Cooperation in public goods games predicts behaviour in incentive-matched binary dilemmas: evidence for stable pro-sociality

We report the results of an experiment in which subjects completed second mover public goods game tasks and second mover binary social dilemma tasks. Each task was completed under three different incentive structures which were matched across tasks. The use of nonlinear incentive structures, along with a novel categorization method, allowed us to identify behavioral subtypes that cannot be distinguished using conventional linear incentive structures. We also examined how well behavior could be predicted across tasks. Subjects’ average conditional cooperation levels showed significant cross-task predictability and stability. However, almost a third of responses (28%) demonstrated unambiguous preference reversals across tasks. We argue that pro-sociality is best described as an individual-level trait, similar to risk aversion in choice under risk.