William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. Wittgenstein’s externalism William Child 1. Introduction A central feature of Wittgenstein’s later writings is a form of externalism about mind and language. The content of a subject’s thoughts and words, Wittgenstein thinks, depends not just on what is explicitly present before her mind – mental images, sensations, experiences, visualized formulae or sentences, and so on – and not just on the bodily movements she makes or is disposed to make, or the words she is disposed to utter, but also on the context in which these things occur. And, like modern externalists, Wittgenstein sometimes supports his externalist claims by Twin-Earth-style thought experiments. He describes imaginary subjects who are just like us in all individualistic respects but are situated in very different contexts and argues, or simply takes it for granted, that their thoughts would be very different from ours. From the many passages where Wittgenstein makes such points, I quote some illustrative examples. First: Let us imagine a god creating a country instantaneously in the middle of the wilderness, which exists for two minutes and is an exact reproduction of a part of England, with everything that is going on there in two minutes. Just like those in England, the people are pursuing a variety of occupations. Children are in school. Some people are doing mathematics. Now let us contemplate the activity of some human beings during these two minutes. One of these people is doing exactly what a mathematician in England is doing, who is just doing a calculation. – Ought we to say that this two-minute-man is calculating? Could we for example not imagine a past and a continuation of these two minutes, which would make us call the process something quite different? (RFM VI-34) Wittgenstein thinks we evidently can imagine a context with respect to which we would deny that the two-minute-man was calculating. What would in one context be a case of calculating would, in a different context, be something quite different. A similar point is made by a second example: It is, of course, imaginable that two people belonging to a tribe unacquainted with games should sit at a chess-board and go through the moves of a game of chess; and even with all the appropriate mental accompaniments. And if we were to see it we should say they were playing chess. But now imagine a game of chess translated according to certain rules into a series of actions which we do not ordinarily associate with a game – say into yells and stamping of feet. And now suppose those two people to yell and stamp instead of playing the form of chess that we are used to; and this in such a way that their procedure is translatable by suitable rules into a game of chess. Should we still be inclined to say they were playing a game? What right would one have to say so? (PI §200) It would obviously be wrong to say that the two people were playing a game of chess if they were merely yelling and stamping their feet in a way that was translatable into a game of chess. 1 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. It would be less obvious that they were not playing a game of chess if they were behaving in a way that gave the external appearance of a game of chess. But, Wittgenstein implies, they would not in fact be playing chess in that case either. Since they are wholly unacquainted with games, nothing they do can count as playing a game of chess, however closely they resemble two people in our society who are playing chess. Their behaviour takes place in the wrong context: it has no connection with any practice of playing games. In a third passage, Wittgenstein argues that possession of particular mental states requires the right social context: Now suppose I sit in my room and hope that NN will come and bring me some money, and suppose one minute of this state could be isolated, cut out of its context; would what happened in it then not be hope? – Think, for example, of the words which you perhaps utter in this space of time. They are no longer part of this language. And in different surroundings the institution of money doesn’t exist either (PI §584). One reason for the context-dependence of mental states is highlighted in the final sentence. The existence of thoughts about money depends on there being such a thing as money. And that, Wittgenstein thinks, requires the actual existence of a complex social institution. 1 But he has a further reason for thinking that the possession of attitudes and emotions is contextdependent: ‘Grief’ describes a pattern which recurs, with different variations, in the weave of our life. If a man’s bodily expression of sorrow and of joy alternated, say with the ticking of a clock, here we should not have the characteristic formation of the pattern of sorrow or of the pattern of joy. ‘For a second he felt violent pain.’ Why does it sound queer to say: ‘For a second he felt deep grief’? Only because it so seldom happens? (PI p. 174). And the continuation of that passage explicitly compares the context-dependence of feeling grief with the context-dependence of playing chess. In each case, something is true of a person at a particular time (he is feeling grief; he is playing a game of chess). But a condition for its being true, Wittgenstein thinks, is the presence of an appropriate context that extends beyond the particular time. It seems clear that he would say the same thing about the case of hope. In this paper I shall discuss three issues about the kind of externalism exemplified in these passages. First, I address a question about the relation between Wittgenstein’s views and a standard modern form of externalism: the model of natural kind terms developed by Putnam and Kripke. 2 Some supporters of Wittgenstein think that Wittgenstein’s views about meaning pose a direct challenge to this form of externalism. I shall argue that the KripkePutnam model, properly understood, is entirely compatible with those views. Second, I consider a problematic aspect of Wittgenstein’s externalism: the apparent suggestion that facts about a subject’s past intentional states may be constituted by things that happen after the event. I argue that Wittgenstein’s views in this area do not threaten our ordinary, realist intuitions about past attitudes. Third, I discuss the objection that Wittgenstein’s arguments do not demonstrate the need for the actual existence of the kinds of contextual conditions he discusses, since those conditions could in principle be internalized in a subject’s intentions and dispositions. I draw on a suggestion of Wittgenstein’s about self-knowledge to respond to this criticism. 2 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. 