Vladan Djordjevic

An incorrect interpretation of Goodman’s theory of counterfactuals is persistently being offered in the literature. I find that strange. Even more so since the incorrectness is rather obvious. In this paper I try to figure out why is that happening. First I try to explain what Goodman did say, which of his claims are ignored, and what he did not say but is sometimes ascribed to him. I emphasize one of the bad features of the interpretation: it gives counterfactuals some formal properties that neither Goodman nor (usually) the interpreter would accept. The usual interpretation (UI), which I claim should not be ascribed to Goodman, considers a counterfactual A>C true iff A, together with natural laws and all contingent truths cotenable with it, entails C. (UI) makes valid the law of conditional excluded middle, which Goodman clearly rejected. Among possible reasons for which the interpreters might find (UI) adequate is that (UI), as I argue, smuggles in the idea of minimal change, which is otherwise attractive, natural to many, but not to be found anywhere in Goodman’s paper. At the end I stress the significance of Goodman’s theory by arguing that we still need some of his notions to test the adequacy of our modern theories.

This paper is about the distinction between arguments and conditionals, and the corresponding distinction between premises and antecedents. I will also propose a further distinction between two different kinds of argument, and, correspondingly, two kinds of premise that I will call "assumption" and "hypothesis." The distinction between assumptions, hypotheses, and antecedents is easily made in artificial languages, and we are already familiar with it from our first logic courses (although not necessarily under those names, since there is no standard terminology for the distinction). After explaining their differences in artificial languages, I will argue that there are ordinary-language counterparts of these three notions, meaning that some formal properties of the artificial notions nicely capture some features of the ordinary-language counterparts and their behavior in contexts of reasoning. My next crucial claim is that these three notions often get confused in ordinary language, which leads to problems for translation into symbols. I will suggest a solution to the translation problem by pointing to some distinctive characteristics of the three notions that link them to their artificial-language counterparts. Next, I will argue that this confusion is behind some well-known philosophical problems and puzzles. I will apply the distinctions in order to explain away some famous paradoxes: the direct argument (also known as or-to-if inference), a standard argument for fatalism, and McGee's counterexample to modus ponens. As Stalnaker also solved the first two of these paradoxes by using his theory of reasonable inference, I will elucidate the similarities between our solutions, and also explain why my distinctions apply more broadly, to some cases involving indicative and counterfactuals conditionals, where reasonable inference does not apply.