DIRECT REFERENCE AND ASCRIPTIONS OF BELIEF

It is often supposed that demonstratives and indexicals are devices of direct reference that they are, as David Kaplan puts it, terms which “refer directly without the mediation of a Fregean Sinn [or individual concept, set of properties, etc.] as meaning.“’ Most of the resistance to this view, I think, arises from the suspicion that it is not possible to give an acceptable treatment of the semantics of belief ascriptions and other so-called propositional attitude contexts which is consistent with the thesis of direct reference. For it seems that a straightforward construal of the thesis (along with some plausible semantical assumptions) requires that demonstratives, when co-referential, be intersubstitutable everywhere, even in belief contexts, sulvu veritute. But many feel that it is obvious that such substitutions do not always preserve truth. The purpose of this paper is to motivate and present a semantics for a first-order treatment of belief ascriptions which is both consistent with the thesis of direct reference and intuitively satisfactory. The paper is structured as follows: In Section I, I discuss semantical consequences of the thesis of direct reference in particular, what it does and does not require with respect to the overall form of a semantical treatment of belief ascriptions. I also discuss a view about belief, championed by Kaplan and John Perry, which I call the triadic view of belief. Crudely put, it is the view that belief is a triadic relation among a person, a proposition, and a sentential meaning, the latter entity a different sort of thing than a proposition. On this view, to believe a proposition is to do so ‘under’ a sentential meaning. The champions of the triadic view of belief have shied away from using the view to motivate a semantic account of belief ascriptions.’ But the triadic view of belief suggests, as I note at the end of Section I, that ascriptions of belief not only imply that a proposition is an object of belief, but that it is believed in a certain way. The purpose of Sections II and III of the paper is to show that an account of belief ascriptions, on which they behave in just this way, can be formalized rather easily, and that it nicely handles certain cases which, at first blush, seem quite problematic for those who