Wittgenstein: Meaning and Mind (Volume 3 of an Analytical Commentary on the Philosophical Investigations), Part 1: Essays

By P. M. S. Hacker

Wittgenstein: Meaning and Mind is the third volume of a four-volume analytical commentary on Wittgenstein’s Philosophical Investigations. It consists of two parts. Part 1 is a sequence of fifteen essays that examine in detailall the major topics discussed in Philosophical Investigations §§243-427. These include the private language arguments, privacy, private ostensive definition, the nature of the mind, the inner and the outer, behaviour and behaviourism, thought, imagination, the self, consciousness, and criteria.

The first edition of this volume of essays was published in 1990 to widespread acclaim as a scholarly tour de force, providing a comprehensive survey of these themes, the history of their treatment in early modern and modern philosophy, the development of Wittgenstein’s ideas on these subjects from 1929 onwards, and an elaborate analysis of his definitive arguments in the Investigations.

The new second edition has been thoroughly revised by the author and features four new essays. These include a survey of the evolution of the private language arguments in Wittgenstein’s oeuvre and their role within the developing argument of the Investigations, a comprehensive essay on private ownership of experience and its pitfalls, a detailed examination and defense of Wittgenstein’s repudiation of subjective knowledge of one’s experience, and an overview of the achievement and importance of the private language arguments. New objections to Wittgenstein’s arguments are examined—and found wanting—and new materials from the Nachlass that were not known to exist in 1990 have been incorporated into the text of these essays. All references have been adjusted to the revised fourth edition of the Investigations, but previous pagination in the first and second editions has been retained in parentheses.

These revisions bring the book up to the high standard of the extensively revised editions of Wittgenstein: Understanding and Meaning (2005) and Wittgenstein: Rules, Grammar and Necessity (2009). They ensure that this survey of Wittgenstein's private language arguments and of his accounts of thought, imagination, consciousness, the self, and criteria will remain the essential reference work on the Investigations for the foreseeable future.

• Language: English
• Publisher: Wiley
• Release date: Apr 5, 2019
• ISBN: 9781118951828

Book preview

I

Introduction to the private language arguments

1. The Augustinian conception of language and Wittgenstein’s early commitments

§§243 – 315 of the Philosophical Investigations are commonly referred to as ‘the private language argument’. The name is not Wittgenstein’s — indeed he never used this phrase. But he did talk of a ‘private language’ both in the Investigations (§256, §269, §275) and in its immediate manuscript antecedents (MS 180a, 10v–11r; MS 124, 225 f.; MS 120, 42), and he referred to ‘the discussion of a private language’ (MS 165, 101). He wrote of the putative ‘privacy’ of representations or Vorstellungen (MS 109, 296; MS 110, 300; PI §251); of the privacy of sense‐data (MS 110, 7; BT 224; LPE 221, 237; LSD 290), of experiences or ‘inner experiences’ (LPE pp. 208, 234 f.; PI §243, §272), of visual images (LPE 214), of impressions (LPE 220) and colour‐impressions (PI §§272 – 9), and of sensations (PI §246). To use the singular definite description ‘the private language argument’ to describe the long and complex discussion in Investigations §§243 – 315 is misleading. For there is not one continuous argument in this sequence of remarks, but many discontinuous arguments. If we are to employ a definite description to refer to this part of the book, it would be more appropriate to speak of the private language arguments. Many closely interwoven themes are investigated, and a wide variety of grammatical clarifications emerge. They are all connected more or less directly with the prima facie curious idea of a language which cannot, logically cannot, be understood by anyone other than its speaker. On first encounter, this is bewildering, since it is not obvious why the idea of such a language should be of any interest to philosophers or indeed to anyone else. After all, our languages, the natural languages of mankind, are spoken by many human beings, and can be learned by any human being.

Wittgenstein characterizes a ‘private language’ at the very beginning of the discussion as ‘a language in which a person could write down or give voice to his inner experiences — his feelings, moods, and so on — for his own use’. The words of this language ‘are to refer to what only the speaker can know — to his immediate private sensations’ (PI §243). In earlier discussions, Wittgenstein speaks of ‘the language in which someone speaks to himself about his own private experiences’ (MS 165, 101), and later in the Investigations, he focuses upon colour‐impressions (PI §§272 – 8). So his net is intended to catch not only pain and other sensations such as tickles and itches, but ‘inner’ experiences in general, such as sense‐impressions (including colour‐impressions), moods and emotions. Of course, we can, in our natural languages, speak of our pains and other sensations, and of our experiences in general. That by itself does not render a language private in Wittgenstein’s special sense. In addition, the words of the language, or at any rate of the descriptive vocabulary of the language, are supposed to refer to what can be known only to the speaker of such a language.¹ This condition is satisfied only if two further commitments are in place. First, that experience is uniquely and non‐transferably owned by its subject. Only I can have my pains. No one else can feel my pain. Different people cannot have the very same pain, only a very similar one. Second, that precisely because no one else can have my pain, no one else can know whether I have a pain or, indeed, know what precisely I have. But even this does not suffice, for one might suggest that words may indeed refer to such private and unshareable inner ‘objects’, as long as others can understand what they mean (i.e. grasp their sense) even if they cannot experience their referent. So one must add a further, final, condition, namely that what the words of such a language mean is determined for oneself alone by a mental, private, ostensive definition which attaches a name of any given private experience to a private sample that only the subject has and only the subject can know (be acquainted with). Given these specifications, then such a language would indeed be a ‘private’ one, not in a contingent sense in which, say, Esperanto was private until its inventor Zamenhof started to teach it to others, but in a logical sense. For Wittgenstein’s supposed private language is one which it is logically impossible to teach another and similarly impossible for anyone else to understand.

