Epiphany: A Search Theoretic Approach to Modelling Boundedly Rational Agents

In this paper we use search theory to model the decision making process of boundedly rational agents. In the canonical (sequential) search approach, each decision maker is deemed to acquire new information, at random intervals of time, regarding the external environment. In the simplest of such settings, the individual's behavior is completely characterized by an optimal stopping rule in which the search process terminates as soon as information is revealed that indicates that the agent's payoff is expected to exceed his reservation value. In our approach, the individual cannot "see" all of the actions that are available to him at a given point in time. However, through search he can learn of some of the courses of actions available to him. In this sense, the individual is conceived of as introspectively "searching for the right idea." Furthermore, in our framework the agent's optimal course of action is decided in an epiphany. Thus, action takes place as soon as an idea generates an (expected) payoff in excess of his reservation value. The paper explores several applications of the modelling framework. The approach makes precise, albeit in a simple way, notions of intelligence. Thus, an intelligent player is one who generates many (and better) ideas in a given interval of time. We apply the model to better understand: (i) advertising (ii) incomplete contracts and (iii) authority relations