João Miranda
University of Lisbon
Contrastivism about epistemic reasons is the claim that reasons to believe are relative to sets of alternatives. Sets of alternatives are the content of questions (Hamblin, 1958). So, reasons to believe are relative to questions. Contrastivism about epistemic reasons is motivated by puzzles that generate inconsistencies when a non-contrastive account of epistemic reasons is assumed. There is, however, a deep issue with contrastivism. By relativizing reasons to sets of alternatives, it runs the risk of losing the power to validate inferences that we take to be perfectly fine. Too much relativity won’t allow for inferences across different sets of alternatives and, at least some of those, seem to be valid. If that is right, then not allowing those inferences greatly impoverishes the model’s power. In this talk, I’ll discuss Snedegar (2017)’s solution to the problem and argue that it can be improved on. Snedegar appeals to a notion of promotion – x is an epistemic reason to believe that p if and only if believing that p on the basis of x promotes knowledge. Snedegar’s solution can provide the constraints we want, but it faces two challenges. The first, is that it is not illuminating enough: the notion of ‘promotion’ is not less obscure than the notion of ‘reason’, and so explaining one at the expense of the other won’t get us very far. The second problem is that the reader is left with a feeling of ad hocness: promotion is said to have the right formal properties and that it does have those properties is a “minimal assumption about promotion” (Snedegar, 2017: 71). That, however, is something that the opponent of contrastivism just can’t accept lightheartedly. I’ll present an account that, while compatible with promotion, offers a solution to the problem of relativity that does not need to appeal to promotion. It appeals, rather, to conditional probability: x is a reason to believe that p if and only if the conditional probability of p given x is greater than some threshold τ. In particular, I’ll argue that that threshold is set by the alternatives, such that the notion of ‘reason to believe’ can be characterized by a Bayesian confirmation measure. I’ll conclude by discussing how the contrastivist should approach the problem of conciliating the probability values obtained relative to the different probability spaces generated by different questions.
Chair: Natalia Tomash
Time: September 8th, 16:00-16:30
Location: SR 1.004
SOPhiA 2026 