2. Externalism, meaning and use It is sometimes claimed that Wittgenstein’s views about meaning and mind fundamentally conflict with the externalist view of natural kind terms to be found in the work of Kripke and Putnam. But, I shall argue, that is a mistake. Whatever Wittgenstein’s own view of the meanings of terms such as ‘lemon’, ‘tiger’, ‘gold’ and ‘water’, an account of meaning and content that applies Wittgensteinian principles is entirely consistent with the standard, modern externalist view. The main lines of the externalist view of natural kind terms are familiar. They can be briefly summarized as follows. Some terms are what Putnam calls ‘one-criterion words’: they are associated with a single criterion or description; understanding the term is a matter of knowing the criterion; and it is necessary and sufficient for something’s falling into the extension of the term that it satisfies that criterion. But natural kind terms do not function like that; for there is no set of descriptive properties possession of which is necessary and sufficient for membership of the kind. On the one hand, something may possess all the observable properties normally associated with a kind without belonging to that kind (e.g. fool’s gold); on the other hand, something can lack some of the properties associated with a kind but still be a member (e.g. three-legged or albino tigers). How, then, do natural kind terms function? The canonical way to introduce or explain such a term involves identifying examples of the kind: ‘This is gold (pointing to an example). This is gold (pointing to another example). And this is gold (pointing to a third example). Something else is gold just in case it is the same kind of stuff as these examples. It is the same kind of stuff as the samples just in case it shares the underlying nature that explains their observable properties and causal behaviour. And what that real underlying nature is is a matter to be determined by empirical investigation.’ There are at least three ways in which this Putnamian account has been thought to conflict with Wittgenstein’s views about meaning. 3 First, Putnam’s externalism may seem to conflict with the idea that ‘the meaning of a word is its use in the language’ (PI §43) or that ‘the meaning of a word is what is explained by the explanation of the meaning’ (PI §560). It is said that ‘externalists distort the conceptual connections between meaning, on the one hand, and use, explanation and understanding, on the other’ – that they fail to see that ‘what a word means is determined by convention, not discovery’4. For externalism, it is claimed, ‘implies that the “real” meaning of a word must be discovered by science’ – that ‘scientific discovery can disclose . . . what the words we use, such as “gold” and “water”, “fish” and “lily” really mean’. 5 But, it is objected, that conflicts with the principle that the meaning of a word is a matter of its use in the language. Second, externalism may seem to imply that ordinary speakers are ignorant of the meanings of their words, and thus to undermine the commonsense thought that we are authoritative about our own meanings: ‘Putnam mistakenly thinks that ignorance about the scientific nature of the referents of my words or thoughts entails ignorance about the meaning of these words’, whereas ‘Wittgenstein’s account respects FirstPerson Authority’ about meaning. 6 Third, Putnam’s views may seem to imply that there is a single, correct set of concepts that we must use if we are to describe the world as it really is. But such a view conflicts with Wittgenstein’s insistence that grammar is arbitrary and that it makes no sense to talk of the meaning of word’s being right or wrong, correct or incorrect. 7 I 3 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. shall consider these points in turn. 8 First, the objection that Putnam’s account conflicts with the principle that the meaning of a term is its use in the language. The objection might be put like this. ‘The way we use a term is a matter of the criteria we employ in applying and withholding it. Now for educated modern speakers of English (we may suppose), the criterion for being gold is having atomic number 79. And what makes the property of having atomic number 79 necessary and sufficient for being gold is just that that is the criterion we use for something’s being gold. According to Putnam’s externalism, by contrast, the meanings of words are determined by the actual nature of the things to which speakers apply them. Now the nature of gold, for example, has not changed over time. So, according to Putnam, the extension of the ancient Greek word “chrysos” is the same as the extension of the modern English word “gold”. 9 But to say that, it is argued, conflicts with the link between meaning, use, and criteria. Whether or not the ancient Greek word “chrysos” is applicable to something is a matter of the criteria the Greeks associated with the word. But the Greeks knew nothing of modern chemistry and physical theory, and did not have the concept of atomic number. So their criteria for the applicability of their word “chrysos” were different from our criteria for the applicability of our word “gold”. Accordingly, on a Wittgensteinian view, the words “chrysos” and “gold” have different meanings. And, pace Putnam, a thing may satisfy the criteria for being chrysos (and hence, be chrysos) without satisfying the criteria for being gold (and hence, without being gold), and vice versa.’ The critic is right to insist that words have the meanings they do only because of the way we use them. And, suitably understood, she is right to say that whether or not a word applies to a thing is a matter of whether or not the thing meets the criteria we associate with the word. But Putnam’s externalism is consistent with those principles. Putnam makes three key claims. First, that our term ‘gold’ picks out a kind defined in terms of real, underlying similarity to paradigm exemplars. Second, and crucially, that a necessary condition for the word ‘gold’ to pick out a kind of that sort is that we use it with the intention that it should pick out such a kind. Putnam is explicit on this point: whether a term is synonymous with a description, he thinks, or whether, instead, it refers ‘to whatever things share the nature that things satisfying the description normally possess’ is a matter of ‘the empirical facts concerning the intentions of speakers’. 