ISSN 0439-755X
CN 11-1911/B

心理学报 ›› 2023, Vol. 55 ›› Issue (7): 1099-1114.doi: 10.3724/SP.J.1041.2023.01099

• 研究报告 • 上一篇    下一篇


袁博(), 王晓萍, 尹军, 李伟强()   

  1. 宁波大学心理学系暨研究所, 浙江 宁波 315211
  • 收稿日期:2021-12-30 发布日期:2023-04-21 出版日期:2023-07-25
  • 通讯作者: 袁博, E-mail:;李伟强, E-mail:
  • 基金资助:

The role of cross-situational stimulus generalization in the formation of trust towards face: A perspective based on direct and observational learning

YUAN Bo(), WANG Xiaoping, YIN Jun, LI Weiqiang()   

  1. Department of Psychology, Ningbo University, Ningbo 315211, China
  • Received:2021-12-30 Online:2023-04-21 Published:2023-07-25


基于联结学习理论(associative learning theory), 通过3项实验考察了跨情境(公平−信任)的刺激泛化在面孔信任形成中的作用。实验1a和实验1b分别从直接互动和观察学习视角, 发现了面孔信任形成中跨情境的刺激泛化效应(stimulus generalization effect), 即相比于中等不公平条件, 随着被信任者的面孔与先前互动中公平(不公平)分配者面孔相似度的增加, 个体对其信任程度逐渐增加(降低); 并且这一效应具有不对称性(asymmetry), 对不公平分配者面孔的泛化强度高于对公平分配者面孔的泛化强度。采用漂移扩散模型(Drift-Diffusion Modeling, DDM)分析发现, 不公平条件下的漂移率v显著小于中等不公平或公平条件下的漂移率v, 且大多分布在小于0区间; 表明在对与先前互动中不公平分配者面孔相似的陌生面孔进行信任决策时, 个体更倾向累积不信任的证据。实验2结果发现, 行为意图在刺激泛化效应的产生中起到调节作用; 在无意图条件下, 上述跨情境的刺激泛化效应消失。上述结果表明, 个体采用联结学习机制将不同情境中习得的刺激价值联结泛化到新的互动情境中, 进而指导随后的信任决策。

关键词: 信任形成, 联结学习, 刺激泛化, 行为意图, 漂移扩散模型


How do humans learn to trust unfamiliar others? Decisions in the absence of direct knowledge rely on our ability to generalize from past experiences and are often shaped by the degree of similarity between prior experience and novel situations. A previous study suggested that people use stimulus generalization from the same situation as a mechanism for learning to trust towards strangers. However, it is still unclear whether this stimulus generalization effect exists across different situations, and the role of intention perception in this effect. Here, we leverage a stimulus generalization framework to examine how perceptual similarity between known individuals and unfamiliar strangers across different interactive situations shapes people’s trust towards strangers. Given that the strong adaptability of the stimulus generalization mechanism, we assume that the faces associated with different degrees of unfairness will affect the individual's trust towards similar unfamiliar faces, and intention perception modulates this process.

Three experiments were conducted to examine the above hypothesis. In Experiment 1a and Experiment 1b, participants play or observe an iterative ultimatum game with three partners who exhibit highly unfair, medium unfair, or highly fair behavior. After learning who was the fair/unfair allocator, participants select new partners for a trust game. Unbeknownst to participants, each potential new partner was parametrically morphed with one of the three original players. In Experiment 2, participants play a similar iterative ultimatum game with three partners, nevertheless the allocations were generated by a computer algorithm which excludes the intention of the allocator.

A mixed linear regression was conducted, with both (un)fairness type (whether faces were morphed with the original fair, medium unfair, unfair allocator’ face) and perceptual similarity (increasing similarity to the original face, 23%, 34%, 45%, 56%, 67%, 78%) were entered as predictors of choosing to play with the morphed face. The result of Experiment 1a and Experiment 1b show that compared with the medium unfair condition, as the perceptual similarity between the morphed trustee’s face and the face of the fair (unfair) allocator in the previous interaction increases, the degree of trust (distrust) towards the trustee gradually increases. In addition, this effect is asymmetrical, participants preferentially avoided more the unfair morphs in comparison with the fair morphs. This suggests an asymmetric overgeneralization toward individuals perceived to be morally aversive. Using Drift-Diffusion Modeling (DDM), we found that the drift rate ν under unfair condition is significantly smaller than that under medium unfair or fair conditions, and most of them are in the range of less than 0. This suggests that individuals are more likely to accumulate evidence of distrust when making trust decisions about unfamiliar faces that are similar to the allocator who was unfair in previous interactions. In Experiment 2, under an unintentional situation, the above-mentioned cross-situational generalization effect disappeared.

Together, our results demonstrate that the individuals use the associative learning mechanism to capture the moral information of the interactive objects from the past experience, and then guides subsequent trust decision- making. This mechanism draws on prior learning to reduce the uncertainty associated with strangers, ultimately facilitating potentially adaptive decisions to trust, or withhold trust from unfamiliar others.

Key words: trust formation, associative learning, stimulus generalization, behavioral intention, Drift-Diffusion Modeling