Having thus pinned down the quarry, one may well still be baffled. Why is Wittgenstein interested in the bizarre idea of the possibility of a language that no one other than its speaker can understand? To this there are four answers. First, he associated this array of misunderstandings with the Augustinian conception of language and linguistic meaning with which the Investigations began. So finally to extirpate this pervasive conception requires a thorough examination and pruning of these poisonous offshoots. Second, he had once committed himself not merely to the possibility but to the actuality of such languages and to the analysability of all language into the ‘primary language’ of sense‐data. So, in exploring the idea, he was examining where and why he had himself gone wrong. Third, the majority of early modern and modern philosophers had unwittingly committed themselves to such ideas in so far as they conceived of ideas or impressions as what is ‘given’ and of all else as constructed or problematically inferred. Although Wittgenstein’s knowledge of the history of his subject was minimal, he certainly knew that much; in particular, he was well acquainted with the writings of Frege, Russell and Moore that succumb in one way or another to these misconceived ideas. Fourth, the existence, or even the possibility, of a private language thus envisaged would challenge the conclusions of his lengthy discussion of rules and following rules, namely that following a rule presupposes a public practice manifest in regularities of behaviour, recognition of a regularity and a normative attitude towards the recognized regularity exhibited in normative activities of checking, correcting mistakes, and so forth. Rules need not be shared, but they must be sharable (see Volume 2, ‘Following rules, mastery of techniques, and practices’ and ‘Private linguists and private linguists — Robinson Crusoe sails again’).

To these four answers, we can, from a twenty‐first‐century position, add a fifth important reason for taking great pains in studying the strange idea of a logically private language. Just as Descartes and Locke, Berkeley and Hume, Bentham and the Mills, Frege and Russell, the phenomenologists in the wake of Brentano and Husserl, and the phenomenalists of the Vienna Circle all constructed their systems of philosophy, logic and metaphysics on an array of presuppositions that unwittingly committed them to the intelligibility of a private language, so too have contemporary sciences of mind and brain — theoretical linguistics, cognitive neuroscience, empirical psychology and psycholinguistics. If Wittgenstein’s arguments are correct, as I shall endeavour to show them to be, then it is important to grapple with them. For they have a significant bearing on these empirical sciences, facilitating the elimination of conceptual confusion and providing them with clear conceptual frameworks within which to pursue empirical investigations of human powers and their exercise.²

That Wittgenstein linked the idea of a private language with the Augustinian conception of meaning is evident both from remarks in the Investigations and from Wittgenstein’s notebooks. In the quotation from the Confessions with which Wittgenstein opens his book, Augustine writes: ‘When grown‐ups named some object and at the same time turned towards it, I perceived this, and I grasped that the thing was signified by the sound they uttered, since they meant to point it out.’ In the same context Augustine elaborated further on how, as a child, he had tried to express what he meant, and how he taught himself to speak (see Exg. §1). Of this Wittgenstein remarks that

Augustine describes the learning of human language as if the child came into a foreign country and did not understand the language of the country; that is, as if he already had a language, only not this one. Or again, as if the child could already think, only not yet speak. And think would here mean something like talk to himself.

(PI §32)

Here, it is reasonable to suppose, the child is envisaged as having a private language in which he names his desires and sense‐impressions, which he must do if, as Augustine says, he tries to convey to adults what he means by his cries and gestures. In MS 140, 7 Wittgenstein linked ‘Augustine about expressing the wishes inside him’ with the sentence ‘as soon as ever he had learnt enough of their language, the stranger informed his hosts of…’

Furthermore, the child is supposed to be able to guess what the adults signify by the words they use, e.g. whether they are pointing at the colour of a thing when they say a word, or at its shape, or number, and so forth (PI §33). So he must already possess the concepts of colour, shape, number, pointing, word, naming and meaning. In MS 111, 15 f. Wittgenstein observes that Augustine’s conception makes it appear as if naming is the foundation and the be‐all and end‐all of language (‘Hier scheint also das Bennenen Fundament und Um‐und‐Auf der Sprache zu sein’). This, Wittgenstein continues, is equivalent to the conception according to which the form of explanation ‘This is’ is conceived to be fundamental. This is made perfectly explicit in PI §§256 – 8, in which the imagined speaker of a private language is said to ‘simply associate names with sensations, and use these names in descriptions’ and, being a genius (let us suppose), he ‘invents a name for the sensation by himself’ and gives himself ‘a kind of ostensive definition’.

It is also important to bear in mind that one of Wittgenstein’s targets in his discussions of our psychological vocabulary is the idea that it consists of names of psychological objects or experiences. Three issues preoccupied him:

First, what does ‘giving a name to a sensation’ mean? If giving a thing a name is akin to hanging a label or nameplate on it (PI §15), how can one label a pain? One can name a child in a baptismal ceremony or at a registry office; one can name a ship at a launching ceremony; but how does one name a sensation? What analogy is there between the various forms and contexts in which we name things and the putative naming of sensations? — Hence the exploration and repudiation of the idea of naming sensations by a kind of private ostensive definition which seems analogous to a public ostensive definition.

