We use linear regression to estimate the players' strategies based on the co-player’s behavior in the previous round (Table 1). Here, we visualize the resulting conditional cooperation probabilities. C and D represent the decision to cooperate or to defect, respectively. In the same-partner treatment, participants link their behavior across games. As a result, a player’s cooperation probability depends on the previous outcome of both games. In comparison, behaviors in the different-partners treatment are largely independent across the two games.
