Decision makers can reveal their preferences for alternatives using different methods or response modes. For example, a decision maker might choose between two alternatives or match them by adjusting a feature of the first so that it is equally as preferred as the second. According to normative theories of choice, preference orderings should be invariant across response modes, a property called procedure invariance (Tversky, Sattath, & Slovic, 1988). Empirically, however, different response modes can reveal very different preferences. The most well-known inconsistencies are demonstrations that preferences for simple gambles differ systematically across response modes (Slovic & Lichtenstein, 1983). Because these preference reversals are contrary to the most basic principles of rational choice, they have drawn substantial attention from both economists and psychologists.

The theories that have been proposed for preference reversal posit psychological mechanisms that could create different evaluations of the same gambles in different response modes. These theories have generally been tested through the extent to which they account for empirical patterns of inconsistent responses. By this standard, most of the proposed theories achieve some success, accounting for reversals in at least one pair of response modes. This research has largely used the characteristics of stimuli to manipulate these hypothesized psychological mechanisms. More recent research has studied these mechanisms through process tracing measures (Johnson, Payne, & Bettman, 1988) or by introducing choices involving riskless outcomes and then analyzing response patterns to reveal the influence of intransitivity and the tendency to overprice or underprice certain types of options (Tversky, Slovic, & Kahneman, 1990).