Second, although, as Wittgenstein has argued, one can say that every word in the language signifies something (PI §13), and hence that words for sensations or more generally for experiences signify something, the price of this concession is vacuity. There is no single ‘name‐relation’ and there are as many logically different kinds of name as there are different kinds of nominata. To say that ‘pain’ is the name of a sensation, if it amounts to anything, is to say that ‘I have a pain’ is an expression of sensation (RPPI §313) — which cannot be said for ‘Her name is Jill’, ‘This colour is called magenta’, or ‘This truth–functional connective is called the Sheffer stroke’, and so on.

Third, the model of object and name — Jack and ‘Jack’, a chair and ‘chair’, perhaps even (when stretched) red and ‘red’ — is wholly misleading when it comes to psychological expressions (PI §293). For even if we innocuously call the nouns ‘pain’, ‘anger’, ‘understanding’, ‘knowledge’, ‘belief’, ‘intention’, ‘hope’, ‘expectation’ names of psychological attributes, they are not names of objects of any kind. Nor does replacing the expression ‘object’ by other categorials, such as ‘state’, ‘activity’, ‘process’, ‘act’, save the model for the purpose of analysis of psychological propositions. These categorials, far from being the hardest of the hard, are the softest of the soft. Knowledge and belief are not states, understanding and thinking are not processes or activities, intending and meaning something are not acts. In the case of many psychological concepts, such as knowing, believing, intending, meaning something, there simply are no phenomena to capture by means of a name. In the case of others, such as fear, expectation and hope, there are sensible phenomena all right (feeling of the accelerating heartbeat, tension, shortness of breath, feeling of dryness in the mouth, etc.), but these phenomena are not the fear, expectation or hope, and are not what the expressions signify, let alone name. The reasons are manifold — the most important of which is that whatever sensible phenomena there are, they lack intentionality. In short, the analysis of psychological expressions is not aided, but obscured, by conceiving of them as names.

This suffices to show that one reason for Wittgenstein’s interest in the idea of a private language is that it extends his criticisms of the Augustinian, referential, conception of language and linguistic meaning right to the heart of philosophy of psychology.

Had Wittgenstein himself envisaged the language each of us speaks as a private language in the requisite sense? The question is of interest. The evidence here is murky, and has to rely on the Notebooks, Wittgenstein’s letters, the Tractatus, the 1929 lecture ‘Some Remarks on Logical Form’ and the post‐1929 notebooks in which he criticized what he had previously thought. The young Wittgenstein’s ideas on two themes need to be collated and interpreted.

First, what he assumed simple objects were and how he thought the names of simple objects are linked to the objects of which they are names. This is notoriously controversial and need not be discussed here. I have elaborated elsewhere why I believe that among the simple objects of the Tractatus are spatio‐temporal points in the visual field and minima visibilia in the domain of colour (space and time being forms of objects (TLP 2.0251) and being coloured a form of visual objects (PTLP 2.0252). Whether these ‘objects’ are to be understood in Platonist terms (beyond existence and inexistence) or in phenomenal terms or both depends upon Wittgenstein’s remarks on the second theme that bears on our concern.

These are the opaque remarks on solipsism and the self on 23.5.1915 and between 11.6.1916 and 19.11.1916 and their correlates in the Tractatus 5.6 – 5.641, as well as the observations on death in 6.4 – 6.4312. These exercise a certain perverse fascination. They are exceedingly obscure and their interpretation is as contentious as their obscurity. They resonate powerfully:

The World and Life are one

Physiological life is of course not Life. And neither is psychological life. Life is the World.

(NB 24.7.1916)

The thinking subject is surely mere illusion. But the willing subject exists. [Cf. TLP 5.631]

If the will did not exist, neither would there be that centre of the world, which we call the I, and which is the bearer of ethics.

What is good and evil is essentially the I, not the world.

The I, the I is what is deeply mysterious. (5.8.16)

The I makes its appearance in philosophy through the world’s being my world. (12.8.16; cf. TLP 5.641)

What has history to do with me. Mine is the first and only world! (2.9.16)

I am my world. (12.10.16; cf. TLP 5.63)

This is heady stuff,³ It is hardly surprising that he wrote ‘I am conscious of the complete unclarity of all these sentences’ (2.8.16). It is doubtful whether we shall ever know for sure what Wittgenstein had in mind and what such obscure remarks committed him to. Certainly, spatio‐temporal points and minima sensibilia are good candidates for ‘objects’ that are beyond existence and inexistence, thus constituting the substance of the world. Minima sensibilia are also good candidates for phenomenal objects constituting the world as idea. Again, I doubt whether we shall ever know for sure what he meant. But we do know for sure that what he meant makes no sense, since he said so again and again in the 1930s. The most that one can hope to do is not to make sense of what he thought, but to make sense of his thinking what he thought.

