To: Robert H.
From: Geoffrey Klempner
Subject: Wittgenstein on sensation 'S'
Date: 14 July 2006 09:36

Dear Robert,

Thank you for your email of 3 July, with your first essay for the Philosophy of Language program, on the subject of Wittgenstein's sensation 'S'.

Wittgenstein's point is that we SEEM to understand the idea of writing 'S' in a diary, but further questioning reveals that the idea is in fact nonsense. Hence, objections that 'Wittgenstein has failed to give an adequate definition of a private language' miss the point. Wittgenstein is arguing that cannot be such a language. 'S' cannot have the meaning ascribed to it in the thought experiment.

Wittgenstein is not talking about the 'feelings and resonances assigned to a symbol' but strictly about 'the association of a symbol or word with an object'. Associating words with objects is something we do all the time. So what has gone wrong in the 'S' case?

It has been claimed by some critics of Wittgenstein that the argument against the possibility of a private language depends crucially on a 'verificationist assumption'. You say something that suggests this: 'If language depends on rules, and these rules have to be verifiable then Wittgenstein's thought experiment suggests that the private naming of sensations cannot occur.'

It would be far too strict a requirement to say that the application of linguistic rules should always be verifiable. Suppose I doubted whether I am using a certain colour word consistently. I can look at colour charts, ask other people's opinions and so on. But there is no way to remove the doubt entirely. In practice, however, we do not usually entertain such doubts. We take it for granted that we are using words correctly, even when there is no opportunity to check that this is so.

Wittgenstein makes a far less exacting requirement. The correct or incorrect use of a word must be capable of having some consequences, but it needn't always have consequences in practice (e.g. when you use a word incorrectly and no-one corrects you). The problem with the way 'S' is defined, however, is that there is no way that a 'mistake' in using 'S' could ever have any consequences. I can say what I like and never be 'wrong'.

The crucial sentence in the quote is, 'I will remark first of all that a definition of the sign cannot be formulated.'

Wittgenstein is using 'definition' in a much broader sense that the usual one. For example (quoting from the program), 'giddy' is what you feel when you go on a roundabout, 'pain' is a feeling sometimes caused by injury which makes you cry out and rub the effected part.

In a later paragraph, Wittgenstein gives the example of the discovery that a certain feeling is associated with a rise in blood pressure. You go to a doctor and he confirms that 'S' usually occurs when your blood pressure is too high. So now it is NOT the case that 'a definition of the sign cannot be formulated'. This is like your story of the 'computer physician'.

The philosopher who wants to say that 'a definition of the sign cannot be formulated' is in the grip of the idea that my subjective impression of 'S' has a reality which is completely independent of the physical world. Even if no changes in my body corresponded to the occurrence of S, there would still be such a thing as 'following my own rule for the use of S'. But this is an illusion according to Wittgenstein. There is no 'rule' because I am free to say what I like. 'Right' and 'wrong' are defined purely in terms of what I say, in which case there is no such thing as 'right' or 'wrong'.

Does God use a private language? That's a very interesting question. Would it be enough if God 'created the world and humans to ensure that she (?) had something to talk to'? The problem is that God is supposedly omniscient and therefore not in the position of a language speaker who stands to be corrected by others. God can never be wrong about what he/she means by a word.

Similar things have been said about bees and dolphins. It appears from ample evidence that bees through their 'bee dance' are able to transmit information which is used by other bees. Similarly, dolphins emit underwater sounds which other dolphins interpret as a description of objects in the vicinity.

However, the transmission of information is not the same as language use. The bee does not have a choice about 'how' to dance. It does what it does by instinct, and the other bees similarly 'interpret' what it does by instinct. In principle, all we are dealing with here is a chain of causes and effects which produces information in the recipient.

Voice recognition programs could get very sophisticated. At some point, a computer armed with voice recognition would be a suitable candidate for the Turing Test. If you could keep up a conversation with it indefinitely, then no further question arises as to whether it is 'really' intelligent'. In that case, we would be treating the computer as a member of the linguistic community, correcting it's mistakes and allowing it to correct ours.

In other words, the point where we are dealing with language, rather than mere transmission of information is the point where 'right' and 'wrong' come in, the point where 'mistakes' are possible.

All the best,

Geoffrey