This paper investigates a model of preference evolution in which agents are paired to play a sequential prisoner’s dilemma under incomplete information. The population consists of two types of agents: conditional cooperators and defectors. We demonstrate that cooperation can be sustained when the first mover predicts the second mover’s behavior by projecting themselves into the second mover’s role, following the principles of social projection theory. However, when the first mover uses theory of mind to anticipate that different second movers may act differently, cooperation breaks down.
Jiabin Wu (Thu,) studied this question.