3 Belief and truth

In defining a belief’s strength by its effects on our actions, Ramsey foreshadows later theories which define mental states by their causes, effects and interactions (see Functionalism). In ‘Facts and Propositions’ (1927) he extends this idea from degrees to contents of beliefs, taking (for example) ‘the equivalence between believing ‘‘not-p’’ and disbelieving ‘‘p’’ … to be defined [by their sharing] many of their causes and … effects’ (1990a: 44). But after failing to define the contents of beliefs as effectively as their degrees, Ramsey concludes that his view, that ‘the meaning of a sentence [expressing a belief] is to be defined by reference to the actions to which asserting it would lead’, remains ‘very vague and undeveloped’ (1990a: 51).

It is, however, developed enough to stop Ramsey’s theory of truth being, as is usually supposed, that truth is definable by the fact that for all p, it is true that p iff p (see Truth, deflationary theories of). He does say that ‘there is really no separate problem of truth’ since (for example) ‘ ‘‘it is true that Caesar was murdered’’ means no more than that Caesar was murdered’, so that ‘if we have analysed [belief] we have solved the problem of truth’ (1990a: 38–9). But the solution will not be the so-called redundancy theory if our analysis of beliefs includes a substantive analysis of their truth conditions, as Ramsey’s needs to do.

Ramsey starts by observing that we can‘say that a chicken believes a certain sort of caterpillar to be poisonous, and mean by that merely that it abstains from eating such caterpillars on account of unpleasant experiences connected with them’. Since this action is ‘such as to be useful if, and only if, the caterpillars were actually poisonous … any set of actions for whose utility p is a necessary and sufficient condition might be called a belief that p, and so would be true if p, i.e. if they were useful’ (1990a: 40, see Truth, pragmatic theory of).

Unfortunately Ramsey drops this idea when dealing ‘with those beliefs which are expressed in words … or other symbols, consciously asserted’, although it can apply to them too. In fact its only fault is to identify a belief with a set of actions, like abstaining from eating caterpillars, instead of with one of their causes. But any theory that makes beliefs entail causal functions from desires to actions can remedy this. For then the ‘set of actions’ of a full belief b will be all those that b would combine with some desire to cause; and p will be the condition in which every such action would succeed – that is, achieve the object of the desire involved, say to eat without dying. But this is obviously the condition that b be true, that is, b’s truth condition.

Ramsey can therefore let this ‘success semantics’ (Whyte 1990) give the truth condition of any belief definable by ‘the actions to which … it would lead’. Indeed he must do so, since a belief’s truth condition cannot be given just by how it makes us act, for that will be the same whether it is true or false. What does depend on a belief’s truth is whether the actions it causes succeed: hence success semantics. But this, while vindicating Ramsey’s claim that analysing belief solves the problem of truth, rules out the redundancy theory: for success semantics, since its contribution to the analysis of beliefs is to say what makes them true, is itself a substantive theory of truth.