Moderately Fine-Grained Theories of Content
Let me end by making some general remarks about the scope of this project. Firstly, the account of fine-grained propositions I have proposed here does not solve, and is not intended to solve, all of the problematic issues surrounding propositional attitude verbs like ‘knows', ‘believes', ‘desires’, and so on.
For example, the theory distinguishes between the proposition that Harry is bald and necessarily equivalent propositions stating the situation on Harry's head in more precise terms, but will identify the propositions expressed by any pair of tautologies, even if one tautology is significantly less obvious than another.[153] Call theories like this, that distinguish between necessary equivalents but not logical equivalents, moderately fine-grained. To see why a moderately fine-grained theory of propositions is motivated, I think it is important to distinguish two problems.One of these is a problem in the philosophy of language about propositional attitude reports. This is a puzzle about how we succeed in reporting facts about a person's mental states in English using verbs like ‘believes' and ‘desires' and so on. In my view, this is fundamentally a puzzle about how we use words to ascribe attitudes—relations between people and propositions—and not a puzzle about the attitudes themselves. Indeed, many of the most promising solutions to the problem of attitude reports say something distinctive about the kinds of words we use to ascribe attitudes, whilst in principle remaining compatible with various theses concerning how fine-grained the arguments of these attitudes are.[154] A contextualist, for example, postulates the existence of a number of distinct but related propositional attitudes, which the word ‘believes' can pick out in different contexts; when it appears as though we are making conflicting attitude reports about the same proposition, the Contextualist says we are really attributing different attitudes to the same proposition.
This is how it should be—if we took all of our immediate judgements about differences between propositional attitude reports to show that there is a corresponding difference between the objects of the propositional attitudes, propositions would be as fine-grained as sentences, or maybe even more fine-grained.[155] According to the theories I just listed, one cannot infer much about the fine-grainedness of propositions from judgements about attitude reports.
To make this vivid, note that these theories seem to have the explanatory power to reconcile our judgements about attitude reports with the view that propositions are sets of worlds, or even truth values![156]The reasons I take to motivate a moderately fine-grained theory of propositions have nothing to do with accommodating attitude reports, or self-ascriptions of belief. The linguistic data on belief reports, despite the attention it receives, is only a small aspect of a full account of propositional attitudes. It is important not to forget that a person's propositional attitudes are also important for explaining and evaluating all kinds of behaviour, both verbal and non-verbal. An instructive example is the view that there are only two propositions: the true and the false. According to this view, it is impossible to explain a person's behaviour in terms of their beliefs and desires, for there are only sixteen types of people depending on the combination in which they believe or desire the two propositions, and this is certainly not enough to explain or evaluate all the possible kinds of behaviour they exhibit. Thus in order to explain or evaluate a person's behaviour in terms of their propositional attitudes, propositions must be more fine-grained than truth values. This line of reasoning plausibly generalizes to rule out sets of worlds as well—two people with necessarily equivalent beliefs and desires can rationally behave very differently. On the other hand, the fact that propositional attitudes determine rational behaviour gives us a defeasible reason to think that propositions cannot be so fine-grained as to distinguish logical equivalents. This is because our best theories of rational action—decision theory and probability theory—assume that propositions are moderately fine-grained. Decision and probability theory typically begin by assigning probabilities and utilities to indices of some sort and from this one assigns probabilities and expected values to arbitrary sets of these entities.
To treat propositions as isomorphic to sets of entities is just to assume that propositions are structured like a Boolean algebra which, in turn, guarantees that equivalence in classical propositional logic suffices for identity.VAGUE PROPOSITIONS 221 (Note that it is possible to do decision theory and probability theory with a very finegrained account of propositions by assigning values and probabilities to equivalence classes of logically equivalent propositions. Those who insist on such a fine-grained account of propositions can still accept my theory as an account of a kind of theoretical entity: entities that play the theoretical role equivalence classes play in the more finegrained account.)