:: Article

Wittgenstein and the Limits of Science

Interview by Richard Marshall.

Bill Child is an expert on the philosophy of Wittgenstein. Here he discusses Wittgenstein’s anti-scientism, how he thought it might infect approaches to philosophy of mind, his use of ‘Inner’ and ‘outer’, the orthodox view about  his ideas about phenomenal concepts and Child’s unorthodox position, on whether Cora Diamond’s view about continuities between early and late Wittgenstein are right, whether there’s a private language argument in the Tractatus,  whether Wittgenstein was a verificationist and whether the answer helps us answer the question as to whether he was a Behaviourist, whether Wittgenstein was a realist or anti-realist about the past and future, and why it matters, the problem of understanding action or perception in causal terms, Cartesianism, Interpretationism, anti-Cartesianism, Donald Davidson and Interpretationism, and whether Wittgenstein can help answer the hard question of consciousness.

3:AM:  What made you become a philosopher?

Bill Child: I was always interested in philosophical questions. I remember at primary school discussing the conceivability of inverted spectrum cases and the question of whether every action is in the end self-interested. I had a teacher when I was 10 who insisted that nothing is impossible. I was sure he was wrong. But I was frustrated that I couldn’t come up with a counterexample that would persuade him.

Later on, when I was thinking about what to study at university, I watched a memorable BBC TV series about modern philosophy – ‘Men of Ideas’.  Each episode involved a single, long conversation between the presenter, Bryan Magee, and a contemporary philosopher. I particularly remember the programme with Isaiah Berlin, who was discussing what philosophy is, and the one with Iris Murdoch, about philosophy and literature. It was that series that introduced me to the idea of philosophy as an academic discipline. Around the same time, I worked through a reading list that had been suggested for prospective philosophy students. Russell’s Problems of Philosophy, I thought at that time, was pretty hard going. But I was gripped by Berkeley’s Three Dialogues Between Hylas and Philonous; I didn’t believe Berkeley’s conclusions, but I couldn’t work out how to resist his arguments. I was involved in various political activities, too, and I noticed with interest the role that was played by a number of philosophers in public life in the UK. Mary Warnock chaired an national inquiry into special educational needs in 1978, which was much discussed in my home (my father had a professional interest in special education). Bernard Williams chaired a famous report on obscenity and film censorship in 1979. The following year, Michael Dummett published an independent report on the events at Southall in West London in 1979; a teacher called Blair Peach had died there after being assaulted by a police officer at an Anti-Nazi League protest. I liked the idea of philosophers using their professional skill in practical ways.

I studied Philosophy, Politics, and Economics at university. When I applied for the course, I actually expected to focus more on politics than philosophy. But I was quickly captured by the precision, clarity and intellectual rigour of philosophy. And I found its problems more compelling; I loved the fact that there was a whole academic subject devoted to the kinds of question that I’d always liked thinking about. All the same, philosophy didn’t seem a very plausible career path when I was an undergraduate – or for most of my time as a graduate student. Margaret Thatcher was Prime Minister and there were deep cuts to university philosophy departments. So my plan was to study philosophy for as long as I could, but I didn’t expect to be able to get a job as a philosopher. Happily for me, the completion of my doctoral thesis coincided with a thaw in the freeze on university appointments and I had the amazing privilege of moving straight from being a doctoral student to having a permanent appointment in philosophy.

3:AM: You’re a Wittgensteinian expert. So let’s start with Wittgenstein’s anti-scientism. You argue that there were three levels of hostility towards science from Wittgenstein. So what was wrong with scientism according to Wittgenstein, and did this bleed into his attitude to science itself?

BC: As you say, I think we can distinguish three levels of hostility in Wittgenstein’s remarks about science and scientism. There is hostility to the scientism that treats science as the only respectable form of enquiry and ignores the value of other kinds of investigation. There is hostility to the spirit in which contemporary science is conducted, which Wittgenstein thinks is the spirit that informs western civilization as a whole. And, occasionally, there are signs of an attitude that goes further than the anti-scientism expressed in these first two forms of hostility: hostility to science itself. I will say a bit about each of those.

First of all, Wittgenstein’s hostility to the scientistic tendency to take science as the model for all enquiry. His objection here is not to the institution or methodology of science as such; it is to the overgeneralization of scientific thinking to forms of enquiry where it is not appropriate. He is thinking of two features of scientific thinking in particular: its focus on causal explanation and its aspiration to achieve generality in its explanations. There are many areas of human enquiry, Wittgenstein insists, where neither causal explanation nor the search for general laws are appropriate.

There is a nice illustration of this theme in Wittgenstein’s critique of Frazer’s anthropological magnum opus, The Golden Bough. Wittgenstein highlights two respects in which Frazer pursues anthropology as though it is a kind of science. In the first place, Frazer aims to understand the meaning of contemporary practices or ceremonies by causally explaining them: tracing them to their causal origins in earlier practices. In the second place, he thinks that anthropology should look for general explanations: he thinks it counts in favour of a putative explanation that it reveals ‘a harmony and consistency’ (Frazer 1994, 748) in the practices people have pursued at different times and places. Wittgenstein rejects both these ideas. He thinks that understanding a practice requires the anthropologist to achieve a sympathetic understanding of the significance it now has for its participants. Tracing the causal origins of the practice, he thinks, is irrelevant to that kind of understanding. Similarly, Wittgenstein rejects Frazer’s search for general rules in anthropology. He acknowledges that there are similarities between the customs and rituals that we find in different times and places. But, he says, ‘besides these similarities, what seems to me to be most striking is the dissimilarity of all these rites’ (Wittgenstein, Philosophical Occasions, p. 143). It is a mistake to think that there must be an underlying feature or motive common to every such rite or festival.

Wittgenstein finds a second form of scientism in Frazer’s anthropology, too. It’s not just that Frazer’s own anthropological practice pursues anthropology as if were a science. It’s also that he treats the magic-involving beliefs and practices of early societies themselves as being a primitive form of science: a system of beliefs about the causes of natural phenomena and a system of practices whose function is to control those phenomena. Wittgenstein agrees that there may be some cases where it is correct to construe a society’s beliefs and rituals in that way. But there are many others where it is not. In most cases, he thinks, magic is not a primitive form of causal understanding; it is, rather, an autonomous system of thought and action that people engage in for its own sake. He likes to press that point by drawing parallels between the magical practices of other societies and things that we do ourselves. For instance: ‘Kissing the picture of one’s beloved. That is obviously not based on the belief that it will have some specific effect on the object which the picture represents. It aims at satisfaction and achieves it. Or rather: it aims at nothing at all; we just behave this way and then we feel satisfied’ (Philosophical Occasions, p. 123). In the same way, he thinks, the action of burning an effigy of one’s enemy need not be based on the belief that it will cause harm to the enemy. It may simply be something people do for its own sake: a way of venting or expressing their hatred.

Now for the second level of Wittgenstein’s hostility: his objection to the spirit of ‘the typical western scientist’ (Culture and Value, p, 7). Science, he writes, is driven by a commitment to making progress, to ‘onwards movement [and to] building ever larger and more complicated structures’; it involves an endless quest for novelty, ‘add[ing] one construction after another, moving on and up, as it were, from one stage to the next’ (Philosophical Remarks p. 7). Science values knowledge only as a means to an end. And ‘the spirit in which science is carried on nowadays’, he complains, is incompatible with a sense of wonder at nature: ‘Man has to awaken to wonder . . . Science is a way of sending him to sleep again’ (Culture and Value, p. 5). We can recognize the kind of spirit that Wittgenstein is talking about, and we may share his distaste for it. For myself, though, I think it’s tendentious and uncharitable to think that this objectionable spirit is associated specifically with science. The scientists I know have as keen a sense of the wonder of nature as anyone else. And restless system-building is not exactly unfamiliar in disciplines that have nothing to do with science. Still, at this level, Wittgenstein is not objecting to scientific method or scientific understanding itself but only to a particular spirit in which, in his opinion, science happens to be pursued. He may be right to object to the spirit he identifies, even if he’s wrong to think it’s a specially associated with science.

But – and this is the third level – Wittgenstein does sometimes display outright hostility to science as such. ‘It isn’t absurd’, he says, ‘to believe that the age of science and technology is the beginning of the end for humanity; that the idea of great progress is a delusion, along with the idea that the truth will ultimately be known; that there is nothing good or desirable about scientific knowledge and that mankind, in seeking it, is falling into a trap. It is by no means obvious that that is not how things are (Culture and Value, p. 56). That goes well beyond the first two levels of hostility. Admittedly, he doesn’t commit himself to the claim that there is literally nothing good or desirable about scientific knowledge; he only says that that idea ‘isn’t absurd’. But even that suggests a real hostility to science.

