Absent-minded drivers in the lab: Testing Gilboa's model

This note contributes to the discussion of decision problems with imperfect recall from an empirical point of view. We argue that, using standard methods of experimental economics, it is impossible to induce (or control for) absent-mindedness of subjects. Nevertheless, it is possible to test Gilboa's (1997) agent-based approach to games with imperfect recall. We implement his model of the absent-minded driver problem in an experiment and find, if subjects are repeatedly randomly rematched, strong support for the equilibrium prediction which coincides with Piccione and Rubinstein's (1997) ex ante solution of the driver's problem.