10 We could have used the term ‘gold’ as a short-hand description or set of descriptions. But in fact we do not; we use it with the intention that the word ‘gold’ should apply to x iff x has the same underlying nature as standard samples. Third, an individual can use a word with that intention even if she herself has no way of telling whether a particular thing does bear the appropriate similarity relation to standard samples; and a linguistic community can use a word with the relevant intention even if the community as a whole has no definitive test. So, Putnam insists, we can know what ‘gold’ means even if we do not (or could not) know, in every case, whether or not the word applies. All of that is consistent with the principle that the meaning of a word is determined by our criteria for applying the word – provided we understand that principle in the right way. The notion of a criterion is ambiguous: it has a constitutive and an epistemic sense. In the constitutive sense, our criterion for a thing’s being gold is our standard of what it takes for something to be gold. In the epistemic sense, our criteria for a thing’s being gold are the tests we actually apply in judging whether or not something is gold. It is a feature of Putnam’s 4 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. externalism that something may meet our constitutive criterion for being gold (i.e. that it may be similar to standard samples in respect of real internal constitution) even if, using our best tests, we cannot tell that it does. But that does not show that externalism violates the principle that the meaning of a term is a matter of the way we use it or explain it. On the contrary, as we have seen, Putnam insists that, for a term to pick out a kind defined in terms of real, underlying similarity, it is not sufficient that there should be some such kind in the vicinity; speakers must use the term with the intention of picking out such a kind. So externalism does not challenge the connections between meaning and use, or between meaning and criteria. What it does challenge is the assumption that our constitutive criterion of what it takes for something to be an F must coincide with the epistemic criteria we employ in telling whether something is an F. That assumption runs through many Wittgenstein-influenced discussions of these matters. 11 But it has no essential connection with Wittgenstein’s principal that the meaning of a word is its use in the language. The second complaint about externalism was that it wrongly implies that ordinary, scientifically-ignorant speakers do not know what their words mean. For the externalist, it is objected, ‘ignorance about the scientific nature of the referents of my words or thoughts entails ignorance about the meaning of these words’. Whereas, it is argued, the truth is that ‘we commonly distinguish between understanding [a term such as “water”] and having expert chemical knowledge of the stuff it refers to’. 12 But the objection is misconceived. As before, it fails to distinguish between our standard of what it is to be a member of a natural kind and our methods of telling whether or not something meets that standard. Putnam’s externalism does not imply that someone who knows no chemistry will be ignorant of the meaning of their word ‘water’. Such a person may know perfectly well what the word ‘water’ means; she knows that it refers to anything that has the same underlying nature as the stuff that she and others standardly identify as water. What she lacks is a general way of telling whether liquids that seem like water really are water: whether they do have the same kind of internal constitution as the ordinary samples. But that is not ignorance of the meaning of the word; it is ignorance of whether or not the word applies in this or that particular case. The third objection was that externalism about natural kind terms implies that there is a single, correct set of concepts – the set of concepts that classify things in terms of their scientifically-discoverable real essences – so that concepts that classify things in different ways will be wrong, and descriptions that employ such concepts will fail to describe the world as it really is. So, it is said, externalism conflicts with a series of familiar Wittgensteinian claims: that grammar is arbitrary; that concepts themselves cannot be right or wrong; that people who had concepts very different from ours would not be classifying things wrongly but simply classifying them differently; and so on. But, once more, the objection overlooks the role of speakers’ intentions in Putnam’s externalism. As I have said, it is an explicit part of Putnam’s position that, say, in order for the word ‘water’ to be a rigid designator of H20, it is necessary that ‘speakers intend that the term “water” shall refer to just those things that have the same lawful behaviour and the same ultimate composition as various standard samples of actual water’. 13 It is an empirical question whether an individual or community does intend to use a word with that intention. And Putnam thinks that sameness of microstructure often is the dimension of similarity to which we intend to attach our words. But nothing stops us picking out kinds of other sorts. Suppose that we did use words like ‘water’ not as Putnamian natural 5 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. kind terms but, instead, with purely descriptive meanings – so that the word ‘water’ applied to anything that has most of the qualities standardly associated with water. It is no part of Putnam’s externalism that we would have given our words the wrong meanings, or incorrect meaning. The externalist agrees with Wittgenstein that meanings or concepts are not themselves right or wrong: that it is what people say when they use those meanings or concepts that is right or wrong, correct or incorrect, true or false. I conclude that Putnam’s externalism is entirely compatible with Wittgenstein’s principles about the link between meaning and use, about a speaker’s knowledge of the meanings of her own words, and about the arbitrariness of grammar. 3. The retrospective constitution of intentional connections? One principle that emerges from the discussion in section 2 might be put like this: The mere fact that someone inhabits an environment that contains natural kinds – kinds whose members are united by their underlying nature – does not itself suffice for her words and thoughts to pick out such kinds. For that, there must also be something about the subject herself that brings the external context into the constitution of her meanings and concepts. In particular, the subject must have the right linguistic intentions. Now a parallel issue arises in connection with an interesting and little-discussed element of Wittgenstein’s externalism. As we saw in section 1, a central theme in Wittgenstein’s writings is the idea that the content of a subject’s thought is not determined by what was going on in his mind at the time – which may be exactly the same in cases where the person is thinking about quite different things. The object of someone’s thoughts or words, Wittgenstein thinks, depends on the context in which they occur. So, for example, what makes it the case that someone is thinking about one thing rather than another may be a matter, not of anything going on at the time, but of what came before the episode of thinking, or even of what came after. It is this last claim that I want to focus on. Wittgenstein’s writings contain various instances or applications of the claim. For example: If the meaning-connexion can be set up before the order, then it can also be set up afterwards (Z: 289). Again: If I have two friends with the same name and am writing one of them a letter, what does the fact that I am not writing it to the other consist in? In the content? But that might fit either. (I haven’t yet written the address.) Well, the connection might be in the antecedents. But in that case it may also be in what follows the writing (Z: 7). And a third case: ‘Suddenly I had to think of him.’ Say a picture of him suddenly floated before me. Did I know that it was a picture of him, N.? I did not tell myself it was. What did its being of him consist in, then? Perhaps what I later said or did (Z: 14). What exactly is Wittgenstein proposing when he says that the difference between writing to one person rather than another may ‘consist in’ what happened after the writing; or that it may be in virtue of what I went on to do that my image counts as having been an image of N? One way of reading such remarks is to see them as proposing the thesis that intentional 6 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. relations can be retrospectively constituted. On this view, it can be true that, at time t, S intended to write to N, even though there was nothing at all about S at that time that made it true that she intended to write to N rather than M; what makes it the case that she had that intention at t is something that she said or did later on. Now it is entirely plausible that it may take something that happened afterwards to show what someone meant or intended. But the claim that intentional relations can be constituted retrospectively is a metaphysical thesis – a form of anti-realism about the past. And such a thesis is very hard to accept. How could it be true that something that happened after t made it the case that, at some earlier time, the subject had the intention she did? After all, suppose she had dropped dead after t, before going on to do or say anything else. If it is true that she had that intention at t, surely that could not be affected by whether or not she lived to tell the tale? That worry about Wittgenstein’s suggestion has a certain similarity to the Wittgenstein-inspired objections to Putnam’s externalism that we considered earlier, in section 2. The earlier objection might be put like this: ‘How can the mere fact that someone’s words are surrounded by a given context make it the case that their meaning is partially constituted by that context? Surely there must be something about the subject herself, in virtue of which the context has some content-determining role.’ We can present the current worry in similar terms. ‘How can the mere fact that an utterance or an action happens in one context rather than another – a context in which the subject goes on to do one lot of things rather than another – make it the case that the original utterance or action has one content rather than another? Unless there is something about the subject at the time, which already links her utterance or action to a particular object, shouldn’t we rather say that her utterance or action did not, at the time, have that object at all?’ The basic concern is essentially the same in each case: intentional relations cannot be constituted by contextual relations between a subject and anything else unless there is something about the subject that brings the context into the constitution of her intentional states or properties. How should Wittgenstein respond to this worry about his suggestion that what gives an utterance or action its intentional content may be something that came after? Wittgenstein’s discussion is complex, and there are various different kinds of case to consider. In some cases, the idea that a subsequent judgement gives content to an earlier item is entirely consistent with common sense. One such case would be this. I draw a face – without intending it to represent anyone in particular. You ask who the picture represents. At that point it strikes me for the first time that the picture looks like my friend N; so I decide to use the picture as a picture of N, and I reply ‘It’s a picture of N’. In such a case, the later judgement ‘It’s a picture of N’ really does establish the connection between the picture and N. It confers a specific singular content on something that did not previously have one. But the judgement has no retrospective effect on what was true at the earlier time; it does not bring it about that, at the time when I drew the picture, it was already a picture of N. However, we normally think that cases of that kind are the exception rather than the rule. To hold that it is quite generally true that words and actions have no specific content at the time they are uttered or performed, and that their content is conferred on them by later judgements, would be deeply counterintuitive; it would be a radical revision of our ordinary view. Fortunately, there is no reason to ascribe such a revisionary position to Wittgenstein. In the passages where Wittgenstein suggests that what gives something its content may 7 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. be something that preceded it or came after, what he is principally arguing against is the idea that a thought is linked to its object by something going on in the subject’s mind at the time. A major theme in Wittgenstein’s later philosophy is that no conscious mental item – no image, or visualized sentence, or formula, or anything else – is in itself sufficient to tie a thought to its object. Any ordinary image or description, he argues, is susceptible of indefinitely many different applications or interpretations. And it is no good appealing to a ‘queer’, or ‘occult’ mental item: something that is like an ordinary image in being present in an instant, but which is interpretable in only one way. For there is no such thing. What, then, does make my thought about N a thought about N? Very crudely, my ability to identify N as the person I am thinking about. As Wittgenstein puts it: What makes this utterance [of the expression ‘I see him now vividly before me’] into an utterance about him? Nothing in it or simultaneous with it (‘behind it’). If you want to know whom he meant, ask him (PI p. 177). But this picture is quite compatible