You ask whether Wittgenstein’s hostility to scientism bleeds into his attitude to science itself. I think his antipathy to scientism and his criticisms of science itself are mutually reinforcing. On the one hand, the idea that there is nothing good at all about science and the pursuit of scientific knowledge looks like an intemperate overreaction to the imperialism of science. A more measured view would combine resistance to scientism with a healthy appreciation of the value of science in its own domain. On other hand, Wittgenstein’s anti-scientism is plausibly reinforced by a distorted conception of what science is. In many of his remarks about science, he treats science as being fundamentally essentialist and reductionist. But that is surely a misrepresentation. The actual practice of science is much more pluralistic than that. If Wittgenstein had had a more realistic view of science, he might have been less antagonistic towards it. That wouldn’t diminish the force of his point that science does not have a monopoly on the truth and that not every question is a scientific question. But it might have made him less pessimistic about the value of scientific knowledge and less hostile to the spirit that motivates actual scientists.

3:AM:  Approaches to philosophy of mind is an area which Wittgenstein thought was in danger of being infected by scientism wasn’t it? Can you say something about what Wittgenstein thought was the threat here?

BC: To see how Wittgenstein thinks the threat of scientism arises in philosophy of mind, think back to what I said a moment ago about the threat of scientism in anthropology. Wittgenstein opposes two kinds of scientism in anthropology: the scientism of doing anthropology as though it is a science; and the scientism of treating magical beliefs themselves as a primitive form of science. In a similar way, he opposes two kinds of scientism in philosophy of mind: the scientism of doing philosophy of mind as though it is a science; and the scientism of treating common-sense psychology as though it is a primitive form of science.

Wittgenstein thought that philosophy in general has a tendency towards scientism; ‘philosophers constantly see the method of science before their eyes and are irresistibly tempted to ask and answer questions in the way science does’ (Blue Book, p. 18). Scientific questions arise, he thought, when we are ignorant of the natural world. Answering them involves discovering new facts: a causal explanation; a natural law; the internal constitution of a kind of stuff. A philosophical problem is quite different; it is ‘a muddle felt as a problem’ (Blue Book, p. 6). The characteristic of a philosophical question, he says, is ‘that we express an unclarity about the grammar of words in the form of a scientific question’ (Blue Book, p. 35). But, he thinks, answering (or responding to) a philosophical question never involves discovering new facts or finding causal explanations. Rather, it involves unravelling the ‘muddle’ that created the sense that there was a problem in the first place.

Here’s an example from the philosophy of mind of how that’s supposed to work. Wittgenstein thought a lot about the phenomenon of seeing-as or, in his terminology, seeing an aspect. That’s the phenomenon we get with ambiguous pictures, when we see the duck-rabbit figure as a picture-duck or a picture-rabbit; and Wittgenstein explores a host of other cases. A constant refrain in that discussion is the question of whether seeing an aspect is a kind of seeing or a kind of thinking; is it an experiential phenomenon or a cognitive phenomenon? Philosophy’s tendency to scientism, he thinks, tempts us to treat this as a factual question, which can in principle be resolved by investigation of the brain and visual system. But Wittgenstein thinks that’s a mistake. Information about what goes on in the brain when someone sees an aspect is interesting in itself. But it cannot settle the question, whether seeing an aspect is or is not really a kind of seeing. For the original question was not about what is going on in a person’s brain when she sees an aspect. It was about whether – or to what extent – the phenomenon of aspect-seeing falls under the existing concept of seeing. And that question can only be addressed at the level of our ordinary psychological concepts; it cannot be answer by shifting to a different set of concepts and a different level of description.

Wittgenstein is clearly right to distinguish between philosophical questions and empirical, scientific questions. To take another example. It is a philosophical question, what it is for a person to have the concept of others’ minds: to think of others as possessors of experiences, beliefs, intentions, and so on. That philosophical question is distinct from the various empirical questions we might ask: how and at what age infants come to think of others as having minds; what regions or mechanisms of the brain are relevant to such thinking; and so on. The empirical information that helps to answer those latter questions does not answer the philosophical question. On the other hand, I think Wittgenstein goes too far in pronouncing that information about the brain and nervous system can never be relevant to addressing the problems that arise in philosophy of mind. Of course the solution of a philosophical problem has to address the question we originally asked; it must not change the subject and answer a different question. But it is a mistake to think that any appeal to information about the brain automatically shifts the subject-matter of our investigation from mental phenomena to something else. We cannot rule out the possibility that figuring out whether or not some particular kind of case involves a change in experience in the ordinary sense of ‘experience’ may involve empirical information that is not available to ordinary common sense.

Now a word about the second kind of scientism that Wittgenstein finds in much philosophy of mind: the scientism of treating common-sense psychology (that’s to say, our ordinary mentalistic talk about ourselves and others) as if it were a kind of science: a primitive theory of the causal mechanisms that produce behaviour. That conception of common-sense psychology comes in two versions: there is the idea that common-sense psychology is a theory about the literally internal, physical causes of behaviour; and there is the more nebulous image of common-sense psychology as a theory about a ‘mental mechanism’ that produces behaviour – a mechanism whose nature is left unspecified, but which we are tempted to picture as ‘gaseous’ or ‘aethereal’ (see Blue Book, p. 47). Opposition to that way of thinking is central to Wittgenstein’s rejection of what he calls the ‘inner-outer picture’. That’s what we’re going to talk about next.

3:AM:  Wittgenstein used ‘inner’ and ‘outer’ in two ways and one of the ways was to say that they were delusions. What is he rejecting and how does this help us understand his own approach to psychology and philosophy of mind?

BC: One way in which Wittgenstein uses the terms ‘inner’ is as a general label for a particular class of phenomena: sensations, thoughts, intentions, emotions, and so forth. To say that something is an ‘inner’ phenomenon in this sense is equivalent to classifying it as a mental phenomenon; it carries no particular philosophical commitments about the nature or status of the mental. That is how Wittgenstein is using the term when he says that ‘the inner […] is sensations + thoughts + images + mood + intention, and so on’ (Last Writings on Philosophy of Psychology I §959). In other contexts, however, he employs the terms ‘inner’ and ‘outer’ to express a specific philosophical picture of the mental: a particular way of conceiving of the mental and its relation to behaviour and the non-mental world. That’s the sense of ‘inner’ he’s employing when he says that ‘the “inner” is a delusion’ (Last Writings on Philosophy of Psychology II, p. 84). He doesn’t mean it’s a delusion that we have sensations, thoughts, and so on. What is a delusion is the conception of the metaphysis and epistemology of mind that Wittgenstein sums up as the ‘picture of the inner and the outer’.

The metaphysical aspect of the inner-outer picture is the idea that mental phenomena exist in an internal realm, ontologically distinct from the ‘outer’ realm of behaviour and external circumstances. The epistemic aspect involves two key ideas. First, there’s the idea that each of us is directly acquainted with our own thoughts and sensations and thus that we all have certain knowledge of what we ourselves are thinking and feeling. Second, there’s the idea that no-one can be acquainted with another person’s thoughts or experiences and thus that one person can never really know what another person is thinking or feeling; others’ thoughts and feelings are hidden behind their outer behaviour. (We’ll see later on that there’s a conception of the semantics of words for mental phenomena that’s closely related to this picture of the epistemology of the mental.)

Wittgenstein rejects the inner-outer picture for multiple reasons. He thinks it misunderstands the metaphysics of the mental and that its account of the epistemology of the mental fundamentally misrepresents our epistemic relation both to our own and to other people’s experiences and attitudes. I’ll talk in more detail about some of reasons later on – when we get onto discussing the private language argument. But, at the most general level, he takes it just to be obvious that in talking of people’s thoughts, intentions, sensations and so on we are not talking about phenomena going on in an inner realm, hidden behind people’s behaviour. Common-sense psychology, he thinks, is an autonomous scheme of description and explanation that we employ in talking of ourselves and others. The terms it uses – ‘pain’, ‘irritation’, ‘intention’, ‘belief’, and so on – are defined by their places within that scheme. And the ordinary practice of applying mental terms to ourselves and others does not involve the idea that, when we are talking about people’s experiences, beliefs, wishes and so on, we are talking about states and events that are literally internal, in the style of the inner-outer picture.

3:AM: There have been many push backs against Wittgenstein’s picture of phenomenal concepts. Can you first of all sketch out the landscape here before looking at how you defend the Wittgensteinian perspective on this? What is the orthodox view of Wittgenstein’s position?

BC: I’ll start by saying something about Wittgenstein’s views and about the contemporary idea of phenomenal concepts. Then I’ll talk about the relation between them. But before all that, I should say something about what the topic is here.

