Measuring the cooperativeness of behavior in quantity variation duopoly games

Abstract

An economic game (Cournot quantity variation duopoly) was employed to create a laboratory market environment in which there were elements of cooperation and competition. Paper profits (scores) were received by players in repeated play of a game which can be defined as a nonzero-sum two-person game with no preplay communication. Each of 12 subjects was matched simultaneously with two robots, one which was programmed to respond cooperatively and one which was programmed to respond noncooperatively toward moves made by the subject. Subjects believed they were matched against other human participants in groups of three. Indices of cooperativeness were defined within the context of the economic theory of the game in such a way that the differential cooperativeness of a subject toward the two robots could be measured dynamically over the path of the game. A two-part hypothesis was formulated which survived statistical tests. It was concluded that: 1) in experimental Cournot markets in which human subjects play with behaviorally stable robots there is a tendency toward stability of behavior in the human players, and 2) there is a tendency for subjects to be more cooperative with cooperative robots than they are with noncooperative robots.