James E. Corter Teachers College, Columbia University
Single, Multiple, and Repeated Decisions: Why They are Different

Since Samuelson (1963), it has been assumed in the decision literatures that people become more rational when they make multiple decisions. The empirical evidence on this proposition has been mixed, but recent behavioral decision research has begun to focus on the question of when (repeated) experience with outcome feedback in decisions under uncertainty leads to better or more "rational" decisions. In one set of experiments (Chen & Corter, 2006), we found that making multiple prospective decisions under risk did not lead to increased rationality, and in fact revealed a new type of irrational behavior in such tasks -- a preference for mixed options. In a related set of experiments, and in new modeling work, we contrast behavior in prospective risky decisions with repeated risky decisions, in both cases where the probabilities of outcomes are known in advance. Initial behavior was not more rational, but rationality increased across trials, and even generalized across a shift in domain (from gains to losses or vice-versa). A follow-up study, trial level sequential analyses, and a modeling exercise were conducted to explore mechanisms underlying the effects of experience. In the modeling exercise, the best fit to the data was obtained for a model that allows for experience-based changes in the shape of the decision weight function over repeated trials of Loss problems, towards more linear decision weights.