Evidence from experiments on individual decision making reveals that people's choices violate a variety of compelling principles. For example, while it seems compelling that people's choices should be transitive (e.g., that if a person chooses health plan A over health plan B and B over C, it should be the case that the person prefers A to C) there are instances where this is not the case. One possible explanation is that people's decision making machinery is flawed. An alternative possibility is that when people make decisions in experimental settings, they are bringing assumptions and inferences into the process that the researcher is not aware of. This research project explores this possibility. Specifically, the research re-examines experimental violations of three widely accepted principles of rational choice: joint-separate equivalence, independence of irrelevant alternatives, and transitivity. A novel "options-as-information" model is developed, spelling out the conditions under which these three principles are, and are not, applicable to the normative analysis of experimental data. This options-as-information model suggests a revised interpretation of some important findings from the psychological literature: joint-separate reversals traditionally explained in terms of "evaluability", and context effects involving asymmetric dominance. A series of experiments to test the options-as-information explanation of these effects will be conducted. The model also generates new choice problems in which transparently "rational violations of rational norms" (including rational intransitivities) are warranted. These choice problems supply a useful new tool for investigating the benefits and drawbacks of conscious reflection in decision making. A final series of planned studies tests the prediction that, in these choice contexts, conscious deliberation will sometimes lead to systematically poorer decisions.
In terms of broader impacts, the research will help inform efforts to assist people in making better decisions. If observed violations of normative principles of decision making are "biases," then interventions are in order. On the other hand, if they arise because subjects are answering questions different from what the researcher intends, they may not be. Understanding which it the case and when is critical to knowing when interventions to assist people in making decisions are in order and what features of the decision problem those interventions should be sensitive to.
