What trans-world causation could and could not be

Eduardo García-Ramírez has offered a reductio of the counterfactual analysis of causation. The argument purportedly shows that, given a natural generalization of Lewis’ semantics for counterfactuals, statements expressing the existence of causal dependence across worlds are satisfiable. The aim of the present paper is twofold. In the first part, I show that the purported reductio is flawed, as it relies on an overly strong construal of the semantics for counterfactuals. In particular, it is assumed that we can assign a degree of similarity to any given pair of possible worlds. As it turns out, that assumption reduces to the thesis that the relations of comparative similarity featured in the standard semantics for counterfactuals define an interval scale of measurement on the set of all possible worlds. It will be argued that such a thesis is incompatible with a viable understanding of comparative similarity. The second part of the paper is devoted to a new proof of the possibility of trans-world causation. Nevertheless, the new proof does not amount to a reductio of Lewis’ account of causation per se, but rather of the conjunction of several substantive theses (the counterfactual analysis of causation, modal plenitude, the existence of mereological sums and the best theory account of natural laws).