Operationalizing Theories of Theory of Mind: A Survey

Human social interaction hinges on the ability to interpret and predict the actions of others. The most valuable explanatory variable of these actions, more important than environmental or social factors, is the one that we do not have direct access to: the mind. This lack of access leaves us to impute the mental states beliefs, desires, emotions, intentions, etc. of others before we can explain their behaviors. Studying our ability to do so, our Theory of Mind, has long been the province of psychologists and philosophers. Computational scientists are joining this research space, however, as they strive to imbue artificial intelligences with human-like characteristics. We provide a high-level review of Theory of Mind research across several domains, with the goal of mapping between theory and recursive agent models. We illustrate this mapping using a specific recursive agent architecture, PsychSim, and discuss how it addresses many of the open issues in Theory of Mind research by enforcing a set of minimal requirements.