It’s obvious to everyone that we can think and talk about sensations and the qualities of sensations. I can say or think that I have a pain in my finger; that the pain is throbbing, or stabbing, or dull, or intense; that it’s worse than some other pain; that I feel a tingling sensation; that something looks bright red to me; and so on. When I entertain or express thoughts like these, I use words and concepts that refer to sensations and the qualities of sensations. How do those words and concepts function?

Wittgenstein addresses that question in the celebrated private language sections of Philosophical Investigations: roughly §§243 to 315 of the book. It’s helpful to start with the picture of sensations and sensation language that he’s rejecting – which, he thinks, exerts a powerful hold on our imaginations. (Basically, it’s the application to the particular case of sensations of the more general picture of the mental as ‘inner’ that we were just talking about.) The picture starts with the idea that sensations are individuated by their intrinsic, subjective character, and that that subjective character is entirely independent of anything to do with the subject’s behaviour or external circumstances. (That means that it is perfectly possible for two people to be subject to all the same external stimuli, and to be exactly alike in every behavioural respect, but for the subjective character of their sensations to be entirely different: it is possible, for example, ‘that one section of mankind [has] one visual impression of red, and another section another’ (PI §272).) That view of what individuates sensations goes hand in hand with a view about our knowledge of sensations and about the meanings of sensation words. Concerning knowledge, the tempting picture is that the only person who can really know what sensation someone is having is the person herself: ‘Only I can know whether I am really in pain; another person can only surmise it’ (Philosophical Investigations §246). Concerning sensation language, the tempting picture is that words or concepts for kinds of sensations get their meanings by direct, introspective attachment to the sensations themselves. But then, since no-one can know the nature of anyone else’s sensations, it seems that no-one can know the meanings of anyone else’s sensation words. So the language that a person uses to talk about her own sensations is a private language: a language that only she can understand.

Wittgenstein rejects each element of that picture. I’ll say something later on about the detail of his discussion of the supposed possibility of a private sensation language. For now, the important point is just that he rejects the picture of sensation language that goes with the tempting picture of sensations that I’ve just sketched. Kinds of sensations, he insists, are individuated by their links to the external circumstances in which we experience them and the behaviour by which we express them. They aren’t, and couldn’t be, individuated in the completely introspective way it’s tempting to imagine. Of course, the normal links between sensations, circumstances, and behaviour can break down in particular cases. But if we try to imagine there being no such links at all we completely lose our grip on what individuates kinds of sensations.

Now for the idea of phenomenal concepts in current philosophy of mind. We can draw a general distinction between properties on the one hand and concepts of those properties on the other hand. For example, there’s a property of being circular. And there are various different concepts or ways of thinking of that property: we can think of the property of being circular in geometrical terms, for instance; or we can think of it on the basis of the characteristic visual appearance of circular things (‘To be circular is to be that shape’); and so on. It’s plausible that we can draw the same distinction for sensations. There are kinds of sensation: pain, pins and needles, itching and so on. And for each kind of sensation, there are various different concepts or ways of thinking of that sensation. In particular, it’s suggested, there are phenomenal concepts of sensations: concepts or ways of thinking of kinds of sensation that are available only to someone who knows what it’s like to experience them. For instance, consider someone who’s never experienced the kind of tingling feelings that are typically caused by a mild electric shock. The fact that she’s never experienced such sensations doesn’t mean that she has no concept of them and can’t think about them at all. She can, for instance, think of them descriptively – as the kind of sensations that are caused by mild electric shocks. What she can’t do is to think of tingling sensations on the basis of how they feel: she can’t think of them as sensations that feel like this; nor can she think of them on the basis of an ability to recognize such sensations when she feels them. In the jargon: she has a descriptive concept of tingling sensations; but she doesn’t have a phenomenal concept of them.

It’s generally assumed that the idea that there are phenomenal concepts of sensations is incompatible with Wittgenstein’s position. That’s the orthodox view; it’s the consensus both amongst writers on phenomenal concepts and amongst writers on Wittgenstein. And standard accounts of phenomenal concepts do indeed set up the view in a way that Wittgenstein would clearly reject. In particular, they assume that sensations are individuated in a way that is completely independent of any links to external circumstances or behaviour. And they assume that we can set up a phenomenal concept by pure introspective ostension: we feel a sensation, focus on its intrinsic character, and introduce a concept to refer to all sensations of that introspected type in a way that depends on nothing other than our own, introspective resources. If that’s how we think of phenomenal concepts, then they are certainly incompatible with Wittgenstein’s views.

3:AM: And what is your unorthodox understanding of phenomenal concepts and how does it accommodate critics of the orthodox reading?

BC: I think a Wittgensteinian conception of sensations and sensation language is entirely compatible with the idea that there are concepts of kinds of sensation that are available only to those who know what those sensations feel like. I agree that Wittgenstein would reject the central commitments of some of the standard accounts of phenomenal concepts. But we can separate those commitments from the idea that there are distinctive phenomenal concepts.

Here’s the kind of case that I think makes the point. Think again about the kind of tingling sensations that are typically caused by mild electric shocks. Wittgenstein’s view is that the individuation of tingling sensations is tied to the objective circumstances in which they are typically experienced; what a tingling sensation is is the sensation of a mild electric shock. The link is not exceptionless. We can on occasion experience tingling sensations without having an electiric shock. And we can have an electric shock without experiencing tingling sensations. But those are exceptions to the rule. To imagine there being no association at all between tingling sensations and electric shocks is to undermine the individuation of tingling sensations altogether.

Now imagine someone who has never had an electric shock and has never felt tingling sensations. As I’ve already said, that doesn’t stop her having a concept of the tingling sensations. After all, she can discuss those sensations with other people; she might even use their occurrence as a way of diagnosing neurological conditions in others. Then we give her a series of electric shocks, without telling her that that’s what we are doing. She experiences unusual and distinctive feelings. We know, though she doesn’t, that she’s having electric shocks and feeling tingling sensations. She learns, perhaps with the encouragement of others who know the set-up, to classify the new feelings as ‘φ sensations’. So she now has two ways of thinking about tingling sensations: on the basis of her knowledge that electric shocks cause tingling sensations; and on the basis of her experience of tingling sensations themselves. And these seem genuinely distinct concepts or ways of thinking; the discovery that her φ sensations are tingling sensations is genuinely illuminating. It seems right to say that the person I’ve just described acquires a phenomenal concept: a distinct concept of tingling sensations, which depends on her knowing what tingling sensations feel like.

I say that nothing in this story is challenged by Wittgenstein’s views. In particular, it’s consistent with Wittgenstein’s insistence that kinds of sensation are individuated by their links to external phenomena: in this case, the occurrence of electric shocks. Of course, when the subject learns to apply the concept ‘φ’ to herself, she’s not applying it on the basis of the occurrence of the external phenomenon; she applies the concept ‘φ’ in response to the feeling itself. Nonetheless, what she’s picking out – tingling sensations – are individuated by their link to the external phenomenon, just as Wittgenstein requires. If her application of the concept ‘φ’ to herself did not generally keep track with the occurrence of the electric shocks, it wouldn’t qualify as a concept of tingling sensations at all.

Naturally, there’s more to say at many points. Some readers of Wittgenstein will think that my account makes assumptions that he would reject. Some defenders of phenomenal concepts will think that the kind of Wittgenstein-friendly phenomenal concept that I’ve outlined doesn’t pass their tests for being a phenomenal concept. I have replies to such objections. But I hope the basic picture is clear without pursuing those details.

3:AM: Cora Diamond finds continuities between the Tractatus and the Philosophical Investigations. Are you generally sympathetic to this approach to Wittgenstein?

BC: It’s clear to anyone who reads Wittgenstein’s work that there are continuities between the Tractatus and Philosophical Investigations as well as discontinuities. In the history of the interpretation of Wittgenstein’s work some readers have put more weight on the continuities: others have highlighted the discontinuities. In the early years after the publication of Philosophical Investigations in 1953, the discontinuities were more commonly stressed. The standard view was that Wittgenstein was the author of two distinct and diametrically opposed philosophical visions. With the passage of time, many commentators came to see that view as doing insufficient justice to the continuities in his work. Anthony Kenny, for instance, writing in 1973, wrote that ‘the posthumous publication of the works of the thirties shows that [the standard view] is too simple. There are many connections between the earlier and the later work, and many assumptions common to both’ (Wittgenstein, Penguin, 1973, p. 219). In particular, Kenny argued that the ‘theses which made up the logical aspect of the picture theory’ (p. 227) in the Tractatus were modified rather than abandoned in Wittgenstein’s later work. And he drew attention to ‘the permanence of [Wittgenstein’s] general conception of philosophy throughout his work (p. 229). Kenny is just one example. Other writers have highlighted other kinds of continuity between Wittgenstein’s earlier and later works.

