Economics Seminar - Yusufcan Demirkan

I propose a procedural model of random choice called visual choice based on the eye movements of individuals. The decision maker fixates randomly on alternatives, and performs saccades to other alternatives based on the fixation point. The choice probabilities are determined using a random tie-breaking between these saccaded alternatives and weighting according to the fixation probabilities, both formulated using the well-known Luce rule. I discuss the cases of fixation-independent and fixation-dependent saccades. The former connects to the recent literature extending the Luce rule giving a procedural interpretation for these choice rules. For the latter, I discuss a special case called satisficing visual choice which determines the choice probability of an alternative x by calculating the weighted average of the contribution coming from the fixated alternatives strictly dominated by x. The main results of the model are related to the rationality properties of this procedure, and a behavioral characterization result based on conditions imposed on the choice data.