Subscribe to Elucidations:

This month, we are joined by R.A. Briggs (Stanford University), who is here to discuss an interdisciplinary area of study called epistemic decision theory. Click here to listen to our conversation.

Epistemic decision theory is an area of study that brings together two sub-disciplines. The first is decision theory, which tries to mathematically study the best principles for deciding what to do: what are the costs and benefits of each option you’re considering, and how can you optimize the decision process so as to at least the worst options? The second is formal epistemology, which tries to mathematically study how much credence to place in some hypothesis once you’ve uncovered new evidence for it. It might seem straightforward when you’re looking at just one hypothesis and one piece of evidence, but it quickly gets complicated once you start juggling multiple competing hypotheses and multiple kinds of evidence.

In this episode, our guest discusses a rule from probability theory called ‘conditionalization,’ which (roughly) is a rule telling you what your belief in a hypothesis would be, if you were to gain some evidence. For instance, maybe I left my cheesecake out. If I were to see tiny rodent footprints in it, I would conclude that there were mice in my apartment. I don’t in fact see any such footprints, but I still have a plan, as it were, for what my belief would be if I did. The conditionalization rule says how to arithmetically compute what my exact level of belief that there are mice in my apartment would be, if I were to see those footprints.

The question is: what is the status of this probabilistic reasoning rule? Are you always more accurate when you follow it? (As it turns out, that’s not exactly true, but something close to it is!) Does following it inherently make more sense than not following it? Etc. Our guest presents a new justification for following this rule that doesn’t depend on the accuracy of your previous beliefs. Tune in to hear about it!

Matt Teichman