How do humans learn about other people? Incorporating social knowledge structures into reinforcement learning
To navigate social interactions successfully, humans need to continuously learn about the personality traits of other people (e.g., how helpful or aggressive is the other person?). However, formal models that capture such complex social learning processes are currently lacking. In this study, we specified and tested potential strategies that humans could employ for learning about others. Standard reinforcement learning (RL) models only capture part of the learning process because they neglect inherent knowledge structures and omit previously acquired knowledge. We therefore formalized two social knowledge structures and implemented them in novel hybrid RL models to test their usefulness across different social learning tasks. We named these concepts granularity (knowledge structures about personality traits that can be utilized at different levels of detail during learning) and reference points (previous knowledge formalized into representations of average people within a social group). In five behavioural experiments, results indicated that participants combined the concepts of granularity and reference points in a rather optimal fashion—with the specific optimal combinations in models depending on the people and trait items that participants learned about. Overall, our experiments demonstrate that variants of RL algorithms, which incorporate social knowledge structures, describe crucial aspects of the dynamics at play when people interact with each other.