I think it’s pointless to argue about whether the continuities or the discontinuities are more important. We do best simply to understand what they are without feeling constrained to produce an overall balance sheet. We mustn’t overlook the genuine continuities between the Tractatus and Philosophical Investigations. But equally, we shouldn’t overstate the degree of continuity.

I’ll mention three of the continuities between Wittgenstein’s early and later work that I find most striking. Of course there are many others.

First, there’s a cluster of ideas about the source of philosophical problems and the proper response to them. In both Tractatus and Philosophical Investigations, Wittgenstein draws a distinction between the surface structure of ordinary language and the deeper structure of language and of the phenomena we use it to describe. In both works he holds that the words that have the same surface structure may function in quite different ways. And there’s a related continuity in his view about the source of philosophical problems. In the Tractatus he says that ‘most of the propositions and questions of philosophers arise from our failure to understand the logic of our language’ (TLP 4.003). In Philosophical Investigations he says that philosophical problems arise when we are misled by ‘certain analogies between the forms of expression in different regions of our language’ (PI §90) into thinking that the phenomena we are talking about when we use those expressions are similarly analogous. And in both works there is the idea that the proper response to philosophical problems is not to take them at face value and try to solve them producing a philosophical theory but, rather, to dissolve them by showing that they were not genuine problems at all. ‘We cannot give any answer to [philosophical questions]’, Wittgenstein writes in the Tractatus, ‘but can only point out that they are nonsensical’ (TLP 4.003). And in Philosophical Investigations: ‘The results of philosophy are the discovery of some piece of plain nonsense and the bumps that the understanding has got by running up against the limits of language’ (PI §119).

A second striking continuity in Wittgenstein’s views is the idea that the only point of view from which we can contemplate language and logic is the point of view internal to our language and logic; there is no external point of view from which philosophy can justify or criticize our ways of thinking and reasoning. That idea is articulated in the preface to the Tractatus, in the observation that we cannot draw a limit to thought, since in order to do so ‘we should have to find both sides of the limit thinkable (i.e. we should have to be able to think what cannot be thought’). And it is a constant theme in Philosophical Investigations. It lies, for instance, behind the claim that philosophy leaves everything as it is: that philosophy can in the end only describe and cannot justify the actual use of language (see PI §124).

Third, there is the idea that words have the meanings they do because we use them in the way that we do. That is explicit in both the Tractatus and Philosophical Investigations. But the conception of ‘use’ in the two works is importantly different. In the Tractatus, the sense of a proposition is a matter of ‘how things stand if it is true’ (TLP 4.022); and ‘to understand a proposition means to know what is the case if it is true’ (TLP 4.024). Accordingly, Wittgenstein’s early account of sense is exclusively an account of truth-conditional content. And the use of words that is relevant to their meaning is exclusively their use in saying that some state of affairs does or does not obtain. In his later work, Wittgenstein’s conception of use is much more pluralistic: embracing, for instance, pragmatic and expressive aspects of our use of words.

You asked in particular about Cora Diamond’s work and the continuities she sees between Wittgenstein’s earlier and later work. I think there are two things that Diamond’s writings have been particularly helpful in highlighting. One is the quietistic, anti-theoretical, therapeutic character of Wittgenstein’s later philosophy. The other is the puzzle posed by the apparent incoherence of the Tractatus. Those two themes come together in Diamond’s idea that the right way to respond to the apparent incoherence of the Tractatus is to see Wittgenstein’s early work as having essentially the same quietistic, therapeutic character as his later work.

On the face of it, the Tractatus advances a specific theory of language and logic and a specific metaphysical view about the nature of reality. At the end of the book, however, Wittgenstein tells us that the propositions of the Tractatus are nonsensical. Furthermore, the nonsensicality of the propositions of the Tractatus seems to be a direct consequence of the theory of meaning that’s advanced in the book. For, by the lights of the Tractatus itself, propositions like ‘A propositional sign is a fact’ (TLP 314) or ‘In a proposition a name is the representative of an object’ (TLP 3.22), which on the face of it articulate the Tractatus’s account of language, are attempts to say something that cannot be said but is shown in the functioning of ordinary, empirical propositions. That is the traditional way of seeing things. But it poses a deep puzzle. For how can a book that is made up of nonsensical propositions succeed in communicating a theory of language and logic – or, for that matter, anything else?

Many readers of the Tractatus have responded to the puzzle by concluding that Wittgenstein’s overall position is untenable. The propositions of the Tractatus evidently do advance a definite and intelligible theory of meaning. So if Wittgenstein is right in saying that one consequence of that theory is that Tractarian propositions themselves are nonsensical, we should reject or amend the Tractatus’s theory of meaning.

Diamond’s reading offers a radical alternative. Her central commitment is to respect Wittgenstein’s claim that the propositions of the Tractatus are nonsense. (She acknowledges that Tractatus 6.54, which says that the propositions of the Tractatus are nonsensical, must be an exception. That proposition, she thinks, is not only intelligible; it is true.) Since Tractarian propositions are nonsense, they do not advance any substantive theory at all. At first glance, Wittgenstein’s remarks about pictures in TLP 2.1-2.225 look like a substantive theory of representation. And at first glance the claim that a proposition is a picture of reality, that names are representatives of objects, and so on look like a substantive theory of linguistic meaning. But, Diamond insists, those appearances are deceptive. The Wittgenstein of the Tractatus is no more concerned to advance a substantive theory of meaning and representation than is the Wittgenstein of Philosophical Investigations. The point of the Tractatus, like the point of some of Wittgenstein’s later writings, is to take us through a therapeutic journey in which we start with a form of words that seems to express an intelligible philosophical view and are gradually shown that they don’t really express any view at all.

I think Diamond is absolutely right to highlight both the problem of finding a coherent overall reading of the Tractatus and the anti-theoretical character of Wittgenstein’s later work. But I’m not convinced by her proposal to resolve the tension in the Tractatus by finding in the early work a therapeutic quietism of fundamentally the same kind as that which he certainly did adopt in his later work. On the evidence of the Tractatus itself, and of what Wittgenstein subsequently said about it (both at the time of its completion and publication and in subsequent writings and discussions), I find it impossible not to believe that the account of language and logic set out in the Tractatus was intended as a substantive theory of meaning and logic, which Wittgenstein was convinced was correct. I agree with Diamond that there are important continuities between Wittgenstein’s early and later work. Indeed, the continuities I mentioned earlier on include some of those that she has done as much as anyone to stress. But her overall take on the Tractatus, in my view, involves reading back into the earlier work features that really only emerge in the later work.

3:AM: Regardless of the general continuity reading, are there good reasons for reading a version of the private language argument into the Tractatus? Could you sketch what the private language argument is and then say whether it is there in the Tractatus?

BC: This takes us back to some of what we were discussing a moment ago, when you were asking about phenomenal concepts. I said a bit about the private language argument there. I’ll fill in some more detail here. Then I’ll say something about Cora Diamond’s suggestion that, long before Wittgenstein put together the private language sections that appear in Philosophical Investigations, the Tractatus already contained a different line of thought, with a different target, that it’s reasonable to call a private language argument.

The private language sections in Philosophical Investigations get going with the question, whether it is conceivable that there could be a language whose words ‘refer to what only the speaker can know – to his immediate private sensations. So another person cannot understand the language’ (PI §243). Wittgenstein takes it for granted that our ordinary language for talking about sensations is not a private language in that sense. If I say ‘I have pins and needles’ or ‘I feel toothache’, you understand perfectly well what I mean. But even if we agree with him about that, it is tempting to think that I could introduce other terms that refer to something that really is knowable only by me: the particular subjective quality of the experience I have when I have pins or needles or when I feel toothache. Unlike our ordinary sensation language, it seems, a language made up of words like that really would be a private language in Wittgenstein’s sense.

The burden of PI §§243-315 is that a private sensation language like that is not conceivable. There have been numerous different interpretations of the thinking that gets Wittgenstein to that conclusion. I agree with the widespread view that a crucial part of Wittgenstein’s dialectic is the following line of thought. First, for a word to have a meaning, there must be a standard of what counts as a correct application of the word and, therefore, a distinction between applying the word correctly and applying it incorrectly. Second, in the case of the putative private sensation language, it would be impossible to establish and maintain such a standard of correctness. But why does Wittgenstein think that would be impossible?

One popular suggestion is that he thinks the only thing that can supply a standard of correctness for an application of any word is communal agreement about which applications are correct and which are incorrect. But there could be no communal agreement on the correctness or incorrectness of applications of words in a private sensation language. For, ex hypothesi, a private sensation language would be a language that could be understood by only one person. It follows, on this line of thought, that there could be no distinction between a correct and an incorrect application of the words of a private sensation language. So the putative private sensation language fails to meet an essential condition on being a language at all.