with our ordinary, realist view of past intentional connections. On Wittgenstein’s view, what makes it true that I meant him is not something that was consciously going on in my mind at the time. But that is not to say that there was nothing at all about me at the time in virtue of which it is true that I meant him. There was something about me at the time that made it true that I meant him. But it was not something that was going on. For Wittgenstein, what makes it true that I meant what I did is the set of abilities and dispositions I had at the time. Thus, for example, the fact that I was writing to one of two friends with the same name rather than the other is underpinned by such facts as that I could at the time have recognized or specified him as the intended recipient. So it is not true that the only difference lay in what I went on to do subsequently; that was merely the overt manifestation of dispositions or abilities that I already had at the time. The main thrust of Wittgenstein’s position, then, contains no suggestion that intentional connections are retrospectively constituted by a subject’s subsequent judgements. 14 4. Externalism and Self-Knowledge Consider the example from PI §200 quoted earlier. If ‘two people belonging to a tribe unacquainted with games’ move pieces about on a chessboard in a way that mirrors a possible game of chess, and ‘with all the appropriate mental accompaniments’, what they do will not count as playing a game of chess. For that, their actions need to occur in a context in which games are regularly played, where there is an established practice of chess-playing, and so on. But why should such a context be needed? What does the addition of the context supply? Wittgenstein is clear that, though a context in which there is such a thing as chess is a necessary condition for something to count as a move in a game of chess, it is not a sufficient condition. It is also necessary that the person who makes the move should be appropriately related to that context. If her action is to count as making a move in chess, she must know about the game of chess, and intend her action as a move in a game of chess. In this respect Wittgenstein’s externalism parallels Putnam’s externalism about natural kinds. In each case, the external context is important. But it does not do all the work. Speakers’ intentions play a crucial role as well. But, it might be objected, if the subject’s intentions are so important, why is the 8 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. external context needed at all? Couldn’t everything that the context is supposed to supply be built into the subject’s intentions, so that we could do away with any external or contextual condition altogether? Wittgenstein anticipates just that objection. He puts it like this: ‘But it is just the queer thing about intention, about the mental process, that the existence of a custom, of a technique, is not necessary to it. That, for example, it is imaginable that two people should play chess in a world in which otherwise no games existed; and even that they should begin a game of chess – and then be interrupted’ (PI §205). But Wittgenstein is clear that the external conditions for intending to play chess cannot be internalized in that way; the intention to play chess really does require the actual existence of a practice of playing chess. His immediate response to his own statement of the objection is this: But isn’t chess defined by its rules? And how are these rules present in the mind of the person who is intending to play chess? (PI §205) And the point is amplified in a later passage: An intention is embedded in its situation, in human customs and institutions. If the technique of the game of chess did not exist, I could not intend to play a game of chess (PI §337). Wittgenstein’s point is his familiar one. Nothing that is ‘present in the mind’ of a person suffices for her to intend to play chess. For the kind of thing that is present in the mind of a person – a linguistic formulation of the rules of chess, for example – can be applied or interpreted in indefinitely many different ways. To specify the rules of chess, what is required is not just a linguistic formulation of the rules, but a way of taking or applying the formulation. And a way of taking the formulation cannot itself be built into what is present in someone’s mind; it requires the existence of an actual practice of applying the rules. At this point, however, a new challenge arises. Suppose we grant that the rules of chess cannot be present in someone’s mind in virtue of the presence of anything ‘stationary . . . such as a picture that one has in one’s mind’ or a linguistic formulation of the rules. And suppose we grant that a specification of the rules of chess requires something ‘dynamic’: a way of using or applying a formulation of the rules. 15 Still – why couldn’t someone’s way of taking a formulation of the rules of chess be built into her capacities and dispositions without ever actually being exercised? Why can’t it be embodied in facts about the way she would apply the rules if she were to apply them; about the moves she would make if she were to play a game of chess; about the ways she would explain and justify those moves; and so on? And why can’t facts such as these supply what the internalist needs: a determinate way of taking or applying her formulation of the rules of chess, in virtue of which she can count as intending to play a game of chess even in the absence of any actual practice of playing chess?16 Indeed, consider the following remark: ‘What makes our conversation a conversation about him?’ Certain transitions we made or would make (LW i 308). The reference there to transitions that ‘we would make’ introduces an implicit counterfactual: ‘Had we considered such-and-such, we would have gone on to say so-and-so’. And Wittgenstein suggests that the truth of the claim that we were talking about a particular person may depend on the truth of counterfactuals of that kind. So Wittgenstein himself is committed 9 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. to the idea that intentional properties may depend on counterfactuals of that form. Why shouldn’t we extend the same idea in the way the internalist proposes – giving up Wittgenstein’s idea that intending to play chess requires a context in which there is an actual practice of playing chess, and claiming, instead, that it requires only a suitable pattern of counterfactual truths about the subject? One response to this internalist proposal would be to reject the realism about dispositions and counterfactuals that it presupposes. The internalist is assuming that there is a fact of the matter about what a subject would do in circumstances that