中国科学院行为科学重点实验室知识库

摘要：通常人们在做信任决策时，为了降低风险会考虑他人的信誉信息。已知人们对于他人信誉先验的形成方式可以分成两种：通过与对方进行多次交易，观察学习得到(interaction-based)；外界给出的、不需要学习就可以获得的(reputation-based)。但还不清楚这两种形成信誉先验的方式对人们信任决策的影响是否一致。为探讨这一问题，本研究纳入了33名大学生(年龄18.82 ±2.13)完成了改进的、重复信任博弈(iterative trust game)范式。在该实验范式中，我们将对方的信誉量化为五个可信度水平（一星/三星/五星/七星/九星），星级越高代表对方越可信；将从外界获得对方信誉先验的方式设置为有提示的条件（实验开始前提前告知对方的星级水平），将通过观察学习获得对方信誉先验的方式设置为无提示的条件（不告知对方的星级水平，在实验结束后需要评价对方的星级水平）。每名参与者需要作为投资者（investor）分别在有提示和无提示两种条件下，与不同星级水平的被信任者（trustee）在该范式中重复20次的信任博弈。通过重复测量分析及其事后检验，我们发现：有无提示被信任者信誉先验会显著的影响被试的投资次数（p<0.001;η2=0.94），在无提示的条件下，被试投资给对方的次数更多；面对不同可信度水平的被信任者，投资的次数有显著性差异（p=0.002;η2=0.25）,被试更倾向于投资给可信度水平高的被信任者；被信任者可信度水平与有无提示的交互作用显著（p<0.001;η2=0.69），在有对方先验信息的情况下，被试对于可信度水平高（七星/九星）的被信任者投资的次数更多，而对可信度水平低（一星/三星/五星）的被信任者投资的次数更低。本研究发现信誉先验的形成方式对于人们信任决策行为有不同影响，并依赖于被信任者的可信度水平。这一研究结果补充了影响信任的因素相关方面的研究，同时为进一步探索精神分裂症、抑郁症等群体的不良社会适应能力的潜在行为标记提供了新的角度。

Abstract：People often consider the trustworthiness of others to reduce the risk when they make the trust decision. There are two kinds of ways to form the trustworthiness of others:one is to observe and to learn through multiple transactions with others and thus is called interaction-based; the other is reputation-based in which the trustworthiness of others is explicitly given. In order to study whether these two ways has same influence on the trust decision, the present study investigated the behaviors of 33 undergraduate students (age: 18.82 ±2.13) being investor in a modified iterative trust game. In this game, we quantified the trustworthiness of others into five levels (Star one/ Star Three/ Star Five/ Star Seven/ Star Nine). The higher the star rating is, the level of trustworthiness of others is higher. Besides, we set two conditions for the ways to form the trustworthiness of others: for the reputation-based way, we provided the star rating of others for participants before the experiment (with reputation prior condition); for the interaction-based way, we asked the participants to rate the star rating after they finished the game (without reputation prior condition). Each participant needed to play trust game with each trustee who had different levels of trustworthiness under with/without reputation prior conditions for 20 times. Using RMANOVA and its post test, we found that the conditions of with/without prior significantly influenced the amount that investor invested to the trustee（p<0.001;η2=0.94）,and the investor invested more to the trustee under the without reputation prior condition. There were significant differences in the amount that the investor invested to the trustee with different levels of trustworthiness（p=0.002;η2=0.25）,and the investor was inclined to invest to the trustee with higher level trustworthiness. The interaction between the level of trustworthiness and condition was also significant（p<0.001;η2=0.69）.Specifically, under the condition of with reputation prior, investor tended to invest to the trustee with higher level of trustworthiness (Star Seven/ Star Nine) and invested lesser to the trustee with lower level of trustworthiness (Star one/ Star Three/ Star Five). The current study suggested thatthe way for investor to form the prior-reputation of trustee generated different influences on the trust decision, which depended on the level of trustworthiness of trustee. This finding improves our understanding on the factors affecting trust and provides a new perspective for further exploration of the potential behavioral markers for the poor social adaptability of patients with psychiatric disorders.