As I say, that’s a common way of reading the private language sections. But I don’t think it’s right. As I read him, Wittgenstein’s point is rather this. If someone were to have a private language, they would have to individuate the kinds of sensation they were talking about in a way that was completely independent from any links to external circumstances or behaviour. (Otherwise the language wouldn’t be a private language; other people could understand it just as well as the subject herself.) But there can be no such way of individuating kinds of private of sensation types. In particular, we cannot individuate a kind of sensation simply by an act of introspective attention. Putting aside that and other questions of interpretation, however, everyone can agree that the private language argument in Philosophical Investigations is concerned with the first-person use of sensation words; it’s concerned with the language that a person uses for talking or thinking about her own sensations.

What does all this have to do with the Tractatus? Cora Diamond has suggested that the Tractatus contains a private language argument. But the argument she has in mind is quite different from the one I have describing, which appears in Philosophical Investigations. The ‘Tractatus private language argument’ that Diamond describes is concerned not with the language each of us uses to describe our own inner sensations but rather with the language we use to describe other people’s sensations. Here, with inevitable simplifications, is how Diamond sees things.

At the time of the Tractatus, Diamond thinks, Wittgenstein assumed that the language that each person uses to talk about her own sensations was a private language, intelligible only to her. In a slogan, my first-person uses of sensation words refer to private objects in my own mind. (That, of course, was the view he went on to reject in Philosophical Investigations.) But what about the language we use to talk about other people’s sensations? On the current view, when someone else – Angela Merkel, say – talks about her pains, her word ‘pain’ (or ‘Schmerz’) is intelligible to her but unintelligible to me or anyone else. For only she is acquainted with her pains and only she knows the nature of the sensations her word picks out. What, then, do I mean when I say ‘Merkel is feeling pain’? Since I’m not acquainted with Merkel’s sensation, I cannot directly attach my word ‘pain’ to her inner sensation; I cannot bring her sensation to consciousness and lay it down that, when I apply my word ‘pain’ to Merkel, it refers to that kind of sensation. But, you might think (and Bertrand Russell did think), I can attach my word ‘pain’ to Merkel’s inner sensation indirectly. For I can lay it down that, when I use the word ‘pain’ of Merkel, I mean the kind of inner sensation that she is feeling when she truly says ‘I’m feeling pain. That provides a mechanism by which a kind of private object in Merkel’s mind, with which I am not acquainted, can play a role in the language I use to talk about Merkel’s sensations. The truth or falsity of my sentence ‘Merkel is in pain’ is a matter of the presence or absence in Merkel’s mind of an inner object of that kind.

The private language argument that Diamond finds in the Tractatus is directed against that account of the way that sensation words function in the third-person case. And the core of the argument is very simple. Wittgenstein’s view in the Tractatus is that all language is ultimately analysable down to the level of elementary propositions, at which simple elements of language (names) stand directly for simple elements of reality (objects). It is a fundamental feature of that view that if an object plays a role in my language – if it features in the truth-conditions of sentences I understand – I must have a name for that object. So, Diamond argues, the Tractatus directly rules out the idea that the truth-conditions of my sentences about other people’s sensations involve private objects that I cannot name.

I think it’s a mistake to attribute this argument to the Wittgenstein of the Tractatus. I completely agree that, according to the Tractatus, we cannot refer to or quantify over objects that we cannot name. But I don’t think there’s anything in the Tractatus to rule out the possibility of our giving names to private objects in other people’s minds. Of course that would be impossible if giving a name to an object required being perceptually or introspectively acquainted with the object; for, on the view under discussion, we are not and cannot be acquainted with other people’s sensations. But there is nothing in the Tractatus to suggest that he regarded naming as requiring acquaintance in that sense. Is there anything else in the Tractatus that would block my giving names to private objects in other people’s minds? Well, the Tractatus does say that words mean what they do because we use them as we do. Someone might argue that the irrelevance of private objects in others’ minds to our third-person sensation language follows from that idea. For, they may say, the presence in someone else’s mind of an object that I can never detect can make no difference to my actual use of a sensation word: to my actually accepting or rejecting the claim that Merkel is in pain. And it is true that, when Wittgenstein discussed sensation language in Philosophical Investigations, he offered an argument of exactly that form. (I’m thinking of the ‘beetle in the box’ remarks in PI §293.) But to get an argument like that, we have to construe ‘use’ in a way that ties the use of words to evidence, knowledge, and practical consequences. Such considerations clearly play an important role in Wittgenstein’s later conception of use. But I see no evidence that he thought of use in that way in the Tractatus. On the contrary, he was explicit that his concerns were logical, not epistemological. And he offered no account at all of how a particular name might come to stand for a particular object. So I think the idea that the Tractatus contains a private language argument of this sort involves projecting onto the Tractatus epistemic and pragmatic concerns that were important in Wittgenstein’s work but had no part in his early work.

3:AM: When it comes to the mind, was Wittgenstein a verificationist? What does examining verificationism show us about Wittgenstein’s approach to the mind? Does it help answer the question as to whether he was a Behaviourist?

BC: For a while after his return to philosophy in 1929 Wittgenstein held an explicitly verificationist view of meaning. He said that ‘the sense of a proposition’, he said, ‘is its method of verification’ (Wittgenstein and the Vienna Circle, p. 79). Or again, ‘In order to know the sense of a proposition, I should have to know a very specific procedure for when to count the proposition as verified’ (Wittgenstein and the Vienna Circle, p. 47). He soon gave up that extreme form of verificationism. But he didn’t give up the idea that there is a significant relation between the meaning of a proposition and the ways in which we can tell whether it’s true or false. In Philosophical Investigations, for instance, he writes that ‘asking whether and how a proposition can be verified is only a special form of the question, “How do you mean?” The answer is a contribution to the grammar of the proposition’ (PI §353). But exactly how close does he think the relation is between what a proposition means and our ways of telling whether it is true?

The position he reaches in his late writings on philosophy of mind is this. Take the example of thinking. In determining the meaning of the term ‘thinking’, an essential role is played by cases in which we can know what someone is thinking on the basis of what she says and does. But, having grasped the concept of thinking in connection with cases like that, we can perfectly well understand claims about what someone is or was thinking in cases where there is no possibility of knowledge. Wittgenstein discusses a particular nice example in his Lectures on Philosophical Psychology and in Remarks on Philosophy of Psychology. Lytton Strachey, in his biography of Queen Victoria, speculates about what thoughts might have passed through Queen Victoria’s mind in her dying moments. We understand Strachey’s speculations. But we have no way of telling what Victoria was in fact thinking at the time. And we can suppose, for the sake of argument, that there was never any way of telling what she was thinking. Even if we had been present at the time, it would have been impossible to tell what she was thinking. How do we make sense of the ascription of thoughts in cases of this sort, where there is no possibility of verification? We can find a number of different suggestions in Wittgenstein’s remarks.

One suggestion is that cases in which we can tell what someone is thinking on the basis of what she says and does play an ineliminable role in fixing the reference of the term ‘thinking’, but that once its meaning is fixed in that way, the term can be straightforwardly applied without further explanation to pick out the same property in cases where there is no possibility of verification. That is a resolutely non-verificationist view. And some of Wittgenstein’s remarks make it look as he would endorse it. He says, for example, that given a background of cases in which I can tell that someone is in pain, I can make sense of the claim that a sleeping or anaesthetised person is feeling pain even if I have no idea at all about how one could go about telling whether she is in pain or not. (For that claim, see Last Writings on the Philosophy of Psychology II p. 57.)

Other remarks, however, make it look as if he finds such a view unsatisfying. They suggest the idea that, though the meaning of the word ‘thinking’ in the Queen Victoria case is parasitic on its use in more basic cases, it is not fully determined by that use. In particular, Wittgenstein suggests, there is a substantial question about what we are doing when we use mental terms in Queen Victoria cases. On this view, explaining the meaning of the word ‘thinking’ as it’s applied in the Queen Victoria case involves giving a direct account of the nature and point of the practice of speculating about thoughts whose ascription cannot possibly be verified. This second view avoids verificationism. But it proposes of a way of understanding the Queen Victoria cases that seems strikingly revisionary of common sense.

A third suggestion in Wittgenstein’s late writings is that the meaningfulness of a sentence like ‘Perhaps N had the experience E but never gave any sign of it’ depends on its having a ‘possible application’. And Wittgenstein proposes, as one possibility, that such a sentence might be applied on the basis of traces of experience that were found in N’s brain (see Remarks on the Philosophy of Psychology I §157). That suggestion avoids behaviourism; it doesn’t tie the meaningfulness of mental ascriptions in Queen Victoria cases to anything to do with a person’s behaviour or behavioural dispositions. However, it is verificationist in spirit, for it ties the meaningfulness of such ascriptions to there being some method in principle of determining whether or not they are true.

