Sense and Evidence

Abstract

In his writings on philosophical logic, Gottlob Frege distinguished between two semantic factors that, he thought, are associated with any fully meaningful expression.1 The first of these is the expression's reference (Bedeutung); philosophers and linguists now have a good understanding of how the reference of many a complex expression depends on the references of its parts. But Frege also held that any meaningful expression has a sense (Sinn), and that two expressions may differ in sense while sharing a reference (see Frege 1892, 25-27). The project of saying how the sense of a complex expression relates to the senses of its parts is less advanced, and pessimism about the prospect of carrying it to a successful conclusion is a source of doubt whether expressions possess senses whose identities are not fixed by their references, notwithstanding Frege 's arguments and examples. In this essay, I construct a compositional theory of senses for a small but nontrivial fragment of a natural language. I thereby hope to vindicate the thesis that a satisfactory semantic theory should treat of intensional as well as extensional elements against a corrosive source of scepticism.1. Problems in Accommodating SenseSome popular frameworks for semantic theory provide no houseroom for senses. Each of a truth-value - the closest correlate to a complete statement in Frege 's formalized language2 - expresses a sense, a thought. Namely, by our stipulations it is determined under what conditions the name denotes the True. The sense of this name - the thought - is the thought that these conditions are fulfilled (Frege 1893, §32, 50). This famous passage from Grundgesetze suggests that a statement's sense may be identified with its trum-conditions, but a popular account of truth-conditions makes the identification impossible to sustain. According to that account, a statement's truth-conditions are given by the possible worlds, or possible circumstances, at which it is true; 'possible' means 'metaphysically possible'.3 This account, however, immediately generates a problem for a Fregean. Let us imagine a context in which it is not common knowledge - perhaps because it is not known at all - that the names 'Hesperus' and 'Phosphorus' share a reference. Frege would certainly maintain that the statements 'Hesperus is bright' and 'Phosphorus is bright', as used in such a context, differ in sense. Since Kripke's Naming and Necessity (1980), however, most philosophers have come to accept that it is metaphysically necessary that Hesperus is identical with Phosphorus, so that any metaphysically possible world, or metaphysically possible circumstance, at which one statement is true is a world or circumstance at which the other is true. That is to say, the familiar modal gloss on the notion of a truth-condition fails to accommodate a key thesis of Frege's about sense.Another familiar gloss on that notion faces a different problem. Some philosophers take a statement's truth-conditions to be given by its canonical T-theorem in an interpretative truth-theory for the language to which the statement belongs. (See Wiggins 1976 and McDowell 1977.) It is then argued that the difference in sense between our two statements is shown by the fact that the pair of T-sentences(HB) 'Hesperus is bright' is true if and only if Hesperus is bright and(PB) 'Phosphorus is bright' is true if and only if Phosphorus is brightare theorems of a truth-theory that is interpretative for the speakers in the imagined context, whilst the pairs(HB!) 'Hesperus is bright' is true if and only if Venus is bright and(PB,) 'Phosphorus is bright' is true if and only if Venus is bright and for that matter(HB2) 'Hesperus is bright' is true if and only if Phosphorus is brightand(PB2) 'Phosphorus is bright' is true if and only if Hesperus is brightare not interpretative for those speakers.4 The different status of these Tsentences reflects the fact that the pair of reference assignments(H) 'Hesperus' refers to Hesperusand(P) 'Phosphorus' refers to Phosphoruscan serve as axioms of an interpretative truth-theory, whilst the pairs(H,) 'Hesperus' refers to Venusand(P,) 'Phosphorus' refers to Venusand for that matter(H2) 'Hesperus' refers to Phosphorusand(P2) 'Phosphorus' refers to Hesperuscannot so serve. …