Is it Possible to Speak a Private Language?

Wittgenstein's Private Language Argument

In this essay we shall examine the notion of a private language and its implicit embodiment in several of the prevailing theories of the mind. This leads to an exploration of Wittgenstein’s fundamental scepticism concerning the privacy of our mental states and our privileged first person access to them. We shall show how this then denies the possibility of both a private language or of the meaningful use of public language in discussion of our inner mental states. This result has serious implications for mainstream models of mentality which must fundamentally change our view of human consciousness.

Firstly, what do we mean by a private language and how is this concept implicit in several important philosophies of the mind? The idea that our perception, recognition and understanding of our present mental states is intrinsically private to ourselves is a basic axiom of the Cartesian position. Mental states, says Descartes, are directly accessible and therefore knowable only to the person who has them. We have privileged first person access to our inner sensations by which we can have perfect, complete and immediate knowledge without any mediating interpretation or mapping. Descartes starting point is the question: ‘how can I know with certainty the things that I claim to know?’ Introspection upon the contents of his mind led to his famous cogito and its self-verifying nature – only a thing which thinks can question whether it does indeed think. This certainty of perception of my own inner state gives such a priori knowledge an authority which ensures its priority over any knowledge I might have of any other person or thing. Where my own mental states are concerned then what seems to be, is, and what is, seems to be. If I think I am in pain then I am in pain and if I am in pain then I must necessarily think I am in pain. This contrast between my certainty about my inner states and the uncertain knowledge of my corporeal sensations and perception of the world leads directly to Descartes’ conclusion that the mind must be essentially separate from the body and immaterial in essence. The Cartesian mind is a spiritual substance whose essence is thought and which is only contingently connected with the material world, whose essence is extension. When I have knowledge of my inner states I perceive them directly whereas my consciousness of external or bodily events is somehow mediated by the body in a way which must potentially introduce doubt and error.

Descartes hypothesises the logical possibility of an ‘evil genius’ which could in principle deceive him by falsifying the information he is able to gather of the world external to his mind. He retreats from the advances of the evil demon into the inner sanctum of his mind wherein he can engage in safe epistemological experiments in the sure and certain knowledge that his rational conclusions are valid. From this bridgehead he attempts to advance step by step into the uncertain domain of a posterori knowledge armed only with the shining sword of scepticism and the unshakeable belief in the essential incorruptibility of the soul.

To perceive by introspection that we are in some state implies that something is standing in relation to that state. Hence we are driven to the notion of mental experiences as objects which have inner existence and which enter into contingent relations with each other. The statement ‘I know I am in pain’ seems at first sight to talk about the perception of a definite object, my pain, as it becomes illuminated by the searchlight of my introspection. Furthermore this pain is indubitably my own and no-one else’s – only I can possess my pain and whereas objects in the outer world may be owned or public, my pain must by its nature be owned by myself alone.

For us to have such a privileged viewpoint of our inner mental states then we must be able to make statements to ourselves concerning the existence or otherwise of certain inner states and their logical relationship to each other. At the very least this requires some means of internally verifying that a certain inner sensation perceived at time t0 is the same as that perceived at some later time t1. But how can I establish this identity between two mental states without recourse to some symbolic representation by which I label each perceived state? Suppose whenever I am in mental state M then I shall label this with a symbol ð say, where ð is independent of the time of occurrence of M and yet represents for me the entire phenomenal character of M. Then, when I later experience some other state U say, I may compare its character with that of ð and if an identity relation holds it is meaningful for me to then assert that I am again in state M. In short I must evolve a phenomenal language with which I may make meaningful statements about my current mental states and their relation to my experience of past states. Since the objects of any predicates asserted in this language are, according to the Cartesian viewpoint, knowable only by myself, then only I can meaningfully use and understand this language – it is a private language.

Dualism then, with its notion of privacy and privileged access to mental states, rests upon the implicit practicality of such a private language. Since dualism in modified forms also underpins many other views of the mind which, although very different from the largely substance dualism of Descartes, this concept of the private language able to discourse about mental states as private objects is a fundamentally enabling axiom. To show as Wittgenstein did that such a private language is impossible would seriously damage all such first-person viewpoints of the mind. . Descartes point of view is irretrievably linked to the first person – the ‘I’ that questions the certainty of its knowledge and its own existence. If private language is impossible and mental states are not after all private objects, then we are driven towards the third person view of the mind and the philosophical anthropology espoused earlier by Kant in his criticisms of Descartes. To arrive at this conclusion and discuss its implications is the intention of the remainder of this essay.

Let us look first at the idea that mental states are in essence private to the individual mind that experiences them in the first person. Whilst I can ‘know’ that I am in pain, I can only ‘know’ that you are in pain through the imperfect interpretation of outward manifestations in your behaviour that I associate by analogy with my own experience as pain. It is the mediation of this inferential curtain between your inner pain and my perception (or lack of perception) of it that, amongst other things, divides your mind from my own. Let us examine then the nature of this association between a mental state and its indirect perception by a third party observer. How is it that I infer your inner mental state through observation of your behaviour and what does this tell us about the reality of the privacy assertion?

Firstly of course, mental states are only contingently connected to behaviour. A mental state, M say, may be manifested in a number of possible behaviours B1, B2, B3… or indeed in no external behaviour at all. Similarly I may exhibit one of the behaviours Bi without being in mental state M. My inference through observed behaviour then is subject to doubt, error, deceit, deviant behaviour, and all manner of non-causal associations between mentality and behaviour. As highly evolved social animals we often borrow sensation-associated behaviours to indicate totally different mental states – a linguistic use of behaviour akin to metaphor and allusion. I may be groaning and grimacing because I am indeed in the torment of inner pain, but perhaps I am using this behaviour to express my distaste for writer of the television program I am watching, or even that I wish you to believe me to be in pain in order that you then treat me as if I were in pain. This is similar to the analysis of spoken language by H.P. Grice where the distinction between sentence meaning and speaker’s meaning is discussed. Through behaviour you only have access to the sentence meaning of my mental states and must remain forever ignorant about my private speaker's meaning.

Perhaps this inference from behaviour to mental state has a deductive nature: if A then M, where A can be an arbitrarily complex disjunction of empirical observations about the individual and the world context, and M is an inner mental state such as ‘is in pain’. But this cannot be the case for we have already seen that the premise can be denied and yet we do not falsify the conclusion. If not deductive, then is our reasoning using some principle of induction where I infer from past observations and conclusions and assume that the present will behave like the past? But I cannot do this in any general sense for, as we have seen, I do not have perfect and direct knowledge of your mental states when a specific behaviour is manifested. To formulate a reliable inductive inference I must have a fund of cast-iron associations in my memory where I know with certainty that behaviour B ® mental state M. I can only do this for my own case and not for any third party. The use of induction here then would pre-suppose the very knowledge that we are trying to formulate an inductive link in order to ascertain – I can only use induction to perceive your present mental state if I first have direct and perfect knowledge of your mental states in the past.

Whilst I could perhaps derive inductive rules for my own mappings, it is only through a loose and insecure analogy that I reason about the possible mental states of a third party. This is similar to the ‘intentional stance’ of Daniel Dennett where I can analyse the behaviour of some object or system (human, animate or mechanical) by attributing to it beliefs, desires and intentions. I can understand the behaviour of a thermostat say by regarding it as if it had certain mental states, beliefs and intentions.

Alternatively, one could escape the problem of induction by following the path trodden by Karl Popper and consider human interaction to be a process of ‘inference to the best explanation’. I might put forward several hypotheses about your mental state and evaluate them against each other by devising various empirical test criteria. Popper would claim that I am justified in relying on the hypothesis which best survives this refutation process even though I can never be certain that it is a true explanation (statement of) your mental state. It is my best course of action, says Popper, to act according to this hypothesis as if it were true, rather than it being just the best of a bad bunch of discredited explanations.

This then is the position that the Cartesian legacy has led us to. Mental states are by their nature private objects subject only to direct and perfect perception by the owner of the mind wherein they are manifested. Characteristically, Wittgenstein challenged the claim that such a statement is meaningful and asserted that such notions are in fact illusions caused by the misuse of public language terms moulded by shared human experience to describe private entities which, as we shall see, are not ‘objects’ at all.

