What Do We Believe In?

In this article, I will outline, how three meaning theoretic paradigms deal with the content of belief, and what difficulties each of them faces: the Fregean approach, the Neo-Russellian or direct reference approach, and the dualist approach as a mixture of both. The Fregean approach stipulates senses that correspond to linguistic expressions and fulfill various roles such as determining the truth-conditions of a sentence or encoding cognitive value. The direct reference approach can be traced back up to John Stuart Mill’s treatment of proper names, and it is also a part of Russell’s theory of knowledge by acquaintance. In this view, the meaning of a simple assertoric statement can be described as a singular proposition, i.e. an ordered n-tuple consisting of the objects denoted by referential expressions and the relation holding between them.1 The mixed view enriches a singular proposition with some way how particular objects are given to someone. During last century until now, entities that at first only served as a representation of sentence meaning consuetudinarily have been abused to represent the content of what a person believes as well. In this view, belief is a relation between a person and something which specifies what the person believes, whereas the content of what a person believes is closely related to some entity that represents the meaning of a natural language sentence. This is a rather complicated construction. It involves at least four kinds of entities: persons, contents of belief, meanings of sentences and sentences. And there are three kinds of relations involved: A belief relation with a reading like ‘x is in the belief-relation to y’ (or similar), a meaning relation with a reading like ‘x is a meaning of y’, and finally a relation between meaning and belief content with a reading that is hard to specify. Perhaps this relation is best regarded as a function translating meanings into belief contents, and I will useC(x) to denote this function. Likewise, for explanatory purposes, let M(x) be a function from sentences into sentence meanings, and B(x) be a function from speakers and belief contents into truth or falsity.2 (If you insist that belief is gradual, you can make it ‘fuzzy’ by allowing B to be a function into the interval of real numbers [0 . . . 1] indicating the strongness of belief. For reasons that cannot be addressed here, I do not recommend this variation, though.) The general picture is then, that for any sentence S, the fact that a speaker a holds a belief that S is true (to some extent, in the fuzzy view) is described by a mapping from S to the result of B in the following manner: