Humblebrag, bribes or bootstrapping? Act-state dependence in epistemic decision theory

Julien Dutant (King's College London)

Epistemic rationality is about doing the best, from one's perspective, to satisfy the twin goals or imperatives of reaching truth or knowledge and avoiding their opposites. Epistemic decision theory (EDT) uses decision-theoretic tools to turn this idea into a systematic theory of rational belief. It can be truth-oriented (veritistic) or knowledge-oriented (gnostic). When restricted to beliefs in propositions whose truth isn't dependent on what we believe, it's clear what norms veristic EDT entails: Lockeanism, according to which it's rational to believe something iff that is likely enough. But for many propositions, whether they're true depend on what you believe—propositions about what you believe, for one. And arguably, for all propositions, whether you know them to obtain partly depends on what you believe—namely, on whether you believe them. When we expand EDT to consider such cases, several choice points arise. I'll argue that every option must contend at least one of Moore-paradoxical humblebrag ('I don't know p, but p'), epistemic bribes (believing something blatantly false to gain further knowledge), or ignorance bootstrapping (if you don't believe something, it's rational not to believe it). I'll suggest that the best option combines gnosticism and Konek and Levinstein's (2018) way of dealing with epistemic act-state dependence. It entails a mild version of ignorance bootstrapping that we can live with.