have not arisen and will never arise: even though neither the subject nor anyone else has ever played chess, or any other game, there are determinate facts about how she would explain and apply the rules, about the ways she would move the pieces if she came to play a game of chess, about the kind of interest she would take in playing and winning games, and so forth. But, according to this first response, that is an illusion. There simply are no facts of the matter about what a subject would do in such circumstances. That is why it is only a context of actual applications that can supply the necessary way of taking or applying a formulation of the rules. Wittgenstein is sometimes interpreted as taking just such an anti-realist view of counterfactual conditionals. That is how Michael Dummett interprets him. He writes: Wittgenstein says, of a calculation that we have not made and will never make, that it is wrong to say that God knows what its result would have been had we made it, for ‘there is nothing for God to know’. 17 And, as Dummett reads Wittgenstein, the reason there is nothing for God to know is that there simply is no fact of the matter about what would have happened in circumstances that never arose. I agree that there is room for debate about the extent to which Wittgenstein accepts a simple realism about unmanifested dispositions (though I am not convinced by Dummett’s reading of the particular passage he mentions). 18 But I want here to consider a different response to the internalist proposal: a response that is suggested by a passage in Remarks on the Foundations of Mathematics. The RFM passage starts with a statement of the question we have been pondering: What surrounding is needed for someone to be able to invent, say, chess? (RFM VI32) Wittgenstein continues: Of course I might invent a board-game today, which would never actually be played. I should simply describe it. But that is possible only because there already exist similar games, that is because such games are played (ibid.). And he immediately reinforces the point: I may give a new rule today, which has never been applied, and yet is understood. But would that be possible, if no rule had ever actually been applied? And if it is now said: ‘Isn’t it enough for there to be an imaginary application?’ the answer is: No. (Possibility of a private language.) (ibid.) The discussion proceeds: ‘But how often must a rule have actually been applied, in order for one to have the right to speak of a rule?’ How often must a human being have added, multiplied, divided, before we can say that he has mastered the technique of these kinds of calculation? And by that I don’t mean: how often must he have calculated right in 10 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. order to convince others that he can calculate? No, I mean: in order to prove it to himself (ibid.). Wittgenstein’s answer to his own question is that, unless a person has actually calculated right in some sufficiency of cases, he cannot know that he can calculate: But couldn’t we imagine that someone without any training should see a sum that was set to do, and straightway find himself in the mental state that in the normal course of things is only produced by training and practice? So that he knew he could calculate although he had never calculated. (One might, then, it seems, say; The training would merely be history, and merely as a matter of empirical fact would it be necessary for the production of knowledge.) – But suppose now he is in that state of certainty and he calculates wrong? What is he supposed to say [to] himself? And suppose he then multiplied sometimes right, sometimes again quite wrong. – The training may of course be overlooked as mere history, if he now always calculates right. But that he can calculate he shows, to himself as well as to others only by this, that he calculates correctly. What, in a complicated surrounding, we call ‘following a rule’ we should certainly not call that if it stood in isolation (RFM VI-33). We might see the dialectic like this: Wittgenstein suggests that it is only if a person has actually calculated right in some sufficiency of cases that he can know that he can calculate. The internalist denies that: a person who has never calculated, he thinks, can know that he can calculate, provided that he is ‘in the mental state that in the normal course of things is produced by training and practice’. But Wittgenstein rejects that internalizing move: having actually calculated really is a necessary condition for knowing that one can calculate. And if Wittgenstein is right about that, then the same point will generalize: having actually played games is a necessary condition for knowing one can play games; having actually followed rules is a necessary condition for knowing one can follow rules; and so on. Does this line of thought from RFM VI-33 suggest a cogent argument for Wittgenstein’s externalism – for his view that, for example, the intention to play chess requires a context in which there is an actual practice of playing chess, or that it’s being true that someone has the ability to calculate requires a context in which there is an actual practice of calculating? A critic might argue that Wittgenstein’s line of thought does not support externalism at all. In the quoted passage, Wittgenstein is discussing the conditions for knowing that one can calculate. But, the critic says, the debate between the externalist and the internalist is not a debate about the conditions for knowing that one can calculate. It is a debate about the conditions for the existence of the ability to calculate. And even if we agree with Wittgenstein that knowledge that someone can calculate requires a context in which that person actually does calculate (even where that person is oneself), that does not show that a context of actual calculations is required in order for a person to be able to calculate. For all the argument shows, someone could have the ability to calculate even if they had never calculated – and even if no-one had ever calculated. How should the externalist respond to that criticism? One response would be to claim that, for Wittgenstein, the meaning of the proposition ‘S can calculate’ is to be explained in terms of the conditions under which we would be justified in asserting the proposition. On this view of meaning, if Wittgenstein is right that we can only be justified in asserting that a 11 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. person can calculate if she has actually calculated correctly in some sufficiency of cases, it follows directly that we can only make sense of the supposition that she can calculate if she has actually calculated correctly. So, contrary to the internalist’s