Abstract

We have lots of good arguments for a variety of epistemic norms on how you should plan to change your credences or beliefs upon coming to possess new evidence. We don’t have many good arguments for how you should actually change your credences or beliefs in response to receiving new evidence. Sure, we do have some arguments for actual epistemic norms, but none of them are the gold standard in the field, that is, none of them are accuracy-dominance arguments. Here we fill this gap. Doing so requires some conceptual development about good and bad ways to evaluate your epistemic performance. In short: your evidence, while not directly placing constraints on your rational attitudes, places a constraint on how you should evaluate your epistemic performance. If you possess evidence E, it seems, from your point of view, bad to take non-E worlds as relevant to the assessment of your epistemic performance. Using this idea, we develop an accuracy-dominance argument for Actual Conditionalization and a variety of other actual updating norms.