Evaluating ambiguous random variables from Choquet to maxmin expected utility

We introduce a new theory of belief revision under ambiguity. It is recursive (random variables are evaluated by backward induction) and consequentialist (the conditional expectation any random variable depends only on the values attains conditioning event). Agents experience no change in preferences but sensitive to timing resolution uncertainty. provide three main theorems: first characterizes our rule relates it standard Bayesian updating; others show that dynamic behavior an agent who adopts maxmin expected utility with arbitrary set priors.