objection, considerations about the conditions for knowing that someone can calculate are directly relevant to the dispute between internalism and externalism. Many readers would dismiss this response – on the grounds that, though Wittgenstein clearly embraced some form of verificationism in 192930, he had abandoned that view by 1943/44, when he wrote the remarks in RFM VI. I think the status of verificationism in Wittgenstein’s later philosophy is less clear-cut. For instance, it is a notable feature of his discussion of rule-following that he focuses at many points on questions of the form ‘Can we say that someone is following a rule?’ or ‘Are we justified in saying it?’, rather than on questions of the form ‘Is it true that someone is following a rule?’19 And it is not at all clear that, in the kind of case where he says ‘There is not enough regularity for us to call this “language”’ he would be happy to go on: ‘Nonetheless, there is a fact of the matter about whether it is language; and it might really be language even though there is not enough regularity for us to call it that.’ So I do not regard this first suggestion as a nonstarter. But I want here to consider a different argument, which we can summarize as follows: ‘Having the ability to calculate involves knowing that one can calculate. So anything that is a necessary condition for knowing that one can calculate will also be a necessary condition for being able to calculate. In particular, since actually having calculated is a necessary condition for knowing that one can calculate, it is also a necessary condition for being able to calculate.’ Stated as briefly as that, the argument is unlikely to seem convincing. Two points in particular call for expansion and justification: Why does having the ability to calculate involve or require knowing that one can calculate? And why is actually having calculated a necessary condition for knowing that one can calculate? I shall comment on those points in turn. I do not pretend that my comments will amount to anything like a vindication of the argument. But I do think they point to a position or perspective that is suggested by some of Wittgenstein’s writings. Wittgenstein insists on a distinction between following a rule and merely conforming to a rule. Consider someone who responds to our signposts by walking in the direction that they point – but who knows nothing of the practice of erecting and using signposts. She does not recognize our signposts as signposts; nor does she think of them as pointing in any direction. Such a person, Wittgenstein thinks, conforms to the rules of signposts. But she does not follow the rules. Following a rule is not the exercise of a merely practical disposition. It requires, in addition, that one knows what the rule requires; that one intends to follow it; and so on. Now Dummett develops a related point in arguing that mastery of a language is not a mere practical ability. ‘To regard the understanding of a word or an expression purely as a practical capacity’, he writes, ‘is to render mysterious our capacity to know whether we understand’. 20 One can possess a mere practical ability, he observes, without knowing that one possesses it. Swimming, for example, is a merely practical ability. And one might have the ability to swim without knowing that one did. Accordingly, if someone asks you whether you can swim, it makes perfectly good sense to reply that you do not know, but will try to swim and see whether you can. But linguistic knowledge, Dummett argues, has a different 12 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. character. One cannot understand a word without knowing what one understands it to mean. So one cannot understand a word without knowing that one understands it. 21 And a similar point holds at the level of a whole language. One cannot understand a language without attaching sense to the words of the language. And one cannot attach sense to the words of the language without knowing what sense one attaches to them. So one cannot understand a language without knowing that one understands it. Accordingly, if someone asks me ‘Do you speak Spanish?’, Dummett insists, it would make no sense to respond ‘I don’t know: I’ve never tried’. Whereas that response would make perfectly good sense if speaking Spanish were, like swimming, a merely practical capacity. Similar points apply to the case of calculating. The ability to calculate is not a merely practical ability: an ability to respond to calculation problems by coming out with what are in fact correct answers. One could, indeed, have that merely practical ability without knowing that one had it. But in order to calculate, one must not merely conform to the rules of calculating; one must follow the rules. And one cannot follow the rules of calculating without knowing what rules one is following – without knowing that one is following the rules of calculating. And if one knows that one is following the rules of calculating, then one knows that one can calculate. That line of thought evidently needs to be filled out, and defended against various objections. But suppose we agree that the ability to calculate does involve knowing that one can calculate. Why should knowing that one can calculate require that one actually has calculated correctly in some sufficiency of cases? In the passage quoted above from RFM VI33, Wittgenstein does not spell out his argument explicitly. But a plausible interpretation of his view is this. Suppose that someone who has never calculated suddenly finds herself ‘in the mental state that in the normal course of things is only produced by training and practice’. It is not clear exactly what Wittgenstein is supposing the person to have suddenly acquired. But suppose she acquires two things: a sense of certainty that she will be able to respond to calculation problems correctly; and the ability actually to do so. Wittgenstein suggests that, if she has never exercised her disposition to answer calculation problems, then she does not know that she can calculate. And his reason, it seems plausible to think, is that the subject is not entitled to claim that she can calculate. Suppose she asks herself Wittgenstein’s question: Couldn’t I be in this state of certainty but go on to calculate wrong? Nothing entitles her to think that she really can calculate – that, when she tries to calculate, she will in fact succeed. For, intuitively, she might very easily be wrong; she might easily be in just the same epistemic situation position as she actually is but not be able to calculate at all. By contrast, if she has already calculated correctly, and if her calculations have been accepted as correct by others, then she could not easily be wrong when she believes that she can