DAVID PEARS
WITTGENSTEIN'S CRITICISM OF CARTESIANISM
1. INTRODUCTION The history of ideas is often simplified and dramatized, so that it becomes more like folklore than real history. The mythologizing of philosophy is most conspicuous at its crises and turning-points, and the treatment of Wittgenstein's private language argument is a very clear example of it. When I say that, I am not referring to the recent controversy about the identification of the argument: Is it the argument against the possibility of a sensation-language completely detached from everything in the physical world? Or is Kripke right when he identifies it with Wittgenstein's denunciation of certain general theories about rule-following? I am going to avoid that controversy, because my concern is solely with the argument against the possibility of a completely detached sensation-language, and I do not really care what this argument is called. I want to look at the argument because it seems to me that commentators have simplified it and made it into something much less interesting than it really is. It is understandable why this argument has been mythologized. The texture of Wittgenstein's writing makes it difficult to cut neat packages out of it. When he offers a new way of looking at a problem it is not his practice to try to persuade us with a single self-contained argument. He has many different ways of inducing us to get rid of the point of view that he is rejecting and to replace it with something better. He uses realistic sketches of our actual linguistic practices, diagnoses of our misunderstandings of them, illustrations of viable alternatives, and a host of other devices. Of course, there are arguments in his text, but he seldom rests his case on a single one. The argument against the possibility of a completely detached sensation-language - which henceforth I shall simply call "the private language argument" - looks like a welcome exception to his usual strategy of scattering his forces in a kind of guerilla warfare. It is a reductive argument, and it appears to concentrate the whole campaign into a single decisive battle. Synthese 106: 49-55, 1996. (~) 1996 Kluwer Academic Publishers. Printed in the Netherlands.
50
DAVID PEARS
A philosopher working in the Cartesian tradition, like Russell, would maintain that sensation-language comes first in any logical reconstruction of our knowledge of the world around us. Wittgenstein retorts that, on the contrary, it would not be possible to set up sensation-language in complete detachment from the physical world. For in that situation we would have no viable way of distinguishing between two kinds of cases in which we think we are applying a word to a sensation correctly, the cases where we really are, and the cases where we are not applying it correctly. This is a reductive argument and it looks attractively simple: Russell's detached sensation-language would collapse for lack of a viable distinction between correct and incorrect uses of a word. So we cut this argument out of the complex pattern of Wittgenstein's thought, and take it to mark a turningpoint, or, at least, a crisis in the development of western philosophy. I am not saying that this interpretation of Philosophical Investigations is mistaken - that would be absurd - but I do want to show that it is less than the whole truth. So the rest of this paper will be concerned with what has to be added to these bare bones. There are two big additions to be made. One is an account of the place of the argument in the development of Wittgenstein's philosophy, and its connections with other closely related arguments used by him. It is important to add this material, because the private language argument is not the single decisive battle that it is usually taken to be - not the final breakthrough achieved after a lot of false starts and unsuccessful strategies. The second addendum is an investigation of the premisses of the argument. In any convincing reductio, the premisses must be clearly identified, because, if they are not, the absurdity may not be validly deduced, or, if it is, the finger of blame for it may not point clearly at the culprit.
2.
