Infra-Bayesianism is a new approach to / / reinforcement learning theory, which builds on "imprecise probability" to solve the problem of prior misspecification / grain-of-truth / nonrealizability which plagues and Bayesian reinforcement learning.
Infra-Bayesianism also naturally leads to an implementation of , and (more speculatively at this stage) has applications to multi-agent theory, and reflection.
See the Infra-Bayesianism Sequence.