaning when he says that a sensation ‘is not a something but not a nothing either’ and explains that in saying this he has ‘only rejected the grammar which tries to force itself on us here’ (304). These passages are often taken to mean that a word like ‘pain’ can have a public use despite the privacy of sensations. But such an interpretation is wrong; Wittgenstein is rejecting a metaphysical account of what sensations are. When he remarks that he is rejecting ‘the grammar which tries to force itself on us here’, we ought to connect this with his remark further on that ‘grammar tells us what kind of object anything is’ (373). When he says that a sensation is not a something, he is saying that a sensation is not an object of ‘inner sense’, is not something having its essential characteristics apart from a living organism.
What is at issue here can be seen more clearly by recalling Hume’s account of mental states and events, where the grammatical transformation that Wittgenstein is rejecting is made explicit. Hume remarked that if ‘any one shou’d [say] … that the definition of a substance is something that may exist by itself; … I shou’d observe that this definition agrees to everything that can possibly be conceiv’d’.21 That Hume thought of mental states and events (which he called ‘perceptions’) as being substances in this sense is shown by his remark that ‘as every perception is distinguishable from another, and may be consider’d as separately existent; it evidently follows that there is no absurdity in separating any particular perception from the mind; that is, in breaking off all its relations with that connected mass of perceptions, which constitute a thinking being’.22 Thoughts, images, desires, sensations, and so on ‘have no need of anything else to support their existence’.23 This leads Hume to suggest the following thought-experiment as a means of discovering what a ‘thinking being’ or ‘self’ is:
We can conceive of a thinking being to have either many or few perceptions. … Suppose it to have only one perception, as of thirst or hunger. Consider it in that situation. Do you conceive of anything but merely that perception? Have you any notion of self or substance? or For my part, I have a notion of neither…24
If this passage were read apart from the context of Hume’s remarks about ‘perceptions’ being substances, one would suppose that he was here asking us merely to consider what it would be like for someone to have nothing on his mind but his hunger or thirst. In that case it would be natural to suppose that we are to consider a human being who has not eaten or drunk for a good many hours and whose hunger or thirst distracts him from all other matters. In any case, what we should think of here is a living human being, a creature that can want food and drink and can try to get them. It would then be clear what this ‘self’ or ‘thinking being’ is. But if we were to understand Hume’s passage in this way, we would not have understood his proposed thought-experiment. The experiment is to begin by our considering an impression: the impression known as hunger or the impression known as thirst, and we are then to consider whether we can discover any ‘real connexion’ of that impression with anything else. Now the crucial difference between these ways of reading the passage lies in Hume’s notion that hunger and thirst are substances and so can be considered without considering a creature that can eat or drink. Hunger and thirst are to be thought of as ‘distinct existences’ rather than as states of living organisms. They are to be thought of, in other words, in the grammatical category of ‘object and name’. Now it is just this way of thinking of mental states, including sensations, that entails that idea of a private language that Wittgenstein argues against. On the Humean view, the meanings of words like ‘hunger’, ‘dizziness’, and ‘pain’ must be specified without mention of human beings and what more or less resemble (behave like) them. That is, learning these words will not involve learning to use them in such sentences as these: ‘I thought he was hungry, but he didn’t eat the food I brought him’, ‘If I get dizzy, please catch me’, ‘I can’t walk on this foot; it hurts too much’, ‘I dropped the pan because it was too hot to hold’. On the Humean view, if someone never said such things and did not understand others saying such things, this would not count against his knowing what words like ‘hunger’ and ‘dizziness’ mean. The question, then, is: What would count against his understanding such words? Or count for it? If we answer that he understands ‘dizziness’ if and only if he always uses it as the name of the same sensation, we will have begged the question, for we cannot suppose that ‘same sensation’ could, given Hume’s view, have a use when what is in question is whether, on that view, ‘dizziness’ and ‘pain’ could have a use. And for the same reason it will not do to say that he could remember having called this sensation ‘pain’ before: the expression ‘this sensation’ could not have a use if words like ‘pain’ and ‘dizziness’ could not. But all of this is, in a sense, beside the point, for we are already under a misconception if our account of what sensations are entails the consequence that sensations could only have names with ‘private meanings’. If we reject the idea that sensations are, as Hume thinks, individual substances, i.e., objects of ‘inner sense’, and acknowledge that sensations are states of living organisms, then we can allow that words for sensations are tied up with the natural expression of sensations.
