Skip to Main Content
The use of the preference relation `selection of A and rejection of B is preferable to selection of B and rejection of A' is discussed. This contextual relation yields descriptive models of choice under uncertainty that explain many observed violations of classical single attribute utility theory. It is shown that successful use of this relation, for descriptive purposes, requires careful attention to elicitation of realistic decision situation structural models and pre- and post-decision information availability conditions within these models. Reservations with respect to use of the regret model for normative purposes, due to potential nontransitivities of the contextual relation, are discussed.