Counterfactuals, Probability, and the Direction of Time
According to David Lewis’ well known account of counterfactual “If A had been then B would have been” is true if there is a possible world in which A and B are true similar to the actual world and there are worlds as similar to the actual world in which A and not B are true. Lewis then provided an account of similarity that seemed at first to give the right results for a wide class of counterfactuals relevant to accounts of causation, influence, reliability etc. But it turns out his account completely fails. In my talk I propose an alternative account in terms of probabilities that makes no use of possible worlds. The probabilities come from statistical mechanics and so connect the temporal asymmetry of counterfactuals and causation directly to the temporal asymmetry of entropy- something which puzzled Lewis.