Abstract

In this paper, I use the evidential function of chances and counterfactuals to develop accounts of these relations. Chances are objective worldly probabilities that allow us to reason from the state of a system at one time to the state of a system at another time. Counterfactuals are used to reason about what evidence we would have in hypothetical cases—and so, I’ll argue, are evaluated by considering ‘branch points’ where the counterfactual antecedent had a reasonable chance of coming about. An upshot is that counterfactuals play a distinct role from causal relations—and it was a mistake to look for a single set of probabilities that would determine both how we should reason about the world (counterfactuals) and how states of the world depend on others under conditions of manipulation (causal relations). These evidential accounts do better than nearby statistical-mechanical accounts in identifying the function of distinctly modal relations and justifying why we reason about the world in modal terms. They also do better at identifying the source of the temporal asymmetries in these relations—they are due to a probability gradient of the universe, rather than to a special initial state.