Abstract

The main tenet of this dissertation is that a conditional in English is true exactly if there is an argument from its antecedent, in conjunction with certain background assumptions, to its consequent. In chapter 1, I argue that the traditional equation of indicative conditionals with material conditionals is inadequate and cannot be saved by imposing epistemic or pragmatic constraints on assertion. I then show that gappy truth-functional logic also does not deliver a satisfactory analysis. In chapter 2, I discuss a family of theories that measure the assertability of indicative conditionals by the corresponding conditional probabilities. I argue that high conditional probability is not sufficient for assertability and that the foundation of the probabilistic view has implausible psychological ramifications which make the view hard to defend. Chapter 3 is about conditional belief, a kind of mental state that is different from ordinary belief in a conditional proposition. I claim that conditional belief is standardly expressed in conditional sentences, which gives an important clue to the semantics of these sentences. The fourth chapter contains the epistemic analysis of both indicative and subjunctive conditionals. I suggest that a conditional 'If A, then B' is true just in case B is implied by a background theory selected according to the context of utterance and modified twice. First the theory is rendered consistent with the B's denial; then it is expanded or revised by A. I locate the distinction between indicative and subjunctive conditionals in the selection of background assumptions and constraints on revision. In chapter 5, I sketch a sentential framework for modeling background theories and the mechanics of revision and discuss questions about uniqueness, consistency, and closure. This clears the ground for the beginnings of a logic of conditionals. In the last chapter, I propose a revision procedure for theories written in a canonical first order language