We provide a new characterization of Bayesian expected utility maximization. The novel feature of the model is our separation of actions from prizes. The signature of the resulting theory is the impossibility of raising utility by switching wholesale from one action to another. We provide applications to robustness, to the recovery of utility from choice data, and to model classi cation.
Stored only when both papers are in the corpus — 16 citations exist upstream.
Imported from OpenAlex (CC0) — nothing here is generated.