calculate. The fact that she has in fact calculated successfully entitles her to think that she can calculate. That is one reason why it is plausible to think that knowledge that one can calculate requires an actual history of correct calculation. But there is also a second reason. (It is less clear that Wittgenstein has this second reason in mind in the passage we are considering. But it does involve a fundamentally Wittgensteinian thought.) In order to know that I can calculate, I need to know what calculating is. To know what calculating is, I must know what the rules of calculating require. But knowing what the rules of calculating require is itself an exercise of the ability to follow 13 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. those rules; knowing that they require such-and-such and so-and-so is itself an instance of calculating. So a further reason why you cannot know that you can calculate if you have never calculated is that you cannot know what calculating is except by calculating. Wittgenstein, then, seems right to say that one cannot know that one can calculate unless one has actually calculated correctly in some sufficiency of cases. I have suggested, also, that there is at least an arguable case for saying that being able to calculate involves knowing that one can calculate. Taken together, those two points suggest an interesting and plausible response to the internalist who responds to Wittgenstein’s externalism by arguing that the contextual conditions for calculating, for playing a game of chess, and so on, can be built into people’s intentions, and that those intentions can in turn be built into people’s unmanifested dispositions. What I have presented is no more than a sketch of this line of opposition to the internalist: each of the two key points needs further development and defence. But the line of opposition is, I think, a promising one. 22 University College Oxford OX1 4BH UK 14 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. NOTES 1 See section 4 below for discussion of the natural objection that thoughts about money may require a context of certain other thoughts and intentions, but that they do not require the actual existence of the social institution of money. 2 See Hilary Putnam ‘The meaning of “meaning”’, reprinted in his Mind, Language and Reality (Cambridge: Cambridge University Press, 1975); Saul Kripke, Naming and Necessity (Oxford: Blackwell, 1980). I focus here on the version set out in Putnam. 3 Each point is clearly illustrated in a paper by Hans-Johan Glock and John Preston, ‘Externalism and First-Person Authority’, The Monist, 78, 1995, 515-33, which argues against Putnam’s and Burge’s forms of externalism on supposedly Wittgensteinian grounds. 4 The first quotation is from Glock and Preston: 518. The second comes from Peter Hacker ‘Substance: Things and Stuffs’, Proceedings of the Aristotelian Society Supplementary Volume, 78, 2004, 41-63: 54. 5 For the first claim, see Glock & Preston: 516. For the second, see Hacker: 54. 6 Glock & Preston: 518, 527. 7 For the arbitrariness of grammar, see e.g. Z §§320, 331. For opposition to the idea that a meaning or concept could be right or wrong, see e.g. LFM 183 and PI p. 230. 8 It is interesting to note that Donald Davidson criticizes Putnam’s externalism for reasons that quite closely parallel the supposedly Wittgensteinian objections outlined here. I consider his criticisms in my ‘Triangulation: Davidson, Realism and Natural Kinds’, Dialectica, 55, 2001, 267-287. 9 For this claim, see ‘The meaning of “meaning”’: 235. 10 ‘The meaning of “meaning”’: 238. 11 See for instance Hacker in ‘Substance: Things and Stuffs’: ‘Insofar as paradigms play any role in explaining the meaning of a word . . . they must be usable as objects of comparison to guide the application of the word. For the role of the paradigm is as a standard for the correct application of [the word] . . . So the relevant features of the paradigm must be both known and evident – otherwise the paradigm has no normative role’ (55). 12 Glock & Preston: 518. 15 William Child, “Wittgenstein’s Externalism”, in Daniel Whiting, ed., The Later Wittgenstein on Language, Houndmills, Basingstoke: Palgrave Macmillan, 2010, pp. 63-80. 13 Putnam, Reason, Truth and History, Cambridge: Cambridge University Press, 1981: 46-7. See also ‘The meaning of “meaning”: 231, 238. 14 For further discussion of Wittgenstein’s treatment of past-tense intentional statements, see my papers: ‘Memory, Experience, and Past-Tense Self-Knowledge’, Philosophy and Phenomenological Research, 73, 2006, 54-76; ‘Dreaming, Calculating, Thinking: Wittgenstein and Anti-Realism about the Past’, Philosophical Quarterly, 2007, 252-272; and ‘Remembering Intentions’, in A. Ahmed (ed) A Critical Guide to Wittgenstein’s Philosophical Investigations, Cambridge: Cambridge University Press, forthcoming. 15 For the idea of a stationary meaning and a dynamic meaning (and an immediate caveat about this way of speaking), see LFM p. 184. 16 For an example of such a proposal, see Colin McGinn, Wittgenstein on Meaning, Oxford: Blackwell, 1984, ch. 4. 17 Michael Dummett, ‘Wittgenstein on Necessity: Some Reflections’ in Dummett, The Seas of Language, Oxford: Oxford University Press, 1993, at p. 459. The passage Dummett has in mind occurs at pp. 103-4 in Wittgenstein’s Lectures on the Foundations of Mathematics, Cambridge, 1939, ed. C. Diamond, Chicago: Chicago University Press, 1989. 18 For a discussion of the LFM passage, see my ‘Wittgenstein and Common-sense Realism’, Facta Philosophica, 2, 2000, 179-202. 19 That feature is exemplified in the passages from RFM VI quoted in the text above. Other examples include the following (emphases added): PI §200, ‘Should we still be inclined to say they were playing a game? What right would one have to say so?’; PI §207, ‘Are we to say that these people have a language: orders, reports and the rest? There is not enough regularity for us to call it “language”’. 20 Dummett, The Logical Basis of Metaphysics, London: Duckworth, 1991, p. 93. 21 This point needs care. ‘No one’, Dummett writes, is an authority on whether the sense he attaches to a word is really that which it has in the common language’ (Logical Basis of Metaphysics, p. 93). But each person is normally authoritative about the sense that he attaches to the word. 22 Earlier versions of some of this material appeared in: ‘Wittgenstein’s Externalism and Modern Externalism’, Filosoficky C฀asopis, 2002, 459-77; and ‘Wittgenstein’s Externalism: Context, Self-Knowledge and the Past’, in What Determines Content? The Internalism/Externalism Dispute, ed. T. Marvan, Newcastle: Cambridge Scholars Press, 2006. I am grateful to the respective publishers for permission to reuse that material here. 16