The Group as the Container of Generalized Reciprocity

Abstract

The results of an experiment with 91 Japanese participants confirmed the hypothesis that players of a Prisoner's Dilemma game would cooperate more with an in-group member than with an out-group member in the simultaneous game but not in the sequential game. The game used in the experiment was constructed such that each player gave his or her partner a portion of his or her own endowment of 300 yen. The group membership was created on the basis of participant's preferences for Klee's or Kandinski's paintings; each participant played the game once with an in-group member and once with an out-group member. In the simultaneous game, the two players decided simultaneously how much to give to the partner without knowing what the partner would do. In the sequential game, the first player made the decision; then the second player followed with full information on the first player's behavior. All the participants in the sequential game in fact took the role of the first player. The differential effect of the partner's group membership had been predicted on the basis of Yamagishi and his colleagues' argument that expectations of generalized reciprocity from in-group members is the source of in-group favoritism in a minimal group. A group is clearly distinguished from a mere aggregate of people. People who sit in the waiting area of an airport, for example, hardly constitute a group, nor do all people who wear eyeglasses. They share certain characteristics such as sitting in the same place or wearing eyeglasses, but this does not make them a group. The defining feature