This paper provides experimental evidence that biased perception of probabilities affects cooperation levels in repeated games. Subjects completed a prisoner’s dilemma game that continued with a probability. Constant discounted utility (DU) interprets such a probability as time discounting. Probability biases lead to deviations from DU (cf. Halevy 2008), which affects outcomes evaluation. The hypothesis is that biases affect cooperation. Using an incentive-compatible tool, I quantified the biases. About 53% of subjects are expected utility (EU); 47% are prospect theory: 21% overestimate small probabilities and underestimate large probabilities (inverse-S) and 26% underestimate small probabilities and overestimate large probabilities (S-shape). Results confirm that inverse-S (S-shape) subjects cooperate more (less) than EU ones. I explain this result by adopting Halevy's impatience index.

Invitation Link