Extensionality, Indirect Contexts and Frege's Hierarchy

It is well known that Frege was an extensionalist, in the following sense: he held that the truth-value of a sentence is always a function only of (its syntax and) the references (the ‘extensions’) of its parts. One consequence of this view is that expressions occurring in certain linguistic contexts – for example, the that-clauses of propositional attitude ascriptions – do not have their usual references, but refer instead to what are usually their senses. But although a number of philosophers have objected to this result, no one has yet attempted to see what happens to Frege's views – and, in particular, to his theory of sense and reference – if his extensionalism is abandoned while his other views are (so far as possible) maintained. This paper thus does two things. First, it clears the way for such an attempt, by arguing that recent defences of Frege's extensionalism – by Tyler Burge, Saul Kripke, Terence Parsons and Christopher Peacocke – fall short. Second, it sketches a positive proposal for a non-extensionalist application of Frege's theory of sense and reference to sentences of indirect discourse and ascriptions of propositional attitude.