Wittgenstein’s starting point is to examine the way in which we refer to inner sensations using terms to which we ascribe a shared public meaning. How, he asks, can it be meaningful to say that ‘I have a pain’ and that ‘you have a pain’ and that we both agree on the meaning of ‘pain’ while asserting that my pain is separate and distinct from your pain? The problem here is the misuse of the terms ‘my’ and ‘your’ in assigning ownership to ‘pain’. Just as an attribute such as ‘red’ cannot be owned, neither can pain be owned for it is not an object but a description of the state of the person. My book is red and also your book is red and we can agree that both our books are ‘red’ while seeing that they do not have ‘the same’ red. Can we say then that two pains can never be numerically identical – only qualitatively identical? We might share the ‘same’ pain in the sense of bodily location, intensity and frequency but my pain is separate and distinct from your pain. But surely talk of identity conditions is only meaningful for objects and not for attributes? When we talk of the pains of two individuals then surely we need another sense of identity which does not involve ownership? I can say that ‘my pain’ is the same as ‘the pain that I have’ but this is different from ‘the pain that belongs to me.’

The problem here perhaps is again due to a false association of terms such as ‘I’ and ‘me’ to an object. What does the ‘I’ of Descartes’ cogito refer to? Where is the ‘I’ that turns its gaze inward and perceives my inner mental states and ‘sees’ that I am in pain? Public language is inadequate to enforce a logically consistent and objective meaning to terms such as ‘I’. It may be said that ‘I am thinking’ (my mind is thinking?) , that ’I have two legs’ (my body has two legs?), and problematically that ‘I have a pain’ (mind, body or something other?) Person-relative terms such as ‘I’ and ‘me’ are perhaps indicator-words in the same way as ‘now’ and ‘here’ are relative to context and although linguistically they play the syntactic role of nouns, their meaning is subtly different. Were the term ‘I’ to refer to some actual object which performs the introspection then we immediately have the familiar homunculus regress which asks what is it within the ‘I’ that perceives the inner states of ‘I’ as it in turn perceives the first set of mental states?

But surely there is no inner sense when we talk of our knowledge and experience of our inner mental states? There is no sense in which I can change the viewpoint of my introspection upon myself – I cannot move further from my pain, change its perspective or regard it from above or below. When I see the book as red then there is no third person viewpoint within me that watches me in the process of perceiving red – I just do it. When asked whether I am in pain I do not mentally ask my ‘inner eye’ to take a look and see whether I am indeed in pain and report back. I can say that ‘I am in pain’ but, as Wittgenstein asked, what can it mean to say that ‘I know I am in pain’? Neither can I say that ‘he is in pain but he is not aware he is in pain.’ Wittgenstein says that it is meaningless to talk of knowing something unless that perception is contingent and subject to doubt. Since I can never doubt that I am in pain (if it seems to me that I am in pain then I am in pain) it is incorrect to talk of me knowing that I am in pain. In other words, I have no criterion of identity for deciding that I am in pain. The key move here then is to say that if this is so then there is no privileged vantage point from which this knowing becomes possible. Wittgenstein does not refute the concept of our direct and perfect knowledge of our inner states but challenges the meaningfulness of talk of privileged access and of the privacy of mental states. The statement ‘X has privileged access to S’ only has meaning if it is logically possible to assert that ‘X does not have privileged access to S’ and this is indeed not possible in the case of our mental experiences.

Wittgenstein argues that I use the vocal statement ‘ I am in pain’ not to label or point to some inner mental object but to announce to others that I am in some mental state. Our use of public language here is an adaptation of primitive unlearned pain behaviour where we replace crying and wincing with statements such refer to some public concept that stands in place of other pain behaviour. Someone who, when asked whether they are in pain, says ‘I don’t know…’ and proceeds to perform some conscious introspection could be said to misunderstand the meaning of the words ‘you’ and ‘I’.

