Collusion in the Prisoner's Dilemma: Number of Strategies

This experiment is an attempt to determine how behavior is affected when a Prisoner's Dilemma is embedded in a larger decision framework, as is true of many real situations. Twelve pairs of students were confronted with a thirty-price choice matrix. Monetary rewards were used and the situation was couched in terms of two firms in a market. As a control group, eleven pairs of subjects were confronted with a two-choice Prisoner's Dilemma (within the same experimental context). We observed a large difference in payoff between two sessions where the large matrix was run. Although the most obvious explanation was the sex mix of subjects, further investigation seemed to indicate that major field provided a better explanation. Perhaps this result carries over to other attempts to relate PD behavior to sex. It was predicted that (1) there would be more cooperation in the 2× 2 matrix, (2) that cooperators in the large matrix would be spread throughout the Prisoner's Dilemma range, (3) that the larger number of strategies would lead to a more unequal split of payoffs and less stable cooperation, and (4) that subjects would be spread across the entire PD range in the large matrix. All these hypotheses were corroborated by the data, although results were not always significant.