Figure 2.

The architecture of the ToM-SNN model. The ToM-SNN model comprises the perspective taking module, the policy inference module, the action prediction module, and the state evaluation module, which are inspired by the TPJ, the vmPFC, the dlPFC, and the ACC, respectively. The perspective taking module parses current observation to form predictions about other's observation, ${ô}_{other}^t$. The policy inference module parses current observation and past observation to infer other's behavior style, ${\widehat{bs}}_{other}$. The action prediction module parses predictions about other's observation and behavior styles to form other's predicted action, â_other. When combining predictions about other's actions with environmental information, the agent can get other's next state, ŝ_other. The state evaluation module parses other's next state to form state value, ${\widehat{V}}_{other}$. The arrow in the figure indicates information transmission and does not involve the learning of network weight.