On the Consistency and Reversibility of Certain Sequences of Counterfactual Assertions

Abstract

Abstract This paper is about Sobel sequences, which are sequences of counterfactuals that supposedly display two interesting properties: first, they are consistent, as accounted for by the famous Lewis-Stalnaker analysis; but second, they are not consistent in the reverse order, which is not accounted for by Lewis-Stalnaker. I argue that there has been an empirical oversight in the literature on these sequences: there are consistent sequences (which I call true Sobel sequences), and there are irreversible sequences (which I call Lewis sequences), but no sequence is both. The Lewis-Stalnaker theory neatly captures Sobel sequences, and also captures the inconsistency (in both directions) of Lewis sequences. But there is still a dynamic asymmetry at work with Lewis sequences, which I account for by appeal to the pragmatic phenomenon of imprecision. Lewis sequences seem consistent because they do not belie factual disagreements, but rather metalinguistic disagreements about how precise to be. This also accounts for their irreversibility—such metalinguistic disagreements about the standard of precision are only possible in one direction, as famously observed by Lewis (1979), who observed them in discourses not containing counterfactuals.