Figure 2

Associations between decisions in different experimental situations in two studies. Comparison of our results (a, b, anonymous interactions) with those of Peysakhovich et al. (2014) (c, d). (a) Correlation matrix for the decisions in the ten situations in our experiment: Dictator Game (DG), Ultimatum Game (UG1, UG2), Trust Game (TG1, TG2), Prisoner’s Dilemma (PD), Coordination Game (CG), Snowdrift Game (SD), Stag Hunt Game (SH), Punishment Game (PG). Heatmap based on Pearson’s correlation coefficients (transformed through catPCA, N = 168). (b) Principal component loadings of the decisions from the catPCA of our experiment (N = 168). Variance explained: Comp 1 = 25.2%; Comp 2 = 15.4%; Total = 40.6%. The risk-taking game (SH) was not significantly loaded on any of the components, while work share (SD) was loaded on both components. (c) Correlation matrix for the decisions in the games in the study of Peysakhovich et al.¹³: Dictator Game (DG), Ultimatum Game (UG2), Trust Game (TG1, TG2), Public Goods Game (PGG), All-pay Auction (AP), 2nd Party Punishment Game (PG); 3^rd Party Punishment (3PP). The games were relabelled, and their order was rearranged to facilitate comparison with the correlation matrix of our study in (a). (d) Factor analysis loadings from the experiment of Peysakhovich et al.¹³. Total variance explained: 79%. The risk-taking game (AP) was not significantly loaded on any of the factors. The difference in axes labelling in (b) and (d) corresponds to the difference in the methods used in our study (categorical Principal Component Analysis) and the study of Peysakhovich et al. (factor analysis). See Supplementary Table S2 for more details on the study of Peysakhovich and colleagues and our relabeling of their types of decisions to facilitate comparison with our study.