Social interactions are dynamic, requiring individuals to understand not only how their behaviour may affect others but also how others may respond in return. In humans, social interactions are also often dominated by processes such as language and theory of mind which allow us to communicate complex thoughts and beliefs.

Understanding the basic neural processes that underlie social behaviour or by which individuals communicate has remained a challenge. Here, we develop naturalistic approaches in animals and humans to begin investigating these questions at cellular scale. First, by developing an ethologically based group task in three-interacting rhesus macaques, we identify detailed representations of other’s behaviour by neurons in the prefrontal cortex, reflecting the other’s identities, their interactions, social context, actions, and outcomes. We also show how these cells collectively represent the interaction between specific group members and how they enable mutually beneficial social behaviour. Second, by recording from neurons in the human prefrontal cortex during language-based tasks, we identify neurons that reliably encode information about others’ beliefs across richly varying scenarios and that distinguish self- from other-belief-related representations. By further following their encoding dynamics, we also show how these cells represent the contents of the others’ beliefs and accurately predict whether they are true or false. Finally, we show how these cell ensembles track linguistic information during natural speech processing and how language can be used to ask specific questions about the neuronal constructs that underlie social reasoning. Together, these studies reveal detailed cellular mechanisms for interactive social behaviour in animals and humans and highlight the prospective use of naturalistic approaches in social neuroscience.