There is, then, an essential connexion of sensations with living organisms. But to admit this is not to accept some form of behaviourism. We can avoid behaviourism while acknowledging the essential connexion if we also acknowledge a certain complexity in the creatures to which sensation words and other psychological concepts apply. We can see this clearly by taking Hume’s example of hunger. It seems perfectly clear that the concept hunger (employed primarily in the adjectival form ‘hungry’) applies only to a being that eats. Yet there is no need to identify hunger with eating or even with trying to get food, for human beings are also the sort of creatures that can diet and go on fasts and so can properly be said to be hungry even when refusing good food. Now surely there is no paradox here. There can be both the essential connexion and the exceptional circumstances, and this is just because human beings are the complex creatures they are. One can turn down food for a great and indefinite variety of reasons, for example, because one cannot pay for the food, because one suspects the food is tainted, because one is trying to lose weight, because accepting the food would deprive someone else more in need of it, because it is food that is forbidden by divine commandment, because remaining to eat would mean missing a train, and so on. In cases such as these one can be hungry without eating the food that is available. Yet such cases do not destroy the essential connexion of hunger with eating, for in such cases one would be prepared to say (or might think to oneself) that one would eat were it not for such and such or that one would have eaten had such and such a reason not occurred to one. Moreover, if someone does not eat the food available to him, we will judge that he is not hungry, unless we suspect that there is some overriding reason he might have for passing up the food on this occasion. The essential connexion, then, allows of certain exceptions and most certainly does not amount to an identification of hunger with eating or with trying to get food. Now the same considerations hold for a sensation word. There are natural expressions of pain, such as drawing back from the cause of pain, crying, favouring the injured part (e.g., limping), seeking relief from the pain, and so on, but also there are many reasons people can have for restraining the natural expression of pain. A child may hold back his tears in order to show that he is brave; a man may try not to limp or wince in order that the person who stepped on his foot will not have to apologize; a cook may not draw back from a painfully hot pan for fear of spilling its contents; someone may stifle a groan so as not to wake another person or may hold back an exclamation so as not to disrupt a meeting, and so on. Also we learn not to be startled or frightened by pain and so gain some control over our reactions. Now the use of the word ‘pain’ is certainly tied up with the natural expression of pain, but this is not to say that there is pain only in case there is pain behaviour. So there is no reason for thinking that to admit an essential connexion here is to invite behaviourism. There is no plausibility of behaviourism here in any case, for the grammar of the word ‘pain’ is completely different from the grammar of ‘the expression of pain’. For example, a pain can throb but there is no way of behaving throbbingly, and whereas a person in pain can wail loudly or whimper softly, a pain can be neither loud nor soft.
What I have tried to make clear here is that Wittgenstein has rejected the Cartesian-Humean metaphysical account of sensations. This needs emphasizing because it is widely supposed that Wittgenstein accepted that metaphysics and merely argued that that metaphysics does not entail the privacy of the language of sensations. This interpretation gets its plausibility from the assumption that the only alternatives here are the Cartesian-Humean account or behaviourism. But it is just this assumption that Wittgenstein rejects: a sensation is not a something but not a nothing either.
VI
I should now like to return to Descartes once again to ferret out one further source of the whole problem. In the following passage in the Second Meditation he gives us what we can now see to be a metaphysical redescription of a human being. He writes:
First came the thought that I had a face, hands, arms,—in fact the whole structureof limbs that is observable also in a corpse, and that I called ‘the body’. Further, that I am nourished, that I move, that I have sensations, that I am conscious: these acts I assigned to the soul…. As regards ‘body’ I had no doubt, and I thought I distinctly understood its nature; if I had tried to describe my conception, I might have given this explanation: ‘By body I mean whatever is capable of being bounded by some shape, and comprehended by some place, and of occupying space in such a way that all other bodies are excluded; moreover of being perceived by touch, sight, hearing, taste or smell; and further, of being moved in various ways not of itself but by some other body that touches it.’ For the power of self-movement, and the further powers of sensation and consciousness, I judged not to belong in any way to the essence of body…; indeed, I marvelled even that there were some bodies in which such faculties were found.25
Now in one respect Descartes is quite right about this last part: a corpse is not the sort of thing of which we can say that it has sensations, sees, is blind, is conscious or unconscious. And if a soul is that of which we can say these things, then of course a living human being is a soul.
