Abstract: Decision making is a cognitive process during which individuals make a choice after considering all factors in all alternatives. The development of decision theory has gone through several stages: based on expected value; expected utility theory and prospect theory. In the field of psychology, there is increasing interest in the effects of hormones on human decision-making. Testosterone is one of the steroid hormones which plays an important role in human behavior. There is a positive association between testosterone level and risk-seeking behavior in economic decision-making, but the evidence is mixed. In the present study, we aimed to investigate how testosterone affects risk-taking behavior and the computation of value based on the cumulative prospect theory framework. One-hundred and twenty healthy male participants were recruited in our study and engaged in a double-blind, placebo-controlled, between-participant design. Participants in the experimental group received a single dose of 150mg testosterone gel, while the control group received a colorless hydroalcoholic gel (placebo). The decision-making task commenced 3h post-dosing to ensure the effect of testosterone reached peak level. We used a dynamic experiment for estimating preference (DEEP) task consisted of 60 trials where participants made choices between two risky options presented on the screen to measure their risk preference adaptively. Using the hierarchical Bayesian estimation, we fit the data of the experimental group and the control group separately to generate the three characteristic parameters {α,σ,λ} (utility curvature, σ; probability distortion, α; and loss aversion, λ). We also calculated: 1. the proportion of choosing risky option (the options with higher variance); 2. the proportion of choosing options with higher expected value. The participants from control groups showed the tendency of risk seeking as their proportion of choosing risk options were higher than chance level (i.e. 50%); but there was no statistical difference between testosterone and placebo groups. Both groups were more likely to choose higher expected value options, though there was no group difference. For the characteristic parameters, testosterone group was less distorted in probability weighting function (α), suggesting their subjective probability was closer to the actual probability compared to the control group; Testosterone group showed stronger loss aversion (λ) than control group, indicating they were more sensitive to loss and were more likely to avoid loss; Both groups demonstrated identical pattern of utility curvature(σ), suggesting that they were equally sensitive to rewards. Besides, we conducted the model comparison between models based on prospect theory and expected utility theory respectively. Result showed that both groups were better explained by prospect theory model. For the present study, testosterone was not associated with risk taking based on the proportion of choosing risk options. The possible explanations are: 1. Testosterone doesn’t affect risk taking in isolation, it may interact with other factors (sex, cortisol, estradiol, etc.) 2. Different tasks with different decision theory (or paradigms) may lead to inconsistent results. However, it is clear that testosterone influences the computation of value. The neural circuit of reward consist of OFC, ACC, Amygdala, VS and NAcc, testosterone may affect this neural circuit to interfere with value processing. Future studies are needed to investigate the neural mechanism by which testosterone influences risk preference and value processing by using fMRI.

Key words: Testosterone, hormones, decision making, value computation , cumulative prospect theory

摘要： 较多研究支持睾酮和决策中的风险寻求行为呈正相关，但是也有其他的研究未能发现这种关系。基于决策的累计前景理论，本研究使用动态估计参数估计任务(DEEP)，结合计算模型的方法，对120名双盲给药、有安慰剂对照的被试进行睾酮对价值加工过程作用的探究。结果显示，睾酮减少了个体的概率扭曲以及增加了损失规避，但是没有引发明显的风险寻求行为，研究结果表明睾酮对个体的价值加工过程产生了影响，使个体对概率的感知更接近于客观值并且增加了对损失的敏感性。

关键词: 睾酮, 荷尔蒙, 决策, 价值加工, 累积前景理论