Objective rationality and recursive multiple priors
Résumé
We provide an axiomatic characterization of recursive MaxMin preferences that stem from a (possibly) incomplete relation representing preference judgements that are justified by objective probabilistic information. To ensure that her choice behavior be informed by objective information, dynamically consistent, and ambiguity averse, the decision-maker evaluates alternatives by their minimal expected value over the rectangular hull of the objective information set. The characterization builds upon two axioms that naturally combine the three requirements in a behavioral way. Our main result suggests a principled justification for the use of recursive MaxMin preferences in applications to dynamic choice problems.