What are we to say about the public use of words which purport to refer to private mental objects and sensations? Imagine a group of people sitting in a circle, each holding a pin which they then stick into the palm of their hand. Apart from having grave suspicions about possible sadomasochistic tendencies, what can we infer from the nature of the ensuring conversation? They each profess to be in pain and appear to be in perfect agreement with each other that they are indeed in pain and that they all agree on the meaning of ‘pain’. But the pain each person feels is known only to that person and is in truth inaccessible in all its phenomenal detail to all other members of the group. ‘Pain’ here means ‘that which I experience when I stick this pin into my hand, all things being equal.’ But suppose the group includes a zombie who has no phenomenal experience whatsoever when the pin is stuck deep into his hand. Can the zombie talk meaningfully of his empty experience as ‘pain’ in this context? Wittgenstein proposes a similar model where each member of the group holds a closed box containing some hidden object knowable only to the holder of the box. The group as a whole may use some public term, ‘zeebaz’ say, as meaning ‘that which is in my box’ even though my box may in fact be empty. If the holder of the empty box talks of his zeebaz then we can understand him and agree that he is in possession of a zeebaz and that ‘zeebaz’ is a correct description of the thing that he has. In this case then, zeebaz cannot be a private object and, by analogy, neither can descriptions of mental states such as pain. Wittgenstein would claim that terms such as ‘pain’ cannot refer to private objects but designate the set of conditions under which it is correct to describe your condition as being in pain. Use of descriptions such as ‘correct’ immediately implies the existence of socially administered and mediated rules (the ‘language game’) which determine appropriate usage for such terms – a social grammar in other words.

So, if we cannot meaningfully talk about ‘my specific pain’, or ‘the pain I am in right now’ in terms of the public term ‘pain’, then is there some private language with which I can make valid statements about my pain to myself? Is this in fact a coherent idea? Such a language must refer to things that only I can know (note how difficult it is here to escape from inappropriate uses of terms such as ‘know’) and must therefore be incomprehensible and without meaning for anyone else. After all, how can a third person be sure that I am using such a language correctly? More crucially however, how can I be sure that I am using the terms of this language correctly – who can I appeal to for confirmation that I have indicated this mental sensation with the correct private term? If there is no privileged viewpoint at one place removed from the immediate sensation of my mental states then I have no means of verifying the correct usage of my private language. If my usage seems right to me then indeed it must be right. How then could such usage ever be deemed incorrect and if the use of a linguistic term can never be correct or incorrect then, says Wittgenstein, the term is without meaning.

The grammar of sensation words as granted by a public language confers on the user of such a language a guarantee of correct usage by supplying appropriate identity criteria. These criteria allow the speaker to be sure that he has ‘latched on’ to the correct public object by his use of the corresponding public language term. The user of a private language through has no such identity criteria available to him because of the lack of public consensus over the object of references, these being private to the speaker. The private language speaker can never be sure that he has latched on to the same private object that he referred to as ‘S’ previously since he has no criteria for establishing the identity of the two private mental experiences. Private mental language then is impossible.

This conclusion has many implications for our conception of mental states. We cannot refer to private mental objects in a public language – we can only refer to the terms and conditions of correct usage of such public terms as ‘pain’. But neither can we refer to private objects using a private language for the reasons above. How then can be refer to these private objects in any way at all? This is a devastating blow for the Cartesian view of the mind which is built on the foundation of an implicit private language which is free from doubt and error. If I cannot even be sure of making correct or meaningful statements about my private inner mental states, to which Wittgenstein agrees I have private and perfect access, then how can I be sure of anything at all? Hence I cannot defend myself from the delusions of the evil demon and move outwards from the safe haven of my certain private world towards the physical material world?

The solution, as both Kant and Wittgenstein propose, is to abandon the Cartesian reliance on the first-person viewpoint and to adopt a third person perspective. Rather than asking ‘how is it that I can know?’ one should ask ‘how is it that I can have these doubts – what must reality be like if I cannot in fact have certain private knowledge?’ The very fact that I am able to express my doubts about my knowledge in terms of a language which is understood by others and which appears consistent implies that others are able to learn my language and apply it correctly. By thinking about whether I am in fact thinking, I must be talking a public language and this then implies that it is possible for other minds to share this language (without such other minds the public language has no meaning.) For Wittgenstein then, the starting point of our exploration of reality must be the existence of other minds and that we are ourselves part of a public realm which, though the self-verifying function of the public language, must be immune from the corruption of the Cartesian demon.

For Wittgenstein then, the Cartesian view is false and ‘mind’ is a meaningless idea until included as a public concept which somehow supervenes upon the busy traffic of the language game.