Cooperation, Reciprocity, and the Collective-Action Heuristic

Abstract

In laboratory experiments, we manipulate the levels of niceness and reciprocity of seven simulated players in eight-person, iterated social dilemmas. Subjects rationally cooperate most frequently against nice, reciprocal players. However, subjects systematically deviate from optimal responses in intermediate environments that are either nice or reciprocal, but not both. The collective-action heuristic-a simple model of the subject's decision process based on introspection and surprise-driven searchexplains several observed asymmetries of behavior that have important implications for the evolution of cooperation and the theory of social capital: (1) On average, initial cooperators gain a cooperators' advantage over initial defectors due to defectors' inability to take advantage of reciprocal environments; (2) Past experience with reciprocity reduces exploitation even when reciprocity is currently absent, while past experience with nonreciprocity does not ham'per cooperation when reciprocity is currently present; and (3) Institutions that punish noncooperation enhance cooperation by initial defectors, but reduce cooperation by initial cooperators. ollowing the research of Axelrod (1984), Putnam (1993), Taylor (1987), Trivers (1971), and other contributors to the theory of cooperation and social capital, we investigate the role of two characteristics-reciprocity and niceness-generally associated with the development and maintenance of cooperative solutions to social dilemmas. We respond to Ostrom's (1998) request to develop a behavioral theory of collective action by exploring the behavioral relevance of reciprocity and niceness in explaining cooperation from subjects in laboratory collective action experiments. In addition, we examine the interaction between collective action strategies, past experience, and institutions. Our approach to analyzing behavior in collective action dilemmas reflects two general assumptions. First, the collective action strategies of citizens and of experimental subjects are best understood in terms of cognitive heuristics that generate them. In a trivial sense, heuristics can generate any strategy simply by replicating the algorithm that defines the strategy. However, heuristics available to a citizen will reflect the tradeoff between the potential benefits of a repertoire of more complex strategies and the informational and decision-making costs required to apply such a repertoire to appropriate situations (Payne, Bettman, and Johnson 1993). The greater the cognitive costs involved in coping with complex strategic situations, the greater the likelihood that heuristics will produce behavior that differs from predictions based on the analysis of optimal responses. The strategic complexities of collective action dilemmas suggest that such differences will be quite significant, and therefore that optimal responses will be poor predictors of behavior. Thus, the set of heuristics used by citizens will have a profound effect on a society's ability to obtain the potential benefits of cooperation. Second, following the literature on evolutionary psychology (Barkow, Cosmides and Tooby 1992; Caporeal et al. 1989), the set of heuristics in a given society represent specialized cognitive mechanisms for solving social