In the end, I think it’s not clear exactly how Wittgenstein wants to treat the Queen Victoria case and others like it. He is clear that we do understand such ascriptions. But, as far as I can see, he does not commit himself to any particular view about how we understand them. At any rate, we can find a number of different suggestions explored in his discussions of the issue. That said, none of the suggestions I have mentioned here offers a specifically behaviourist account of the meaning of mental ascriptions in Queen Victoria cases. To that extent, his treatment of these cases gives us reason not to regard him as a behaviourist.

3:AM:  You ask this: ‘Did Wittgenstein endorse a form of anti-realism about past thoughts and attitudes?’ How do you answer and what’s at stake here? Does realism/anti-realism shed light on Wittgenstein’s metaphysics, or is there no metaphysics really going on in his philosophical approach?

BC: I’ll first explain why I asked that question, and then say something about my answer.

In the last sixty sections of Philosophical Investigations, Wittgenstein discusses our capacity to remember our earlier intentions, wishes, and emotions. He begins with a phenomenon that was also discussed by William James in The Principles of Psychology. Suppose I’m interrupted while I’m speaking, and I’m subsequently asked what I was going to say. In many cases, I can unhesitatingly report what I was going to say. What goes on in such a case; how do I know what I was going to say? Wittgenstein makes a compelling point about the phenomenology and epistemology of this kind of case. When I report what I was going to say, he observes, I do not normally do it by recalling a conscious rehearsal of the words that I went through at the time – for we don’t normally engage in any such rehearsal; nor do I read off what I was going to say from other remembered experiences; nor do I work out what I was going to say by interpreting the actions I remember doing at the time, or the situation I was in. I just remember what I was going to say. My knowledge of what I was going to say is immediate; it isn’t grounded in anything else.

All of that seems absolutely right. But how should we think of the metaphysics of the situation? (As you hint in your question, many readers of Wittgenstein will think that that’s a thoroughly un-Wittgensteinian question. I’ll come back to that issue later.) We naturally take a realist view of memory in this kind of case. We think there is a fact of the matter about what I intended to say when I was interrupted, and that it’s independent of any subsequent belief or judgement I form about what I was going to say. When I remember what I was going to say I have a true belief that is appropriately related to the past facts. That, as I say, is the ordinary view of things. But it’s sometimes suggested that we should take a quite different, anti-realist, view of the matter. On the anti-realist view, what makes it true that someone was going to say such-and-such is, to put it crudely, simply her retrospective inclination to judge that that is what she was going to say. We have a practice of reporting what we were going to say when we were interrupted. Such reports are standardly accepted as true. And we describe someone who makes such a report as having ‘remembered’ what she was going to say, and someone who cannot tell us what she was going to say as having ‘forgotten’. But, for the anti-realist, the talk of ‘remembering’ and ‘forgetting’ is just another part of the practice; it does not imply that there is an independently-existing past fact of the matter that the subject is recalling or failing to recall.

On the face of it, the anti-realist view is strongly counterintuitive. The idea that it may take something that happened afterwards to show what someone intended at an earlier time is plainly right. But the idea that intentional properties can be constituted retrospectively is hard to accept; how could something that happens now make it the case that the subject had a particular intention at some earlier time? Given Wittgenstein’s general philosophical perspective, you would expect him to endorse this common-sense, realist view. And in many passages, he does precisely that. Think, for instance, about his discussion of the claim that, when I gave someone the order ‘+ 2’, I meant him to write ‘1002’ after ‘1000’:
‘But I already knew, at the time when I gave the order, that he ought to write 1002 after 1000.’ – Certainly; and you can also say you meant it then; only you should not let yourself be misled by the grammar of the words ‘know’ and ‘mean’. For you don’t want to say that you thought of the step from 1000 to 1002 at that time – and even if you did think of this step, still you did not think of other ones (PI §187).

In passages like that, Wittgenstein is not rejecting the realist thought that, if it is true that I meant him to write ‘1002’ after ‘1000’, there was something about me at the time in virtue of which it is true. What he is rejecting is only the idea that what makes it true that I meant the pupil to put ‘1002’ after ‘1000’ is something that was going through my mind at the time I gave him the order ‘+ 2’. Wittgenstein’s own view is, very roughly, that what makes it true that I meant him to put ‘1002’ is the whole complex of dispositions and abilities I had at the time. We could call that view disposition-based realism.

I think Wittgenstein takes a similar view of intentional properties more generally. He accepts the basic realist intuition that what mental properties a subject has at a time is a matter of the dispositions she has at that time. It’s not a matter of her later dispositions; in particular, it is not a matter of her later disposition to self-ascribe earlier intentions, meanings, wishes and so on. However, his discussion of our capacity to remember our past intentions and meanings contains a number of elements that might suggest a less realist view. For instance, there is the idea that we should ‘regard the language-game [of telling someone about what I wished in the past] as the primary thing. And regard the feelings, and so forth, as a way of looking at, interpreting, the language-game’ (PI §656). There is the suggestion that a later judgement may make the connection between an earlier action or utterance and its intentional object, rather than (or, perhaps, as well as) reporting a connection that already existed (PI §682 ff). And there is his attitude to counterfactuals like ‘If you had asked me at the time, I would have said so-and-so’: he suggests in various places that counterfactuals like that don’t say something about the past but rather function in some other way: for instance, as statements of my current view of what I meant or intended in the past. I acknowledge that those and other elements in Wittgenstein’s discussion might be read as pointing to a form of anti-realism about past intentional states. But I’ve argued that that would be a misinterpretation of his position.

But am I myself misinterpreting Wittgenstein’s discussion? Some readers of Wittgenstein will protest that my question, ‘Should we take a realist or an anti-realist view of self-ascriptions of past attitudes?’, is fundamentally non-Wittgensteinian. The right attitude, they will say, is something like the following. We have a practice of reporting our past attitudes. In appropriate cases, we call such reports ‘memories’ and say that they express knowledge of the subject’s past attitudes. But we can go no further than describing these facts about our practice. To press the question, ‘Realism or anti-realism about memory of intentional states?’, is in effect to ask whether the phenomenon we call ‘remembering an intention’ is really a form of memory: whether we are right to call it ‘remembering’. And such attempts to justify, criticize, or explain our practices are always a mistake. If it is correct, by the rules for the use of our concepts ‘memory’ and ‘intention’, to call something a memory of an intention, then it is a memory of an intention; there is, and can be, no more to say than that.

I agree that there is, for Wittgenstein, no question of philosophy criticizing or correcting our ordinary practice. But what philosophy can do, according to him, is to achieve a reflective understanding of a practice that we have mastered but of which we lack a clear overview. Now suppose we concluded that the anti-realist view of the of the practice of reporting our past intentional states was the correct view. That wouldn’t mean concluding that someone who, by the normal standards, counts as remembering what she was going to say is not really remembering. Rather, it would mean concluding that ‘remembering’ in these cases has one kind of use rather than another: replacing a bad picture of its use with an accurate picture. And that seems to me to be a respectably Wittgensteinian position.

3:AM:  What is the prima facie problem with understanding action or perception in causal terms, and does it commit us to some sort of Cartesianism?

BC: When modern philosophers say that a conception of the mind is ‘Cartesian’, they don’t mean that it accepts Cartesian dualism – the view that thinkers are non-physical beings and that mental states are non-physical states. What makes a view Cartesian, in this sense, is that it conceives of mental phenomena as internal entities whose identity is wholly independent of any relation to the subject’s behaviour or external circumstances: entities that could be as they are however things were in the world beyond the subject’s mind. That’s the view of the mental that Wittgenstein was rejecting when he argued against the ‘inner-outer’ model. And a view can be Cartesian, in this sense, even though it rejects Cartesian dualism.

As your question suggests, lots of philosophers have thought that accepting a causal view of perception or action commits us to a view of mental phenomena that is, in this sense, Cartesian. And it is true that many philosophers who have thought of perception and action in causal terms have accepted a broadly Cartesian framework. In this tradition, a perceptual experience is thought of as an inner mental item. What it is to see a tree, say, is to have inner experiences as of a tree and for those inner experiences to be appropriately caused by the presence of a tree. Similarly, beliefs and desires are thought of as inner states of a behaviour-causing mechanism. What it is for one to raise one’s arm intentionally, say, is for one’s arm to go up and for that bit of behaviour to be caused in an appropriate way by the interaction of relevant internal states.

