The only way to settle conclusively what any part of a language means is to discover the circumstances, both linguistic and non-linguistic, in which the speakers of the language are prepared to use it. This is not a new doctrine, but Wittgenstein gave it new life by dramatising the following question: If someone used an expression in a radically non-standard way, could anything he said about his state of mind convince us that he nevertheless meant it in a standard way? (...) To answer ‘No’ to this, and to generalize that answer, is to say that the last-resort criteria for what something means lie in the way in which it is used - a fairly plain statement which I shall call ‘the behavioural theory (of meaning)’ and with which I shall here have no quarrel at all. What I wish to do in the following pages is to consider the relationship between the behavioural theory and some aspects of the concept of proof. It is beyond dispute that one can be led, by one’s acceptance of certain premisses, to accept a certain conclusion. There is, though, a problem about the nature of this ‘leading’. On the one hand, it is usual to think of it as sometimes having the nature of a forcing: ‘If you say that, you are committed to admitting this also; you cannot accept the one and reject the other.’ On the other hand, the behavioural theory of meaning makes it difficult to see how there can possibly be such a relationship between the premisses and the conclusion of any proof. There are many ways of bringing out the apparent clash between the behavioural theory and the notion of logical forcing or committal. Perhaps the clearest of them arises from asking how there can be room for a concept of committal in a purely behavioural study. In this spirit, we might grant that our knowledge of some of the ways in which the parts of a language are used may lead us to expect to find certain sorts of further use and not others; and we could compare such expectations with those of an anthropologist who finds that a certain society has a matriarchal system of authority and then proceeds to investigate its inheritance laws.. (shrink)