IFREE-Sponsored Lecture Series at ESI

David Budescu, Ph.D.
Oct. 7th, 2011

Taking Watson to the Market: Studies of the Watson selection task in competitive markets

"One of the most exciting developments in social sciences is the last decades is the breakdown of disciplinary barriers and the increase in inter-disciplinary collaborations, and the cross disciplinary exchange of ideas. One of the best examples of this trend is the work by behavioral and experimental economics and psychologists, like me, who study human judgment and decision making. The work I am presenting started as an attempt to document the commonality of the two approaches and to highlight the difference between them.

We analyze a classical problem in the literature on reasoning – the Wason (card) task – that was designed to study how people reason about conditional inferences of the type if p then q (or, if someone drinks beer in a bar he is, at least 21 years old). This is a notoriously difficult problem (only about 15% of respondents get it right) but performance improves when the problem is tackled by interactive (and cooperating) groups, and this effect carries over to subsequent individual problems. Our work shows that competitive auctions can have similar effects, suggesting that cooperation is not a necessary condition for learning. We also show that the two effects are synergistic: learning in auctions involving teams of traders is faster than in interacting groups and auctions of individual traders."

Bio:
David V. Budescu is the Anne Anastasi Professor of Psychometrics and Quantitative Psychology at Fordham University in New York. He held positions at the University of Illinois and the University of Haifa, and visiting positions at Carnegie Mellon University, University of Gotheborg, the Kellog School at Northwestern University, the Hebrew University and the Israel Institute of Technology (Technion). His research is in the areas of human judgment, individual and group decision making under uncertainty and with incomplete and vague information, and statistics for the behavioral and social sciences. He is on the editorial boards of Applied Psychological Measurement; Decision Analysis, Journal of Behavioral Decision Making; Journal of Mathematical Psychology; Journal of Experimental Psychology: Learning, Memory & Cognition (2000-2003); Multivariate Behavioral Research; Organizational Behavior and Human Decision Processes (1992-2002); and Psychological Methods (1996-2000). He is past president of the Society for Judgment and Decision Making (2000-2001), fellow of the Association for Psychological Science, and an elected member of the Society of Multivariate Experimental Psychologists.

Abstract:
A vast literature shows that individuals frequently violate normative principles in reasoning. In this paper, we report results of a line of studies designed to determine if information dissemination in competitive auctions can reduce, or even eliminate, logical errors in the Wason selection task at the group and individual level. In general, we find that payoff feedback and exposure to the information flow in the market drive the aggregate (and, under certain conditions, the individual) behavior toward the normative solution.

We link our findings to the literature on cooperative groups, which shows that for intellective tasks – with demonstrably correct solutions - groups perform better than individuals. These studies assume that group members share common goals. We extend this research by replacing standard face-to-face group interactions with competitive auctions, allowing for conflicting individual incentives. We demonstrate that competitive combinatorial auctions induce equally impressive learning effects as standard group interactions, and uncover specific and general knowledge transfers from these institutions to new reasoning problems. However markets have a clear advantage in circumstances where the conflicts between the members of the group are made salient. Finally, we combine the advantages of groups and markets and show that teams of traders learn faster and with less feedback. We explain these results within the theoretical framework of collective induction.

Return to the IFREE-ESI Lecture Series