|Ph.D Student||Michael Sobolev|
|Subject||On Learning Traps and Punishment|
|Department||Department of Industrial Engineering and Management||Supervisor||Full Professor Erev Ido|
|Full Thesis text|
Punishment was, and still is, one of the most studied topics in social and behavioral sciences. In this work, we build on recent studies of decisions from experience in an attempt to clarify the conditions that trigger deviations from effective use of punishments, and determine the impact of information concerning the punishments suffered by others. First, we focus on the overuse of punishment in education, management and personal relationships. We find that the overuse of punishment is most likely when punishments are counterproductive on average but usually effective. Then, we extend our analysis to situations that include excessive and insufficient use of punishment by managers. Our findings imply that both deviations from the optimal managerial punishment policy can be described with the assumption that the decisions to punish reflect reliance on small samples of past experiences. Finally, we explore the indirect effect of punishment. We find that when the participants had limited knowledge on the possible strategies, learning about the punishments (or losses) suffered by others increased the rate of counterproductive risky choices. The effect was particularly strong when the negative outcome of the bad strategies was rare. In this work, we generalize models of decisions from experience to account for the experimental results. We discuss the implications of our findings to the design of effective interventions and useful policy.