There is, I believe, a good case for characterizing his Tractatus views on solipsism as combining empirical realism with transcendental solipsism, while emphasizing that this use of ‘transcendental’, like Wittgenstein’s own (‘logic is transcendental’ TLP 6.13), involves no commitment to Kantian transcendental idealism. There is, for example, no suggestion of any distinction between things as they are in themselves and things as they appear to us, nor is there any claim that space and time are imposed on appearances by our cognitive constitution, and the claim that logic is transcendental does not imply that logic produces the form of the world, but rather that it mirrors it.⁴ One could as well characterize him as adhering to a form of ‘ineffable solipsism and describable realism’. For ‘what the solipsist means [‘meint’] is quite correct; only it cannot be said, but makes itself manifest [‘es zeigt sich’] (TLP 5.62). This remark is elaborated in TLP 5.621: ‘The world and life are one’. Life, in this context, is consciousness itself (see Essay XIV, ‘The World of Consciousness’, section l).⁵ This in turn is followed by the assertion ‘I am my world. (The microcosm.)’ (TLP 5.63), the numerical successor to which is 5.64: ‘Here it can be seen that solipsism, when its implications are followed out strictly, coincides with pure realism. The self of solipsism shrinks to a point without extension, and there remains the reality coordinated with it.’ How does solipsism make itself manifest? Presumably in the form of every empirical proposition or every experiential proposition after analysis (perhaps indexed to the speaker) — a supposition given some support by the later suggestive but obscure remark

((I see, indistinctly, a connection between the problem of solipsism or idealism and the notational system of a proposition. In these cases is the I perhaps replaced by the proposition, and the relation of the I to reality by the relation between the proposition and reality?))

(BT 499)

Be all this as it may, in 1929 – 30, in his notebooks of this period, in Philosophical Remarks, in his lectures (see Desmond Lee’s and G. E. Moore’s lecture notes) and in conversations with, and dictations to, Waismann (WWK & VW), Wittgenstein espoused an account of language and linguistic meaning that was committed not only to the possibility of a private language, but to its actuality. He distinguished between what is ‘primary’ and what is ‘secondary’. To the ‘primary’ belongs immediate experience (expressed, for example, by ‘I have …’, and ‘It perceptually seems to me …’). Immediate experience is described by genuine propositions that are directly and conclusively verifiable by reference to the experience they describe. So ‘I have a pain’ expresses a genuine proposition. ‘He has a pain’, by contrast is a hypothesis which is supported by behavioural evidence. Unlike genuine propositions, hypotheses are never conclusively verifiable. The word ‘toothache’ in my utterance ‘I have toothache’ has a different meaning from the word ‘toothache’ in my utterance ‘He has toothache’. ‘Toothache’ in my assertion ‘I have toothache’ signifies a primary experience or sense‐datum, whereas in ‘He has toothache’ it signifies a behavioural pattern. But if names of primary experience signify what I have when I have an experience, then it makes no sense to say that others have experiences, i.e. have what I have when I have experiences. Moreover,

In the sense of the phrase ‘sense‐data’ in which it is inconceivable that someone else should have them, it cannot, for this very reason, be said that someone else does not have them. And by the same token, it’s senseless to say that I, as opposed to someone else, have them.

(PR §61)

So he was committed to what later became known as a ‘no‐ownership’ analysis of experience. All this becomes much clearer in an alternative form of representation that more perspicuously represents the facts. Imagine a language in which there is a Centre. When the Centre has a pain it is said ‘There is a pain’. When others have a pain it is said ‘Others are behaving as the Centre behaves when there is a pain’. Solipsism stops short of this and says ‘I have pain’ and ‘Others are behaving as I behave when I have a pain’. But (as Hume had already noticed) in my experience of pain I experience no owner (PR §65). So ‘There is a pain’ is a more perspicuous representation here than ‘I have a pain’. Imagine that each of us speaks such a language, of which we ourselves are the Centre:

Now, among all the languages with different people as their centres, each of which I can understand, the one with me as its centre has a privileged status. This language is particularly adequate. How am I to express that? That is, how can I rightly represent its special advantage in words? This can’t be done. For if I do it in the language with me as its centre, then the exceptional status of the description of this language in its own terms is nothing very remarkable, and in terms of another language my language occupies no privileged status whatever. — The privileged status lies in the application, and if I describe this application, the privileged status again doesn’t find expression, since the description depends on the language in which it is couched.

(PR §58)

It is evident that Wittgenstein has moved beyond both empirical solipsism and the apparent transcendental solipsism of the Tractatus to a form of methodological solipsism (not unlike Carnap’s in his Logischer Aufbau der Welt).⁶ Wittgenstein has realized the illusions of private ownership of experience: that if you can’t have my pain, then I can’t have it either; that the subject is not a defining characteristic of a pain; that one cannot explain what it is for another person to have a toothache by saying that he has what I have when I have a toothache (PR §62). But he continues to think, as he had thought when writing the Tractatus, that grammar pays homage to reality — that a perspicuous notation or form of representation will faithfully reflect the objective structure of reality. He has not yet located his grammatical insights in the correct place. He has not yet arrived at his later characterization of avowals or ‘utterances’ of experience or of behavioural criteria, which in due course replace the notions of ‘genuine propositions’ and ‘hypotheses’. Nor has he yet realized the conceptually complementary relation between the groundless first‐person utterance and the third‐person ascription of experience.

Wittgenstein’s struggles with the analysis of experiential nouns and predicates continued in The Big Typescript, in the Blue Book and then in the ‘Notes for Lectures on Private Experience and Sense Data’ in 1934 – 6 (MSS 148, 149, 151) which vividly present his efforts to achieve a correct logical point of view. The skeleton of the later arguments was drafted in 1937 – 9 (MSS 119 – 121, 158, 160, 162(b)). The matter was pursued further in his ‘Notes for the Philosophical Lecture’, probably written for a British Academy lecture scheduled for 1942, but never delivered. Things come to a head in 1945 with MSS 165, 179, 180a, 124 and 129, and the later parts of MS 116, from which the discussion in the Intermediate Draft (1944/5) of the Investigations was derived.

