Fig. 2: Differences in the number of correct choices.

A Three-way interaction of the factors treatment × recipient × visibility. Participants in the placebo group behaved more prosocially (as captured by the prosociality index = correct choices for other – correct choices for self) when being observed than in privacy. Exogenous testosterone eliminated this audience effect. Note that the analyses were performed with raw data in the 2 × 2 × 2 factorial design; the plotted difference score (other minus self) is to improve readability and interpretability. Breaking down the three-way interaction by using treatment contrasts showed that there was no significant effect of the drug treatment and visibility factors, or their interaction, on the number of correct choices made for oneself (B). On the other hand, testosterone, compared to placebo, decreased the number of correct choices made for the NGO when being observed (C). Dots represent the data of individual participants, lines represent mean values per group, and boxes 95% confidence intervals.