Trust and Reciprocity is an exciting compilation of cutting-edge experimental research into the bases of social cooperation. The editors and contributors are prominent and distinguished contributors to experimental and interdisciplinary approaches to modeling cooperation.

The editors quickly identify the central problem of cooperation: while all gain from cooperating, each individual has an incentive to free-ride on the good will of the others. No cooperation takes place unless the free rider problem is successfully addressed. Ch. 2 presents compelling evidence that face-to-face communication alone is capable of sustaining cooperation in small groups, especially when the task is repeated several times. Moreover, if experimental subjects are allowed to fine shirkers, at a cost to themselves, they do so, despite that fact that purely self-interested subjects would never engage in such activity.

There is a fine computer simulation (Ch. 7) that beautifully illustrates the capacity of agent-based modeling to generate interesting hypotheses concerning human behavior--in this case, the synergistic relationship between cooperation and conflict. The experimental chapters of of uniformly high quality, and quite illuminating, including a chapter on the development of trust and reciprocity in children.

The theory side of the book is less successful. Several authors say that the experimental results "contradict game theory." This is simply incorrect. Like mathematics, game theory can be more or less useful in modeling a phenomenon, but it cannot be "wrong." Indeed, all of the experimental results are based squarely on game theory as a tool of experimental design.

More important, despite ample evidence that trust and reciprocity cannot be explained assuming selfish individuals, the theorists in the book attempt repeatedly to assert just that. According to the authors, altruistic behavior is caused by mistakes in reasoning! The major theoretical chapter in the book (Ch. 4), for instance, is an exposition of Robert Trivers' concept of "reciprocal altruism," which is just long-term self-interest. The experimental results on reciprocity are incompatible with this concept, unless one assumes that subjects simply are ignorant and mistake-prone in their behavior. Numerous experiments show that this is decidedly not the case. In Chapter 6, we are told that subjects are only altruistic because they are afraid that the conditions of anonymity in the experiment are highly imperfect--despite the fact that the evidence for this is extremely weak (one study done in 1994, which has not been replicated, using a dictator game, which is not a social interaction or a social dilemma), and the evidence against which is virtually overwhelming. Later in the chapter we are told (p. 162) that people cooperate in one-shot games because they confuse them with repeated games, with which they are more familiar. This argument is specious, however, because many experiments reveal that people are quite capable of distinguishing between one-shots and repeated games, and cooperate much more in the latter than in the former.

So this book will not win any awards for social theory, but it is deep and rich with highly insightful descriptions and analysis of human behavior under conditions where trust and reciprocity increase well being. The Russell Sage foundation, which underwrote the research and preparation of the volume, is to be commended for a job well done.