THE PLACE OF THE PRIVATE LANGUAGE ARGUMENT
The private language argument does not stand alone in the development of Wittgenstein's philosophy. It makes its first appearance in his writings in his notes for a course of lectures given in 1936.1 But it is preceded by three other, closely related arguments, and it is offered as the last member of a quartet. So the best way to understand it is to take it as it is offered, together with its three companion arguments. The first thing that needs to be explained is the division of Wittgenstein's campaign against Cartesianism into four parts. A solipsist is a philosopher who withdraws, or, at least, tries to withdraw into a private world containing only a mental subject and its mental objects. So there will be two parts to any critique of this privatization of the world: it will have to deal with the
WITTGENSTEIN'S CRITICISM OF CARTESIANISM
51
solipsist's account of the subject of experience, and it will have to deal with his account of the objects. Wittgenstein's campaign on the first of these two fronts starts in the Notebooks and Tractatus, and his campaign on the second front starts in the lectures that he gave in Cambridge in 1930. 2 In both cases the issue concerns identity - what are the criteria of identity of subject and objects in the solipsist's private world? Naturally, Wittgenstein is not only attacking the dogmatic solipsist: he is also attacking the classical phenomenalist who dips into solipsism but thinks he can climb out on the other side of it. So far, the field has been divided into only two parts, and another line must be drawn across it to produce a quadripartite division. That is achieved when we ask the following question: "Is Wittgenstein arguing that the solipsist's attempt to carve a private world out of the common public world is incoherent? Or is he arguing that, even if it could be done, there are certain essential linguistic devices that he would be unable to take with him when he retreated into his private world?" The answer to this question is that he argues in both these ways, but, of course, not simultaneously, because the two arguments are quite different from one another. It is one thing to argue that a certain project is incoherent, and quite another thing to argue that even if it were coherent, it would involve a crippling impoverishment of language. We now have a second dichotomy and, therefore, four arguments, of which the last one, in the bottom fight hand corner, is the private language argument. But that is a very brief and abstract way of describing the situation, and so I will now go through the arguments, one by one, placing each in its appropriate square. The best place to start is with the private language argument itself, because it is the most familiar of the four. It is concerned with the objects of experience, and it tries to show that nobody in the solipsist's position could set up a language for describing them. The argument is that a particular linguistic device, reidentification of sensation-types, would not be available to him in his private world. This argument is meant to cap another, more general argument derived from the Tractatus theory of the limit of language: the conclusion of that argument was that the sopipsist's method of specifying the only objects that he experiences is incoherent, while the conclusion of this argument is that, even if it was not incoherent, the solipsist would have lost the ability to reidentify sensation-types. Let us take a step backwards in time and look at the argument for the incoherence of a solipsism based on private objects. According to the solipsist, the limit of the world as he experiences it is the line of his own sense-data. But how will he specify the category of his sense-data?
52
DAVID PEARS
Wittgenstein offers him a dilemma at this point: he must specify them either as things belonging to another empirical category, like tables and chairs, or by tying them to statements about what seems to him to be the case. However, if he takes the second option, his apparently restrictive thesis will really be empty; no description of the world could possibly be restricted to statements about seeming, because they presuppose statements about being. If, on the other hand, he takes the first option, his solipsism will be self-refuting. For it is incoherent to start by placing the line of sense-data within the world that he experiences and to end by treating it as the limit of that world. This argument is derived from a principle stated in the preface to the Tractatus. The limit of language is not a line dividing one field from another and so it has to be drawn from inside language. Of course, a classical phenomenalist would deny the incoherence, because he would point out that his last move is to reduce everything to his own sense-data. But that was not how he began. It is sometimes said that Wittgenstein himself was a classical phenomenalist until 1936, when the private language argument showed him what was wrong with that theory. But that is a misinterpretation. He argues against classical phenomenalism in the Blue Book, and, if he himself ever adopted phenomenalism, it was not that version that he adopted, but something more like Kantian phenomenalism. That fixes the two arguments about sensation-types. The two arguments about the subject of experience were exactly parallel: it is incoherent to identify the ego empirically and then to set up a private world based on it, and, even if this were not incoherent, it would involve the loss of another essential linguistic device, discriminating reference to individuals. These two arguments about the subject of experence are the other half of his critique of Cartesian philosophy. They were developed first, and not much is said about them in Philosophical Investigations. That has created the impression that the private language argument carries more weight than Wittgenstein intended it to carry. The argument for the incoherence of the solipsist's treatment of the ego is given in the Notebooks and Tractatus. It is exactly like the later argument for the incoherence of the attempt to specify the phenomenal limit of language empirically. It takes the form of a dilemma: either the solipsist's ego is identified by its connections with the physical world, or else it is not identified at all. In the first case, solipsism is self-refuting, and in the second case it is empty. This emptiness is explained very clearly in the Notebooks: if the ego is not related to any particular body, it could be anybody's - it could even be the collective consciousness of all human beings - and then there would be no personal restriction put on the only
WITTGENSTEIN'S CRITICISM OF CARTESIANISM
53
things to which the solipsist concedes existence, and so his position would merge with idealism. That is the point reached in the development of these ideas in the Tractatus. But the argument leaves a loophole for the solipsist. He can say that he does not have to identify the only things that exist by describing them as "the things that are presented to a particular ego, namely his." He can simply point to them. Wittgenstein takes up this claim in his writings between 1929 and 1935, especially in the Blue Book. He rejects it, on the ground that the solipsist's private world does not contain the resources for discriminating references to individuals. The argument, put briefly, is that such references have to be made from a base of the same general kind as the objects referred to. This condition is met by the human body in the physical world, but it is not reproduced within the mind. That completes the survey of Wittgenstein's critique of Cartesian philosophy. It shows that the private language argument is only one of its four parts, and that commentators who make it carry the whole burden alone are mistaken.