This makes it clear that in tackling the themes of §§243 – 315, Wittgenstein was, among other things, correcting flaws in his earlier thoughts on the topic of subjective experience, avowals of experience, third‐person ascriptions of experience and the logico‐grammatical character of psychological predicates. The private language arguments in the Investigations are the culmination of recurrent reflections over sixteen years on these great philosophical themes.

2. The place of the private language arguments in the Philosophical Investigations

How does the discussion of the private language from §243 to §315 fit in with the preceding remarks in the book (§§139 – 242)? What part does it play in the overall grand‐strategy? In the opening essay of Volume 2 ‘Two fruits on one tree’, we discussed the fact that the ‘Frühfassung’ (early draft) of the Investigations, composed in 1937/8, contained an early form of Investigations §§1 – 189, followed by an early version of Remarks on the Foundations of Mathematics, Part I. So the discussion of following a rule, which was introduced in order to reconcile the two grammatical observations that the meaning of a word is its use (which is ‘spread out over time’⁷) and that the meaning of a word is what is understood when one knows what it means, and in particular that what a word means can be grasped at a stroke. How can something that unfolds over time be grasped in an instant? Clearly, the meaning of a word is what is given by an explanation of meaning, and an explanation of meaning is a rule for the use of a word. But how can a rule as it were contain its sequential applications in advance of being applied? For example, how can the rule for the series of even integers: ‘Add 2!’ or ‘+2!’ or ‘2, 4, 6, 8, and so on!’ predetermine in advance that the number following 1000 in this series is 1002? This long discussion was originally designed as a prolegomenon to the investigation of the nature of mathematical and logical necessity in Remarks on the Foundations of Mathematics, Part I. In 1944/5, Wittgenstein decided to drop the mathematical sequel of the Frühfassung and replaced it in the Zwischenfassung (intermediate draft) with the discussion of a private language. How it is possible for two such different crowns to grow from the selfsame trunk is explained in ‘Two fruits from one tree’. But it is striking that Wittgenstein did not replace the mathematical example of following the rule of a series (which was apt for the subsequent discussion of mathematical necessity) with a non‐mathematical example of following a rule, e. g. ‘This ◻ is white’, which gives the meaning of a word. A normative variant of Goodman’s new riddle of induction might have been apt. This would arguably have made it much easier to see the continuity between Investigations §139 to §242 and the subsequent discussion of a private language.

However, Wittgenstein did not do this, and left nothing in his Nachlass or in records of conversations with friends to indicate why he didn’t. Nevertheless, the connecting links between the discussion of following a rule in the Frühfassung and in the Zwischenfassung and the private language arguments are forged in what is now §§198 – 242. This is evident in

§202: following a rule is a practice, and to think one is following a rule is not to follow a rule. That is why it is not possible to follow a rule ‘privately’, for then thinking one was following a rule would be the same as following it.

§206: shared human behaviour (Handlungsweise) is the system of reference by means of which we interpret an unknown language.

§207: for there to be a language, there must be regular connections between utterances (sounds emitted) and activities.

§208: there must be a recognized uniformity that determines, in recurrent regularities, what is to count as doing the same in a sequence of actions.

there must be a regularity displayed in a practice, for there cannot be a rule that is followed only once.⁸

the recognition of uniformity must be exhibited in normative activities of teaching, explaining, agreement, rejection, expectation, and encouragement.

The immediate connection is evident in §242, which avers that there must be agreement not only in definitions (explanations of meaning), but also in judgements, just as the existence of a system of measurement requires not only an agreement on the measure (the rule) but also agreement in the results of measurement (on its application). §243 then raises the question of a ‘private’ language in which there could, in principle, be no agreement in measurement, since each person’s measure is known only to himself. One must admit that the connections are anything but immediately perspicuous.

If the smoothness of the connection between the discussion of following a rule, which runs with occasional digressions from §143 to §242, and its sequel leaves something to be desired, so too does the unfolding of the private language arguments in §§243 – 315. Although every sentence is clear enough, the difficulty of the whole is no less than that of Kant’s Transcendental Deduction of the Categories. Many different threads are interwoven: epistemic privacy of experience; private ownership of experience; pretence and the impossibility of animal or neonate pretence; the logical status of such propositions as ‘mental images are private’ or ‘every rod has a length’; private ostensive definition; private samples; criteria of identity for sensations; the commonly supposed privacy of colour impressions and the meaning of colour words; the relationship between pain and pain behaviour; criticisms of behaviourism; the grammatical principle that psychological predicates are predicable only of human beings and what behaves like human beings and the associated criticisms of the homunculus fallacy and the mereological fallacy; the relation between mind and body; ascription of experience to others (and repudiation of the argument from analogy); and criticisms of the misuse of the terms ‘object’, ‘state’, ‘process’ in philosophical discussions of the psychological. Some of these themes are allocated two or three remarks and not mentioned again; others are discussed for a few remarks and then abandoned, only to be resumed later, abandoned again and again picked up later. Some of these headings are nonsensical (like ‘square circles’), for example: ‘epistemic privacy of experience’, ‘private ownership of experience’, ‘private ostensive definition’, ‘private sample’, and the arguments subsumed under these headings are reductio ad absurdum arguments. All of this makes following Wittgenstein’s footsteps exceedingly difficult. The volume of exegesis that accompanies this book of essays is designed to help the reader through this wilderness and up these rock faces.

