Fig. 2: Cooperation varies with payoff values, other-regarding preferences, and option positions.

a Link between payoff values and cooperation. Average cooperation rate across trials and participants for each possible value of the normalized payoffs (payoffs divided by the sum of all payoffs on a given trial). Each dot represents one normalized payoff value, the lines are fitted regressions of average cooperation rates on normalized payoffs. The beta and p-values correspond to fixed-effect coefficients of mixed-effects logistic regressions of payoffs on trial-by-trial cooperation (See Supplementary Table 1). N = 88 participants. b Other-regarding preference (\({{{\rm{\alpha }}}}\)) measured using the choice model. Each dot represents the fitted parameter of one participant plotted against the average cooperation rate of the same participant. The line and coefficients represent the fitted regression between \({{{\rm{\alpha }}}}\) and cooperation (r: correlation coefficient, p: p-value). N = 66 participants. c Cooperation rate plotted against average relative time spent sampling each AOI (time spent looking at the AOI divided by trial duration). Each dot represents a participant, and the lines and coefficients represent fitted regressions of cooperation rate on average relative sampling time, showing higher cooperation rates for participants who sample others’ payoffs longer and lower cooperation rates for participants focus on self-payoffs. N = 88 participants. d Effect of the positions of the rows and columns on cooperation rate. Each dot represents the cooperation rate of one participant, the black error bars represent the average rate across participants and the standard error of the mean (s.e.m). The stars are p-values of two-tailed t-tests between the participant’s cooperation rates for the different positions: p < 0.017 (*) or non-significant (n.s.), using a Bonferroni correction. N = 88 participants. e Fitted model parameters across the four different positions. Each dot represents the difference between the position-specific parameter and the parameter averaged across all positions of one participant, and the error bars represent the mean across participants and s.e.m. Stars represent p-values of two-tailed t-test inferior to 0.017(*) or non-significant (n.s.), applying a Bonferroni correction. N = 66 participants.