The difficulty in Descartes’ remarks lies in the move that looks quite innocent, namely, in ‘I have a body’. And I think we can now see a further source of this move. Descartes’ redescription is a kind of rehearsal of two quite different kinds of language-games. On the one hand, there are those in which human beings are central (complaining of aches and pains, telling dreams, guessing at a man’s motives, etc.), and on the other hand, there are those in which human beings have roughly the same status as sticks and stones (weighing and measuring, etc.). This is a difference that stands out in sharpest relief; it is manifested in hundreds of ways. (Compare: ‘I am as tall as this tree’ and ‘The rock hit a tree, so no one got hurt’.) Now when we come to reflect on this difference it is surely inevitable that we will treat this difference in the use of words, that is, the special status of human beings in the one case and their non-special status in the other, as marking out two different sorts of things composing a human being. This, of course, is the Cartesian account. Behaviourism, then, starting from this account, rejects the language-games in which human beings have a special status. Unlike either of these, Wittgenstein rejects the first step, which escaped unnoticed: the redescription of a human being. We can now express this result as follows: these two kinds of language-games taken together mark off human beings from sticks and stones. If someone now should want to re-open the question, asking: ‘But how can something that lies undressed in bed at night, something of a certain height and weight, have thoughts and sensations?’ we shall have to say, as Wittgenstein suggests (284, 412, 421): Look at someone engaged in a conversation or think of a child just stung by a bee and ask yourself what better subject there could be for thoughts and sensations. In this way we are brought back to earth, turned aside from misleading pictures, and we will find nothing odd in saying that these creatures are thinking or in pain.
Getting these two language-games back together—and in the right way—is not, of course, a simple matter. The problem is rather like that of getting substance and quality to lie down together again: the separation has been so prolonged as now to seem virtually in the nature of things. In each case the difficulty seems to be that we have saddled ourselves with a pair of spurious entities. In the latter case it is the ‘bare particular’ and qualities designed to ‘clothe’ it; in the former case it is the ‘body’ and ‘private objects’. It is only if we let go of these that we can find those ‘real connexions’ that Hume was looking for.—Yet other matters are bound to intrude here. One of these Wittgenstein mentions when he remarks that ‘religion teaches that the soul can exist when the body has disintegrated’ (p. 178). Seen in the context of the philosophical problem we have been considering, it is natural to think of this teaching as requiring an interpretation along lines that now seem impossible. It is natural, that is, to think of this teaching as requiring a Cartesian ontology. Yet it would be obtuse to insist on this, for, as Wittgenstein goes on to remark, the teaching has, after all, a point. It is one way of announcing the promise of a life everlasting. And that promise does not itself specify a Cartesian ontology. If we do not at once see how a non-Cartesian account of the matter is possible, then we can only confess ignorance.26 In any case, it would seem presumptuous of a believer to insist that the promise shall be fulfilled in the way that he has been accustomed to thinking of it. At the same time, it would be equally presumptuous of a non-believer to boggle at this talk of soul and body. After all, we all still speak of the sun rising and setting, and no one is the worse off for that. Indeed, it seems unlikely that we shall ever speak otherwise.
1 At section 316 the discussion does not end but is given a new turn; the investigation of the concept thinking and others in sections 316–76 should be seen as containing a further account of the way in which Wittgenstein means to oppose the idea of an inward or private identification of a mental state or process. He makes this connexion explicit in the next group of passages, 377–97, and then in section 398 the discussion returns to the question raised in 281–7 about the nature of the subject of pain or thought. Here he first discusses (398–413) puzzles about the first person pronoun and the idea that the ‘self’ is discerned by an inward gaze, and he then concludes the discussion of the whole topic by taking up questions about human beings, souls, and automata (414–27). He comes back to the topic in Part II, p. 178.
