Lewis' counterfactual analysis of causation

Abstract

David Lewis offers a counterfactual analysis of causation, limiting his analysis to causation between particular events.1 Exactly what Lewis is analyzing is revealed by the point at which he announces his analysis to be complete. He does so immediately after giving neces sary and sufficient conditions for one event's causing another.2 One event is a cause of another if and only if there is a causal chain that leads from the first to the second. The notion of a causal chain is defined by appeal to causal dependence, which is defined in terms of the truth of certain counterfactuals. The truth value of a counter factual is determined by the truth value of its consequent in those possible worlds in which the antecedent is true which are most similar over-all to the actual world.3 Two sorts of criticisms of Lewis' account have been leveled elsewhere. Jaegwon Kim objects to the account's classification of certain kinds of cases as ones of causal dependence. Among these are ones which Kim says exemplify an 'analytical' or 'logical' relation, and others in which one event is a part of another.4 Bernard Berofsky objects to Lewis' contention that the vagueness of counterfactuals infects causation.5 I agree that the cases cited by Kim and the features of Lewis' analysis cited by Berofsky present problems for his account. However, another potentially far more damaging line of criticism will be offered here. I shall argue that the adequate determination of a set of possible worlds as closest or most similar to the actual must depend upon causal relations. Without appeal to causal relations (though not the particular ones Lewis analyzes here), the truth values expected will not be determined for many counterfactuals whose truth values are pre-analytically clear. And if a set of worlds is determined as closest to the actual without appeal to causal relations, the relevance of those worlds to the causal relations between particular events in the actual world is obscure. Let us begin, however, by elaborating Lewis' account.