| Formal models of political behavior almost invariably employ the orthodox rational actor paradigm. However, a growing body of experimental evidence indicates that each of the central pillars of rational choice theory---maximization of expected utility, Bayesian updating of beliefs, and an invariance to framing effects---is violated frequently and systematically. Existing theoretical work covering such instances of "bounded rationality" almost always assumes that "cognitive costs" are responsible, in other words, that people do as well as they can with the limited attention and mental capabilities they can muster.; This dissertation adopts another approach: to allow elements of choice theory to emerge endogenously in the equilibria of evolutionary games rather than to assume orthodox rationality, as is the common practice. The dissertation consists of three papers, each of which addresses one of the pillars of the rational choice worldview. Each paper presents a model of preference formation in an evolutionary game, and demonstrates that the pillar in question should be expected to topple under particular circumstances---even in the absence of cognitive costs. Thus, the dissertation employs a unified approach to the formal study of behavioral phenomena. There are also binding themes in the structures of the different preference formation models that are studied: coordination and competition, two of the primal themes in the study of politics. Two of the three papers, on non-Bayesian belief formation and framing effects, demonstrate that deviations from classical rationality can be beneficial for actors who are playing coordination games. All three of the papers demonstrate that further deviations can be expected in competitive settings. |