3.
THE PREMISES OF THE PRIVATE LANGUAGE ARGUMENT
I have left little time for the second thing that has to be added to the minimal account of the private language argument - the identification of its premises. So I shall have to be brief, and I shall confine myself to pointing out the importance of one premise that is often overlooked. Near the beginning of his development of the argument in Philosophical Investigations 3 Wittgenstein observes that you can teach a child to substitute the word 'pain' for his natural expressions of pain. This point is important in several ways. For one thing, it shows that pain is a sensationtype that was already discriminated before the advent of language. So Wittgenstein goes on to ask, "How can I go so far as to try to use language to get between pain and its expression?" His idea is that we should not over-intellectualize the acquisition of language. In real life we are not confronted with a lot of sensation-types, each one isolated from everything else and all of them waiting for their phenomenological descriptions. This point has been made by other philosophers from Thomas Reid to Merleau-Ponty and it is developed in an article by Gareth Evans. 4 It can be put like this: solipsists and classical phenomenalists forget that many properties of sensation-types involve theories about the physical world. This is a very important part of the case against this whole school of philosophy, and it appears to be one of the premisses of Wittgenstein's
54
DAVID PEARS
private language argument. Why else should it be placed so prominently at the beginning of his development of the argument? It may seem that there is only a remote connection between the point made by Wittgenstein about the natural expression of pain and the point made by Evans about the implicit references to the physical world contained in certain descriptions of sensation-types. So it is worth dwelling on the connection for a moment, in order to appreciate how close and how substantial it really is. Consider the following three phenomenological properties:painfulness, which is a property of sensations; red, a property of visual impressions; and located in the bottom right hand quadrant of my visual field, another property of visual impressions. Wittgenstein's point is that painfulness is already wired up to its natural expression, avoidance. Elsewhere, he makes the further obvious point that the cause of pain is injury. Putting this in Evans' way, we could say that painfulness involves a theory about its origin. Of course, this theory is only intellectualized later, and in the early stages it remains un-formulated, but, as it were, imprinted on our nervous systems. There are even more striking examples of these pre-linguistic connections in the case of spatial properties. Consider, for example, a man crossing a stream on stepping-stones. He recognizes the positions of the stones, but not just in terms appropriate to his visual field - or, perhaps, it is better to say that the spatial properties of impressions in his visual field are apprehended by him in a way that relates them from the very beginning to the locations of things in the cone of physical space ahead of him. This kind of example is discussed in detail by Evans, and Wittgenstein, as usual, supplies the pieces of the jig-saw without putting them together. The property red is quite different. It has no natural connections with anything biologically important, and so there is no standard reaction to seeing it. No soubt, that is why phenomenalists concentrate on colours and speculate about the possibility that there might be variations between the colours of different people's visual impressions. There are two ways in which these considerations are relevant to the private language argument. First, they undermine the very idea that in the solipsist's private world sense-data could all be reported in the same terms that we now apply to them. That is how Evans uses the fact that so much of our sensation-language carries theories about the physical world. Second, Wittgenstein may have taken these considerations to forearm him against the well-known phenomenalist objection, that a viable distinction between the two cases in which a speaker thinks that he is applying a word correctly actual truth and actual falsehood - can be set up in the solipsist's private -
WITTGENSTEIN'SCRITICISMOF CARTESIANISM
55
world, without any reference to the physical world. Perhaps his answer to this would have been that it would be impossible to set up a descriptive language in this w a y unless some o f its basic f r a m e w o r k had already been imprinted on our nervous systems by nature. I f so, that is a necessary premiss in the private language argument.
NOTES i Wittgenstein Ludwig: 1968, 'Notes for Lectures on "Private Experience" and "SenseData" ', Philosophical Review 77(3), 275-320. 2 Wittgenstein, L.: 1980, in Desmond Lee (ed.), Wittgenstein's Lectures, Cambridge, 19301932, Basil Blackwell, Oxford, UK. 3 Wittgenstein, L.: 1967, Philosophical Investigations, 3rd Edition, translated by G.E.M. Anscombe, Macmillan, New York, I, §§244-5. 4 Evans, Gareth: 1980, 'Things without the Mind', in Zak van Straaten (ed.), Philosophical Subjects: Essays Presented to P.F. Strawson, Clarendon Press, Oxford, UK, pp. 76-116. 5 Evans, G.: 1980, ibid. Oxford University Oxford, England