A model of non-informational preference change

This paper was presented at the Choice Group seminar, LSE, 1/2009, the Preference Change Workshop, LSE, 5/2009, the 3rd Workshop on Decisions, Games and Logic, HEC Lausanne, 6/2009, the 6th Conference of the Society for Economic Design, Maastricht, 6/2009, the Conference on Individual Preferences and Social Outcomes at the University of Maryland, 10/2009, and the Departmental Seminar in Economics, Vanderbilt University, 3/2010.International audienceAccording to standard rational choice theory, as commonly used in political science and economics, an agent’s fundamental preferences are exogenously fixed, and any preference change over decision options is due to Bayesian information learning. Although elegant and parsimonious, such a model fails to account for preference change driven by experiences or psychological changes distinct from information learning. We develop a model of non-informational preference change. Alternatives are modelled as points in some multidimensional space, only some of whose dimensions play a role in shaping the agent’s preferences. Any change in these ‘motivationally salient’ dimensions can change the agent’s preferences. How it does so is described by a new representation theorem. Our model not only captures a wide range of frequently observed phenomena, but also generalizes some standard representations of preferences in political science and economics