Abstract:
The aim of this note is to plug an important gap in our understanding of the epistemic foundations of uncertainty-averse behavior. For Choquet expected utility maximizers (Schmeidler (1989)), the beliefs which motivate uncertainty-averse choice are frequently identified using Dow and Werlang's (1994) notion of support for convex capacities. Building on the work of Morris (1997), we present a new, preference-based belief operator which is is shown to characterize such epistemic inferences. This makes their behavioral foundations transparent, and enables readier comparison with alternative epistemic models for such behavior.