A very influential interpretation of the private language argument (in the singular) was offered by Saul Kripke,⁹ who argued that the private language argument is in fact complete by §202, in which Wittgenstein declares that it is not possible to follow a rule ‘privately’. In Volume 2, Essays IV and V and Exg. §202, it was argued in detail that this interpretation is altogether mistaken. It is neither true to the text of the Investigations nor to its numerous elaborations in the Nachlass that definitively show it to be wrong. I shall not go over that ground again here. But I shall briefly sketch the flaws in Kripke’s account in order to indicate why it is to be dismissed. It will be remembered that Kripke held Wittgenstein’s argument up to Investigations §202 to be advancing a novel form of scepticism about meaning, namely that no past or present fact about a speaker (including oneself) can determine what one meant (e.g. that the pupil expanding the series of even integers should go on to write ‘1002, 1004, 1006, …’ rather than ‘1004, 1008, 1012, …’). To this sceptical challenge, Kripke suggested a ‘Humean’ sceptical conclusion: although it is never true that someone means anything by a word, nevertheless, if the word is used in agreement with the way others in one’s linguistic community use it, then communication can take place. It should be evident that this is not an absurd form of scepticism about meaning, but rather an even more absurd form of semantic nihilism. For this issue, as presented by Kripke, turns not on ignorance (as sceptical questions do) but on there being no fact that determines meaning, and so nothing to either know or be ignorant of.

However, Wittgenstein never advanced any such thesis (for thesis it is). Nor did he embrace any form of semantic scepticism, let alone nihilism. Wittgenstein’s discussion of following rules was not meant to show that there is a warrant for speaking of someone as following a rule only in the context of a community of behaviourally convergent speakers. Rather, it was designed to show that it makes sense to talk of following a rule only in the context of a practice — a uniformity embedded in a regularity of behaviour, informed by normative activities such as using the formulation of a rule as a standard of correctness, rectifying mistakes by reference to the rule, justifying what one said or did by appeal to the rule. Such practices, with us, are typically shared, although they need not always be, and they are typically learned in a social context, though some may be invented in solitude for one’s own private (but not ‘private’) use. But, as Wittgenstein’s numerous discussions of Robinson Crusoe, solitary cavemen, etc. demonstrate, there is no conceptual absurdity in imagining a person following a rule and establishing a practice — a normative regularity — in an asocial context (see Volume 2, ‘Private linguists and Private linguists — Robinson Crusoe sails again’). That a language is learned from other speakers is an important empirical fact (denied by Chomsky and his innatist followers) about the genesis of a linguistic ability, but it does not enter into the grammatical (logical) characterization of the ability (cf. PG 188; BB 12, 97; PI §495), for an ability is characterized by what it is an ability to do, and the criteria for possession of an ability are what the agent does, given an opportunity and reason for acting. The criteria for speaking a language do not require production of a school or parental certificate. We could determine whether a solitary caveman or desert islander could speak or use signs quite independently of determining how he learned to do so (see Exg. §243).

Had the discussion up to Investigations §202 been intended to prove that following a rule, like trade and barter, is only conceivable in a social group, it could not have shown the impossibility of a ‘private language’ in which the words refer to my own private sensations that only I have and only I know. For it would have left open the possibility that a common language in a social group is a congruence of ‘private’ languages built on private ostensive definitions employing ‘ideas’ as samples — as Locke explicitly, and most other empiricists implicitly, had supposed.

Furthermore, if Wittgenstein had been arguing that language, linguistic meaning and meaning something by one’s words are essentially social, essentially dependent on convergent behaviour and agreement in behaviour, then it would immediately follow that a person in solitude (a Crusoe, a Chingachgook (the last Mohican) or a wolf‐child) could not talk about anything. There could be no language at all — never mind a language about private sensations — independently of community agreement in use. In fact, Wittgenstein’s concern in this strand in his web of arguments is not whether one person alone could talk of his experiences in an unshared language, but whether all of us, in our normal social setting, can be conceived to be following rules constituted by mere association of word and mental ‘object’, or by private ostensive definition using an ‘idea’ as a defining sample. Private ostensive definitions are not ostensive definitions which other people do not happen to know about, but putative definitions (rules) which cannot be communicated to other people. It is such rules that were presupposed as the foundations of common languages by the mainstream of philosophy from Hobbes and Descartes to the Vienna Circle and beyond. It is by showing that there can be no such rules that representative idealism (and contemporary cognitive representationalism), classical British idealism, phenomenalism, solipsism and methodological solipsism can be shown to be incoherent.

3. The Great Tradition and its long shadow

The global purpose of Wittgenstein’s discussion of private knowledge of experience, private ownership of experience and private ostensive definition (which might be called the private language argument in a narrow sense) is not to establish that language is essentially social. Of course, human languages are shared, and are learned in social contexts from parents, elders and siblings. That is an anthropological and psychological truth. The mere logical possibility of innate knowledge of a language, of a solitary speaker (Robinson Crusoe), or of a solitary language that in fact only the speaker knows (Chingachgook or Zamenhof) is of little moment for Wittgenstein’s purposes, even though he discussed the matter repeatedly in his notebooks. Rather, his global purpose is to reveal the incoherence of a comprehensive picture or group of related pictures of human nature, of the nature of the mind and of the relation between behaviour and the mental, of knowledge, self‐knowledge and knowledge of the experiences of other people, of language and its alleged foundations in ‘the given’. Despite Wittgenstein’s relative ignorance of the great tradition of Western philosophy, he was familiar with its deep presuppositions through his detailed knowledge of Frege, Moore and Russell, through his contacts with the Vienna Circle and his knowledge of the forms of phenomenalism that some of them favoured, and doubtless also through debates in the Cambridge Moral Sciences club.

