Psychological Economics in the Late 1990s.(Review)

Games and Human Behavior: Essays in Honor of Amnon Rapoport Edited by David V. Budescu, Ido Erev, and Rami Zwick. Based on a Chapel Hill, NC, workshop, August 1996. Hillsdale, NJ: Erlbaum, 1999. xiii + 427 pp. Paper, $39.95.

Alter decades of splendid isolation from its sister social sciences, economics is beginning to engage meaningfully again with cognitive and social psychology. The current volume, a festschrift for Amnon Rapoport, is a landmark on the path of reengagement. A special session of the summer 1996 Mathematical Psychology workshop gathered an interdisciplinary group of researchers at Chapel Hill, North Carolina, to present papers in honor of Rapoport's 60th birthday, and the published collection of 16 chapters (with 25 authors) grew out of the workshop papers.

Progress since the last comparable collection, Hogarth and Reder (1987), is remarkable. That earlier collection revealed a vast methodological gulf between economists and psychologists with a common interest in bargaining and individual choice. Economists insisted on explicit theory centered on rational choice, and they interpreted experiments strictly through that theoretical lens. Psychologists were more casual about how to construct theories and more pragmatic about interpreting experimental data. These differences induced disagreement even on what sort of experiments are informative. Although the arguments of that era were entertaining and occasionally enlightening (I especially favor Vernon Smith's [1991] contribution highlighting the pivotal role of economic institutions), the gulf seemed too wide to allow much interdisciplinary progress.

The present book reveals a very different and more progressive landscape. Economists wisely continue to rely on explicit theory to structure and interpret the data, but have relaxed their insistence on narrow self-interest as the sole motivation for human behavior. Psychologists, if this book is any indication, have become more appreciative of models constructed from widely accepted first principles, and it is no longer difficult to conduct experiments that everyone recognizes as informative. The book reveals only tentative answers to the most interesting questions, but there is a new sense of common purpose and progress. For this, Amnon Rapoport deserves a portion of the credit, both for the example he sets and the students and coauthors he has influenced, as well as for his direct scientific contributions.

The collection has five parts, each with distinctive questions. Part I, "On Psychology and Economics," has a long overview chapter by the three editors and a brief methodological note by Robin Dawes on demand effects. The editors' overview expresses their hope for "a new common core of axioms and unified methodology" (p. 5) and offers several helpful suggestions: that scarcity in economic-style theoretical models include both cognitive and environmental resources, that preferences are sometimes constructed or revised during the decision process, that they are importantly affected by learning and adapting, that the clarity of salient cash payments sometimes makes them useful for psychologists as well as economists, and that economists should consider motives beyond self-interest.

Part II offers a snapshot of the very active subfield known as learning in games, which recently has entered the mainstream of economics and game theory (Friedman, 1998). Three leading approaches are showcased here. Chapter 4, by Ido Erev and Al Roth, offers an expansive version of reinforcement learning, in which rules (or cognitive strategies, in the authors' terminology) rather than direct actions are the objects of reinforcement and learning. Colin Camerer and Teck-Hua Ho in chapter 3 summarize their model of learning responsive to the potential payoff of actions not taken as well as the actual payoff reinforcement from chosen actions, thus nesting belief learning and reinforcement learning. Reinhard Selten and Joachim Buchta in chapter 5 at last publish their paper introducing directional learning, a qualitative theory of learning a continuous action. The idea is that one tends to increase (or decrease) the action following trials when such an adjustment would have increased payoff. It will probably tak e a few more years before a clear consensus emerges on which learning models are most appropriate, but my guess at the moment is that some sort of rule learning model will prevail.

Part III studies equilibrium and learning in coordination games--those with several noncooperative equilibria, some of which are efficient. Rosemarie Nagel in chapter 6 surveys the cottage industry she created in ...