Experiments with N-Person Social Traps I

Abstract

Subjects for single play experiments with n-person social traps (strong and weak versions of Prisoner's Dilemma, Volunteer's Dilemma, Largest Number) were recruited from several populations: students, professionals, business people, employees, and foreign visitors. Performances were compared across the subject pools, across the games, and across instructions. The results indicated that subjects receiving full instructions, including warnings about the social traps, avoid the social traps more frequently than those receiving minimal instructions. However, the differences in frequencies of cooperative choices were all slight except for the strong version of Prisoner's Dilemma. Fewest cooperative choices were observed in business people, the most in foreign visitors. However, the latter effect may have been an artifact due to self-selection. Among the games, Largest Number is distinguished by possessing no individually rational equilibrium. The cooperative solution of this game, however, is salient, namely, for each to name the smallest number. In spite of this salience, the smallest frequencies of cooperative choices were observed in this game, and full instructions did not significantly increase the frequency. When the subjects had an opportunity to organize five-person cooperative groups, each of which had a good chance of winning $1000, only 30 of 60 subjects formed such groups.