Those who object to causal views of perception and action typically think of causal theories in the way I have just sketched. And they have multiple objections to the causal view of mind, thus conceived. For instance, they argue that the causal view of vision commits us to some form of indirect realism, which is objectionable because it would make it impossible to know, or even to think, about an objective, mind-independent world. They argue that the conception of an action as a bit of behaviour that is appropriately caused by a belief and a desire, conceived as internal states, loses the intuitive idea that an action involves someone’s doing something; it represents people not as agents but simply as the receptacles of causal processes. They argue that, whereas causal theories try to analyse perception and action in terms of a general, topic-neutral relation of causation, the truth is that perception and action are irreducibly mental and resist all such analysis. Most generally, they think it’s an error to construe common-sense psychology as an account of the states of an internal mechanism that produces behaviour.

I’m sympathetic to many of these points. I agree that we should reject the kind of causal theories I outlined above, which assume a Cartesian view of experiences and mental states. But that doesn’t mean that we should reject causal theories of perception and action altogether. The problem isn’t the causalism. It’s the Cartesianism. And we can perfectly well develop causal views of mental phenomena in a non-Cartesian framework.

3:AM:  What is Interpretationism, anti-Cartesianism and what consequences does it have for the ontological status of mental phenomena?

BC: When we interpret someone, we aim to make sense of her by attributing beliefs, desires, intentions, emotions, and so forth in the light of which her behaviour is intelligible as, more or less, rational action. The guiding intuition of interpretationism is that we can give an account of what it is to have propositional attitudes by considering what it takes to be interpretable as believing such-and-such, or wanting such-and-such, or intending to do so-and-so, and so forth. As Donald Davidson puts it, ‘what a fully informed interpreter could learn’ about a person’s attitudes on the basis of what she says and does ‘is all there is to learn’ (Subjective, Intersubjective, Objective, Oxford University Press: 2001, p. 148). Or, in Daniel Dennett’s formulation, ‘all there is to being a true believer is being a system whose behaviour is reliably predictable via the intentional strategy’ (The Intentional Stance, Cambridge, Mass.: MIT Press, 1987, p. 15); ‘the intentional strategy’ is Dennett’s term for the approach we take when we make sense of a person by ascribing intentional states to her. The interpretationist understands common-sense psychology as an autonomous scheme of description and explanation that we use in talking about ourselves and others. And she thinks of the concepts of the propositional attitudes as being fully defined by their place in that scheme of description and explanation. In ascribing beliefs, desires, and intentions to a person, we are describing the person, and explaining her actions, in terms of that scheme. Of course, at the physical level, human behaviour issues from a complex physical system: the brain and nervous system. When anyone perceives anything, or does anything, there is a vast array of minute physical processes going on inside her: without them, perception and action would be impossible. But, the interpretationist insists, when we use the concepts of common-sense psychology to talk about a person, we are not talking about those internal, physical states and processes; we are talking about her beliefs, desires, and so on. And there is no assumption that her inner physical states are organized in a way that mirrors the intentional states we interpret her as having. All that matters for the truth of our common-sense ascriptions of propositional attitudes is that they meet the ordinary standards of interpretation. So the only constraint on the internal processing is that it produces behaviour that makes her interpretable as a rational agent.

The interpretationist’s way of thinking of the mental contrasts sharply with the Cartesian view, on which propositional attitudes are taken to be internal, causally-interacting states of an internal mechanism, and common-sense psychology is understood as a system for describing those internal states. Interpretationism insists that common-sense psychology should not be construed as a rudimentary science of the internal causes of behaviour; Cartesianism maintains that that is precisely what it is.

I have illustrated the interpretationist position with quotations from Davidson and Dennett. But there are close and striking similarities between their forms of interpretationism and Wittgenstein’s approach to the mental. The fundamental interpretationist idea that mental concepts belong to a sui generis system of description and explanation, governed by norms distinct from those that govern the physical scheme of description and explanation, echoes Wittgenstein’s idea that the language-game of applying mental terms to ourselves and others is autonomous and that, as he puts it, ‘the inner differs from the outer in its logic’ (Last Writings on the Philosophy of Psychology, p. 62). And the interpretationist’s idea that it is impossible to codify or systematize the principles for reaching a correct interpretation of a person’s words and attitudes echoes Wittgenstein insistence there are no exact rules of evidence for ascribing mental states on the basis of what a person says and does.

3:AM:  Donald Davidson puts them together doesn’t he? How does he try and make them compatible and do you?

BC: Yes, exactly. Davidson takes an interpretationist approach to the mental. He’s also done as much as anyone to make the case for a causal conception of mind. In particular, he argued that reason-giving explanation (i.e. explanation in which we explain an action by citing the agent’s reason for performing it) is a form of causal explanation.

It’s sometimes said that Davidson’s interpretationism and his causalism belong to two distinct projects and that the two projects cannot coherently be combined: if we take the causalism seriously, we can’t be content with an interpretationist approach to the mental; if we accept the interpretationism, we’re not thinking of common-sense psychology in causal terms at all. I don’t think that’s right. As Davidson understands it, the interpretative project is itself a causal explanatory project. That’s the point of his argument for a causal view of action. The argument shows that interpretative, reason-giving explanations explain actions by causally explaining them. So it’s not right to say that, in Davidson’s picture, beliefs and desires have two distinct roles: on the one hand, figuring in interpretative explanations; and, on the other hand, standing in causal relations. The interpretative explanations are themselves causal explanations. (Davidson doesn’t think that every causal explanation is an interpretative explanation. We only have interpretative explanations where we have agents: beings who act for reasons. But he does think that every interpretative explanation is a causal explanation. I agree with him.)

All the same, we can legitimately ask about how exactly we’re to make sense of reason-giving explanation as a kind of causal explanation. For instance, what is the cause in a case of reason-explanation; and what is the effect? Abstracting from the detail, a central feature of Davidson’s answer is his combination of monism about events (‘mental events are identical to physical events’) with his denial that there is any system of strict laws on the basis of which mental events can be predicted and explained (‘the anomalism of the mental’). He calls the combination ‘anomalous monism’. It’s an elegant and appealing view, and it’s an ingenious way of accommodating Davidson’s combination of causalism and interpretationism. I don’t think we should in the end accept it. But it has a lot going for it nonetheless.

Davidson thinks of causation as a relation between events. Wherever there is causation, there are causally related events. And those events can be picked out and categorized in many different ways. In particular, we can pick out events in mental language: ‘Her act of raising her arm’, ‘His deciding to buy bread’. And we can pick out events in physical language: ‘The firing of such-and-such a neuron’, ‘The contraction of her muscles’. In Davidson’s picture, mental and physical vocabulary pick out the same events. That is the force of his monism about events. But there are no systematic relations involving kinds of events picked out in mental terms; in particular, there are no strict laws linking kinds of events picked out in mental terms and kinds of events picked out in physical terms. That is the force of Davidson’s anomalism about the mental. We can illustrate Davidson’s way of seeing things. Suppose I walked to the shops because I wanted to buy some bread. That’s a causal explanation. Where there’s a causal explanation, there are causally related events; the event of my deciding to buy some bread (e1) caused the event of my walking to the shops (e2). There’s no systematic relation between events of those mental kinds; it’s not true that every event of deciding to buy some bread will cause an event of walking to the shops. But e1 and e2 are not just mental events; they’re physical events. And, described in physical terms, e1 and e2 do instantiate a law; every event of exactly the same physical kind as e1 causes an event of exactly the same physical kind as e2.

I agree with Davidson’s interpretationist approach to the mental and with his anomalism about the mental. And I agree with him that we have to understand common-sense psychology in causal terms. But I’m sceptical about Davidson’s event monism and I think we can have a perfectly good combination of interpretationism and causalism without it. After all, in many cases psychological explanation doesn’t have much to do with causal relations between events. ‘She hopes that the UK won’t leave the European Union because she believes that it would make the country more inward-looking and harm the economy’ is a causal explanation; but its status as a causal explanation doesn’t seem to depend on the obtaining of a causal relation between particular events. And even where common-sense psychology does talk about events, there’s no good reason to think either that those events are the same as the events that are talked about in the physical or life sciences or that there’s any natural way of matching up mental events with collections of such events. I think the causal status of common-psychology is a matter of the form of common-sense psychological explanation itself; it doesn’t need to be underwritten in the way Davidson envisages by the thesis that mental events are physical events. Of course, mental causation doesn’t just float free of lower-level causal relations. Causal relations described at the level of common-sense psychology supervene on microphysical causation relations; if you fix the lower-level causal facts, you thereby fix the higher-level causal facts. But we can accept that without accepting Davidson’s event-monism.

 3:AM: Returning to what you discussed at the beginning, does your work on action and perception suggest that scientific approaches that focus exclusively on causality may be in some respects in error?