What is distinctive about early modern philosophy was its unquestioned adoption of the concepts of ideas and impressions as the primary tools for explanation of the sources, nature and limits of knowledge and understanding, and as the foundations of language. Impressions and ideas were what is given in experience. According to representative idealists, they were results of the impact of the material world on our receptive sense organs. The material world, in and of itself, consisted of matter in motion, characterized by primary qualities alone. Sensible qualities arose through the impact of particles on our nerve endings, which generated ideas of colour, sound, smell, etc. in the mind. According to what Kant called ‘dogmatic idealism’ (exemplified by Berkeley and Hume), ideas and impressions were what is given in experience, but also constitute the world itself. In general, it was out of these ideal materials that the mind wove its subjective conception of the world. Ideas were not only the materials out of which our picture of the world is made, they were also held to be the constitutive materials of thought. Thinking consisted in combining and separating ideas by predicating or denying one or more idea of another idea or ideas. Every thought was held to be a complex, consisting of a subject and predicate. All ideas were either simple or complex. Words were held to be names of ideas in the mind. Names of complex ideas were definable in terms of names of simple ideas. Simple ideas could not be constructed by the mind, but were given in experience, and were indefinable. To know what the name of a simple idea means, it was both necessary and sufficient to know which simple idea it stood for. (Simple names linked language and reality by their nexus with simple ideas.) Sentences were essentially complex, consisting of combinations of ideas — represented as conjoined by assertion, and as separated by denial. The combination or separation of names in a sentence reflected the combination or separation of corresponding ideas in the mind. The concept of the mind was uniquely defined in terms of thought, and thought was stretched to incorporate all contents of consciousness. The concept of the body, according to Descartes and his followers, was defined in terms of extension, and was narrowed to encompass everything subject to the physical sciences and alleged to be ultimately explicable in exclusively geometrical terms. Biology was held to be reducible to mechanics, and animal behaviour was no more than matter in motion. Biological life was held to be purely mechanical and biological phenomena were declared to be completely explicable in terms of the laws of physics. But Life, ‘Life as consciousness’ as Wittgenstein put it in Notebooks 1914 –16, was unique, and beyond the domain of physics and its laws. Thus far the Galileian and Cartesian roots of modern thought, variously developed and refined in a complex dialectic by Malebranche, Spinoza and Leibniz, Locke, Hume and Reid, and, despite Kant’s onslaught, by Bentham and the Mills, as well as Frege and Russell.

The vicissitudes of this picture (or these pictures) of the world and of ourselves and our nature within it are not our concern here. What is our concern is that this picture advanced novel anti‐Aristotelian conceptions of both the mind and the body. What is of pivotal importance for the development of Western thought is that both the conception of the mind and the conception of the body were equally defective. Hence repudiating, for example, the conception of the mind as a thinking substance, and instead advancing the supposition that the mind is a bundle of ideas causally associated with a material substance, namely the body, or is a fiction, or is identical with behaviour and dispositions to behave, or is identical with the brain or with the supposed program of the brain, and so forth is futile. For the second half of the classical duality, namely the conceptions of the body, of bodily behaviour and of the relationship between the mental and behaviour, is retained, often in the form of an isomorphic dichotomy between brain and body, or the cortical and the behavioural. But current brain/body dualism is no advance over mind/body dualism if all that has changed is the replacement of aethereal non‐matter by grey glutinous matter while everything else remains the same.

It is obvious that philosophers who adopted such linguistic idealism (namely that the meanings of words are ideas in the mind) did not doubt that it was nevertheless possible to communicate by means of language. Although others cannot have the same sensations or experiences as I have, they can have sensations and experiences just like mine, even though not identical. Given the similarities of our nervous systems and cortex (given that we are all ‘wired up in the same way’), it is highly probable that different people associate qualitatively identical sensations or experiences with the same ideas. All communication is ‘telementation’: a supposition vividly depicted in de Saussure’s notorious ‘speech‐circuit’ diagram.¹⁰ That merely gave pictorial form to a supposition already patent in Hobbes: ‘The general use of speech, is to transfer our mental discourse, into verbal; or the train of our thoughts, into a train of words’ (Leviathan, Pt. 1, ch. 4) and in Locke: ‘The Comfort and Advantage of Society, not being to be had without Communication of Thoughts, it was necessary, that Man should find out some external sensible Signs, whereby those invisible Ideas, which his thoughts are made up of, might be made known to others …’ (Essay, Bk. III, ch. ii, sect. l).¹¹

It was part of Wittgenstein’s genius to put the question marks deeper down than anyone had ever done before. He did not examine the traditional assumption that despite ideas (experiences) being private and their names being explained by private ostensive definition, others can understand what is said in such a language. Rather, he challenged the overall picture at its seemingly unquestionable assumption, namely: could one understand such a language oneself ? Indeed, is there anything to understand at all?

What Wittgenstein aimed to show is not that sensation‐language, like the rest of language, is essentially shared, but that it is essentially shareable. On the received accounts in the dominant philosophical tradition, in current psychology, linguistic theory and cognitive neuroscience, this requirement is not met. But on the privacy assumptions, the speaker himself would not be able to understand what he says. For on those assumptions there are and can be no rules for the use of words, no ostensive definitions, no samples, no techniques of application, no distinction between correct and incorrect application of a rule, no distinction between following a rule and thinking one is following a rule. There is only a Schein‐praxis — an illusion of meaning. To show this is one of the primary roles of the private language arguments.