2 See Philosophical Investigations, p. 178.
3 Descartes: Philosophical Writings (Edinburgh, 1954), eds. G. E. M. Anscombe and P. T. Geach.
4 Frank Ebersole once remarked in another context that philosophers often talk of people as if they were speaking of zombies, which the dictionary describes as corpses that, by sorcery, are made to move and act as if alive. At the time I did not fully appreciate the significance of this remark, but very likely it did something to help focus my thoughts for the present essay. (See also the excellent chapter on human actions in Ebersole’s book, Things We Know (Eugene, Oregon, 1967), pp. 282–304.)
5 See Descartes’ Principles, I, xlvi, lxviii, where he maintains that what is clearly and distinctly perceived as a sensation is something that ‘takes place within ourselves’ and involves nothing of a corporeal nature.
6 This idea is seldom made as explicit as it was by C. J. Ducasse, who wrote: ‘What thought, desire, sensation, and other mental states are like, each of us can observe directly by introspection; and what introspection reveals is that they do not in the least resemble muscular contraction, or glandular secretion, or any other known bodily events. No tampering with language can alter the observable fact that thinking is one thing and muttering quite another; that the feeling called anger has no resemblance to the bodily behaviour which usually goes with it; or that an act of will is not in the least like anything we find when we open the skull and examine the brain. Certain mental events are doubtless connected in some way with bodily events, but they are not those bodily events themselves.’ Is Life After Death Possible? (Berkeley, 1948), p. 7.
7 Locke was bold enough to draw this conclusion. Having said. (Essay Π, I, 4) that we get the ideas of the operations of our own minds from ‘internal sense’, he can later see ‘no contradiction’ in the supposition that God might give to some ‘systems of matter’ the powers to think, feel, and enjoy. (Essay IV, III, 6.)
8 ‘Operationalism and Ordinary Language: A Critique of Wittgenstein’, American Philosophical Quarterly, Vol. II, pp. 281–95.
9 Ibid., p. 281.
10 Ibid., p. 282.
11 Ibid., p. 283. Chihara and Fodor cannot, I think, be putting this in just the way they want, for it would seem to be merely redundant to say that there is a logical connexion between pain-behaviour and pain. What they mean, I suppose, is that Wittgenstein thought that there is a logical connexion between such and such ‘bodily movements’ and pain.
12 Ibid., p. 282.
13 Principles of Behavior (New York, 1943), p. 25.
14 Ibid., p. 292.
15 The foregoing argument is not made explicit in The Blue Book, although the conclusion is stated in several different ways on page 25. That the argument is not made explicit here is perhaps to be explained by the fact that these notes were ‘meant only for the people who heard the lectures’ (p. v), where Wittgenstein may have made explicit use of the argument. In Philosophical Investigations he makes use of the argument in section 153, which concludes: ‘And if I say it is hidden—then how do I know what I have to look for? I am in a muddle.’
16 J. S. Mill, Examination of Sir William Hamilton’s Philosophy, London, 1872, p. 244
17 Op. cit., p. 293.
18 Ibid., p. 294.
19 Ibid.
20 See, for example, Alan Donagan’s interpretation in ‘Wittgenstein on Sensations’, in Wittgenstein: The Philosophical Investigations, ed. George Pitcher, p. 347.
21 A Treatise of Human Nature (Oxford, 1951) ed. L. A. Selby-Bigge, p. 233.
22 Ibid., p. 207.
23 Ibid., p. 233.
24 Ibid., pp. 634–5.
25 Descartes, op. cit., pp. 67–8.
26 Wittgenstein says that he ‘can imagine plenty of things in connexion with’ the teaching, and here we should bear in mind that the promise has been filled out with an account of ‘resurrection bodies’. On this point I have been greatly benefited by conversations with my colleague Robert Herbert. See his essay ‘Puzzle Cases and Earthquakes’, Analysis, January, 1968.