BC: The sober core of Wittgenstein’s anti-scientism is the idea that science isn’t everything. Other modes of enquiry, other forms of description, and other kinds of explanation are equally important and equally legitimate. I think Wittgenstein is completely right about that. And the principle applies in the case of causation just as much as it applies elsewhere. Science investigates causes and effects, and it offers causal explanations. But not every causal investigation is a form of science and not every causal explanation is a kind of scientific explanation. In particular, the investigation of causes and the giving of causal explanations that’s involved in common-sense psychology isn’t a form of science. As I said at the beginning, it’s a mistake to construe common-sense psychology as a primitive science of the internal causes of physical behaviour.

So there is a kind of scientifically-influenced, causal approach in the philosophy of mind that I do think is definitely wrong. A good case is the sort of view that holds that the mind simply is the brain and that the best way of understanding mental phenomena is just to study the workings of the brain. Of course understanding the brain is hugely interesting and important. But it’s a different enterprise, with a different subject-matter, from the enterprise of understanding ourselves and others in terms of the ordinary mental concepts we use in everyday psychological description and explanation.

You asked whether scientific approaches to action and perception that focus exclusively on causality are in some respects in error. My answer is ‘yes’. But as always in philosophy, the nuances are important. The key words here are ‘scientific’ and ‘exclusively’. I say that explaining actions and attitudes in terms of a person’s reasons is a form of causal explanation; when we think about actions we are investigating causality. But our investigation isn’t scientific: we aren’t looking for laws or causal mechanisms; nor are we trying to vindicate a causal story about action by marrying up its elements with those of a lower-level causal story. And our investigation isn’t exclusively causal; we’re looking for an explanation that makes rational sense of a person’s action, not just for a set of causes that produced an effect. That point is often put by drawing a contrast between the causal and the merely causal. Psychological explanation is causal explanation; but it’s not merely causal explanation.

3:AM:  And as a take home, does Wittgenstein help us with the so-called ‘hard problem’ of consciousness?

BC: That may depend on the kind of help we’re hoping for! The so-called ‘hard problem’ of consciousness is the problem of how it is that physical phenomena in the brain and nervous system produce conscious experience. There’s a passage in Philosophical Investigations where Wittgenstein explicitly articulates that problem: or rather, he articulates the kind of thinking we engage in when, as he might put it, we’re gripped by sense that there is a hard problem of consciousness. Here’s some of what he says:

The feeling of an unbridgeable gulf between consciousness and brain process: how come that this plays no role in reflections of ordinary life? This idea of a difference in kind is accompanied by slight giddiness – which occurs when we are doing logical tricks . . . When does this feeling occur in the present case? It is when I, for example, turn my attention in a particular way on to my own consciousness and, astonished, say to myself: ‘THIS is supposed to be produced by a process in the brain!’ – as it were clutching my forehead. – But what can it mean to speak of ‘turning my attention on to my own consciousness’? There is surely nothing more extraordinary than that there should be any such thing! . . .
Note that the sentence which I uttered as a paradox (‘THIS is produced by a brain process!’) has nothing paradoxical about it. I could have said it in the course of an experiment whose purpose was to show that an effect of light which I see is produced by stimulation of a particular part of the brain. . .. (PI §412).

That’s a really interesting passage. There’s a lot more going on in it than I can talk about here. And it’s obviously linked to a couple of the issues we were talking about earlier: the question of phenomenal concepts; and the private language sections. But here are what I take to be some of Wittgenstein’s key points in the passage.

First of all, Wittgenstein thinks that the sense that there is an unanswerable problem about how physical phenomena produce consciousness – an ‘unbridgeable gulf’, an unfillable explanatory gap – is an illusion. The question, ‘How is THIS produced by a brain process?’, asked in a way that expects the answer ‘It can’t be’ or It’s in principle impossible to explain how’ or ‘It’s deeply mysterious how it can be’ is a bad question. The feeling that there is something about conscious experience that is mysterious, or queer, or impossible to understand is an artefact of a bad, distinctively philosophical, way of conceiving things.

On the other hand, Wittgenstein thinks, the claim ‘This experience is produced by that brain process’, when it is made in the kind of straightforward context he imagines, does not seem mysterious at all and doesn’t have the appearance of leaving an explanatory gap between the brain process and the experience. It is interesting to consider how far Wittgenstein thinks we could legitimately press the kind of investigation he describes. I think we could take it a long way without lapsing into what he would regard as illusion. He sees no problem in the idea that we could establish experimentally that a particular ‘effect of light that I see’ is ‘produced by stimulation of a particular part of the brain’. Suppose we ask why that particular kind of experience of light, rather than some other, is produced by stimulation of that part of the brain. There is no reason why Wittgenstein should object to that question, posed as part of the same empirical investigation, or to the kinds of answers that such an investigation might produce. The important constraint, for him, is that the way in which the investigation classifies kinds of experience must keep its roots in our ordinary ways of individuating experience; otherwise, we don’t know what we are asking about. But as long as that constraint is respected, there is no reason why the empirical enquiry he describes has to terminate at the first level, with the claim that this particular kind of experience is in fact produced by stimulation of that particular part of the brain. Of course we’ll reach a point in the enquiry where we can give no further answer: a question of the form ‘How, or why, does this produce that?’ where the answer is ‘It just does’. But that’s true in any enquiry. It’s not unique to questions about the relation between experience and the brain. And it doesn’t mean that that relation involves a unique or distinctive explanatory gap that doesn’t arise elsewhere.

So what exactly does Wittgenstein think is wrong with the question, ‘How can THIS possibly be produced by a brain process?’, asked in the kind of philosophical spirit he describes? I think there are at least two relevant elements in his thinking – neither of which is fully articulated in the passage I quoted, but both of which are behind the scenes. First, there is the idea that, when we focus our attention inward and say ‘THIS’, aiming to point not at some particular experience but rather at the fact of consciousness in general, we really don’t succeed in identifying anything at all. The argument here is related to what Wittgenstein says in the private language sections and in the Blue Book about identifying experiences. Second, there is the idea that the question builds in a philosophical picture of experience that guarantees from the outset that the phenomena will seem to be impossible. But here as elsewhere, Wittgenstein thinks, the right reaction is not to accept the picture and conclude that the phenomena are deeply mysterious, but to give up the picture that’s responsible for the appearance of mystery. In the present case (the ‘hard problem’ of consciousness) we’d obviously need to say much more than I have done in order to articulate and defend Wittgenstein’s case for both those ideas. But Wittgenstein’s attitude towards the ‘hard problem’ is clear. He thinks that the sense that there is an unbridgeable gulf between brain and conscious experience is an illusion. Philosophy should not concern itself with bridging the gulf or bemoan its unbridgeability. It should clear away the way of thinking that makes there seem to be a such a gulf in the first place.

3:AM: And finally, are there five books you could recommend to the readers here at 3:AM that would take us further into your philosophical world?


Donald Davidson, The Essential Davidson, Oxford: Oxford University Press, 2006. Davidson’s writings were an important part of my philosophical education and have been the backdrop to much of my own work. This volume contains some of his most important and influential essays.

Jennifer Hornsby, Simple Mindedness, Cambridge MA.: Harvard University Press, 1996. This collection of essays articulates what Hornsby calls ‘naïve naturalism’ in the philosophy of mind. Her view of the mental seems to me to be basically correct; I’ve learnt a lot from her work.

Iris Murdoch, The Sovereignty of Good, London: Routledge, 1970. I’ve included this slim volume particularly for its first essay, ‘The Idea of Perfection’, which applies to philosophy in general, and moral philosophy in particular, lessons that Murdoch draws from Wittgenstein and his discussion of private language. It helped me to see the significance of Wittgenstein’s later work in a much broader context than is initially evident from his focus in Philosophical Investigations on issues of language and meaning.

David Pears, The False Prison vols 1 and 2, Oxford: Oxford University Press, 1987, 1his. (I’m cheating and counting this as one book.) Pears was one of my teachers, and the first person from whom I learned about Wittgenstein. I had tutorials on the Tractatus with him when I was an undergraduate; later on, he was one of my doctoral thesis supervisors. I’ve read this book many times and it’s always a pleasure to return to it, not least because Pears’s voice and personality come through so clearly in his writing.

Ludwig Wittgenstein, Remarks on the Philosophy of Psychology vols 1 and 2, Oxford: Blackwell, 1980. (Another cheat.) I’m not a disciple of Wittgenstein. But I always enjoy reading his work. His discussions of mental phenomena in these volumes are immensely insightful and phenomenologically acute. And he considers some unusual and interesting cases – including the phenomenon of remembering what one was about to do, which we talked about earlier.


Richard Marshall is still biding his time.

Buy his new book here or his first book here to keep him biding!

End Times Series: the first 302


First published in 3:AM Magazine: Saturday, April 14th, 2018.