Figure 3

Effect of the interaction partner’s facial appearance on prosocial behaviour. (a) Principal Component Analysis of the eight facial attribute scores. PC1 (variance explained: 43.6%): happy, generous, trustworthy, not angry, not greedy; PC2 (variance explained: 23.4%): risk-taking, attractive, not rational. (b) Association between the subjects’ facial appearance (columns) and the behaviour elicited in their interaction partners in seven decision situations (rows). Facial appearance is quantified by the face judgement scores regarding eight attributes (attractive, happy, generous, trustworthy, rational, risk-taking, greedy, angry) and the scores for their principal components (PC1, PC2). The heat map gives a pictorial representation of the R² values for positive and negative associations, based on linear regression models with a single predictor (see “Methods” for details on statistics, and Suppl. Table S5 for numerical values). Significant associations: *p < 0.05, **p < 0.01. The decision situations related to fairness (UG1), demand (UG2), and punishment (PG) are not shown, as the behaviour in these situations was not significantly associated with any of the facial scores (see Suppl. Table S5).