EFFECTS OF TESTOSTERONE ON AVERSIVE LEARNING: A DOUBLE-BLIND, PLACEBO-CONTROLLED STUDY

Abstract Objective Testosterone, a male sex hormone, is found to influence a broad range of human social behavior, but whether it affects people's pro-social decision-making is unclear. The purpose of this study was to examine the effects of testosterone on pro-social decision-making tendencies. Methods This study used a 2 group (testosterone vs. placebo, between-groups) × 2 recipient (self vs. other, within- groups) experimental design in which 120 healthy male university student subjects were recruited and randomly assigned to either testosterone or placebo groups. Three hours after testosterone or placebo administration, participants were required to perform to a task of avoiding electroshock harms for themselves and others, respectively. The task was a probabilistic learning task in which participants were required to choose between a pair of symbols. Each symbol is associated with a lower or higher probability of getting an electric shock. Results Behavioral analyses in this study were conducted using generalized linear mixed models, and the optimal model revealed a main effect of receiver and number of trials, where participants would have a higher probability of selecting the correct symbol (i.e., the symbol with the lower probability of receiving a shock) when performing the task for themselves, and the probability of selecting the correct option increased with the number of trials, demonstrating a learning trend; at the same time, there was a marginal-significant receiver condition versus number of trials effect, i.e., the probability of participants choosing the correct option when performing the task for another person tended to increase more with the number of trials, and this effect was present in both the testosterone and placebo groups. Further analyses using a positive and negative reinforcement learning model of aversion learning revealed that there were higher learning rates when a better-than- expected outcome was encountered; positive learning rates were lower and negative learning rates were higher in the testosterone group relative to the placebo group, suggesting that the testosterone administration reduced positive bias in decision making. Consistent with the mixed-effects linear modeling results, participants generally learned quicker in the others condition. However, in the receiver-as-self condition, the testosterone group presented more unbiased learning, i.e., lower positive learning rates and higher negative learning rates. The testosterone group focused less on worse-than- expected negative outcomes when avoiding harm for others, i.e., lower negative learning rates. Conclusion Testosterone may influence learning about avoiding harm for oneself, leading to a tendency towards more unbiased learning; and less concern for negative consequences when learning to avoid harm for others.