Counterfactuals and the Epistemology of Modality

Admittedly, it is not immediately clear how Lewis’s theory about when such-and-such counterfactuals are true can be turned into a theory about how we evaluate such-and-such counterfactuals as true. Still, to the extent that his theory is plausible and our evaluations of counterfactuals accord with it, we should not be surprised if the ability to correctly evaluate counterfactuals like ‘If I had lit a campfire, I would not have been attacked by a tiger’ brought with it the ability to correctly evaluate counterfactuals like ‘If I had lit a campfire, a small breach of our laws of nature would have occurred’. Thus, the ability to evaluate counterfactuals with nomologically possible antecedents and consequents would bring with it the ability to correctly evaluate counterfactuals with nomologically possible antecedents but nomologically impossible consequents. From the latter, it is a short way to counterfactuals with nomologically impossible antecedents. Generally, (p & q) □→ r may be inferred from p □→ q and p □→ r. Assuming that we are able to reason according to this rule, we may infer ‘If I had lit a campfire and a small breach of our laws of nature had occurred, I would not have been attacked by a tiger’ from ‘If I had lit a campfire, a small breach of our laws of nature would have occurred’ and ‘If I had lit a campfire, I would not have been attacked by a tiger’. Since the conjunction of a nomologically impossible proposition with an arbitrary proposition is still nomologically impossible, the antecedent of ‘If I had lit a campfire and a small breach of our laws of nature had occurred, I would not have been attacked by a tiger’ is nomologically impossible. Thus, competence with nomologically impossible antecedents and consequents is not too difficult to achieve in principle.