4. From grammatical trivialities to metaphysical mysteries

Each person has experiences, in a generous sense of the term. Human beings enjoy or endure sensations, have perceptual experiences (see, hear or feel things), feel emotions, are susceptible to moods and have appetites. They also think, believe, suppose and imagine things, know and understand things — none of which are experiences. There are mental states, such as being in a state of intense concentration, feeling cheerful or being depressed. There are mental activities and mental acts, but, as we shall see, they are not ‘just like’ physical acts and activities, only mental. The experiences that a person has are, tautologically, his experiences. Human beings commonly exhibit their experiences in their behaviour. Tautologically speaking, only I can exhibit my experiences. There is no such thing as one person manifesting the sufferings of another.

A person who has mastered a language can say what experiences he is having or has had. He does so independently of his observations of his own behaviour. What he says does not rest on the evidence of what he does. A person who is in pain cannot doubt whether he is; if he is feeling dizzy, he cannot doubt whether he feels dizzy; if he is thirsty, he cannot doubt that he wants to drink. Though self‐deception is sometimes possible, for example with respect to jealousy or envy, mistake is not. When a person says how things are with him, what sensations and perceptual experiences he is enjoying or undergoing, then, at least in paradigmatic cases, what he avows has privileged status: other things being equal, what he says goes. Truthfulness guarantees truth. There is such a thing as describing one’s states of mind — an activity at which Proust excelled. There is such a thing as introspection.

Judgements about other people’s sensations, perceptions, emotions, appetites, thoughts and beliefs, and so forth, rest on what they do and say in the circumstances of life. They may say or otherwise reveal how things are with them. Sometimes they may keep things to themselves, and sometimes they may conceal or try to conceal how things are with them.

These truisms are not empirical generalizations obvious to all, but for the most part grammatical propositions partly constitutive of the constituent concepts. However, it is but a short step from these grammatical platitudes to metaphysical theses, and thence to intellectual darkness. Philosophers, theoretical linguists and psychologists, hungry for theories about the nature of the mind, typically stray from such narrow and familiar paths into the minefields of metaphysics. Educated people, with no professional axes to grind, are readily induced to go down pathways indicated to them by psychologists, theoretical linguists and cognitive neuroscientists. For what could be more natural than to answer negatively such questions as ‘Can you have my pain?’, ‘Can you really know what I feel?’, and to answer affirmatively such questions as ‘When you are in pain, do you know that you are?’, ‘Are you conscious of your consciousness?’, ‘Are you conscious of your thoughts?’

The opening essay of Volume 1 of this Commentary, ‘The Augustinian conception of language’, section 2, entitled ‘The Augustinian family’, elaborated under five separate heads a wide variety of misconceived philosophical theories that explicitly or tacitly presupposed the Augustinian conception. What I shall now do is elaborate analogously the multiple pathways of error that lead off in different directions from the presuppositions of a private language. As Wittgenstein remarked:

Language has the same traps ready for everyone; the immense network of easily walked false paths. And so we see one person after another walking down the same paths, and we know already where he will make a turn, where he will keep going straight ahead without noticing the turn, etc., etc. Therefore, wherever false paths branch off, I should put up signs which help one get by the dangerous spots.

(BT 423)

The following are groups of pathways down which one should go only out of an interest in the geography of error.

(A) Misconceptions about the mental realm

Parallel to the public physical world, each of us enjoys access to a private realm of our own minds. The mental world consists of objects (such as pains, images, sense‐impressions), processes (such as thinking, imagining, remembering) and states (such as believing, understanding, knowing), which are, logically, just like physical objects, processes and states — only they are mental. They are doubtless mysterious; we talk about them, but with due caution leave their nature undecided — future investigations and theories will dispel our current ignorance (PI §308).

To have an experience, for example a sensation (such as pain) or an emotion (like fear), is to stand in a certain relation to such an object, process or state. ‘A has a pain’ and ‘A has a penny’ (as well as ‘I have a pain’ and ‘I have a penny’) have the same logical form, the latter member of each pair signifying a relation to a physical object, and the former to a mental object. The nature of the mental being undecided, these objects, processes, and states may be (a) sui generis — essentially distinct from and irreducible to the physical, being modes of conscious life with a unique qualitative feel to them (‘qualia’); or (b) neural, the experience of the subject consisting in modifications of the brain apprehended ‘from the inside’; or (c) functional objects, processes and states with a neural ‘realization’.

One person cannot have the identical experience as another, but only a similar one — qualitatively the same but numerically distinct. I cannot have your pain, but only one just like it. So experiences are inalienable ‘private property’.

(B) Misconceptions about names of the mental

One knows what the name of a (simple) mental item means if one knows what it stands for. One knows what the word ‘pain’ means if one knows that it stands for a certain sensation. One can know what this expression means only if one is acquainted with what it stands for, i.e. only if one has or has had a pain. For if the meaning of a word is what it stands for, then to know its meaning is to know or be acquainted with that thing. Alternatively, if the meaning of a word is not what it stands for, it is still plausible to argue that a word that stands for an experience which cannot be defined by analytic definition must be explained by an ostensive definition — and an ostensive explanation or definition of ‘pain’ or ‘see’ or ‘fear’ requires that one have or have had a pain, that one has had the experience of seeing, that one has been frightened. That knowing the meaning of such words presupposes acquaintance with what they signify seems confirmed by the thought that the blind do not really know what colour‐words mean because they