Noble, Jason, Tom Hebbron, Johannes Van Der Horst, Rob Mills, Simon T. Powers, and Richard Watson. “Selection pressures for a theory-of-mind faculty in artificial agents.” (2010): 615-615.
URL1 URL2
To have a theory of mind (ToM) is to anticipate the behaviour of other agents by considering what they want and what they know. It requires a representation of the environment that includes the internal states (e.g., beliefs) of other agents. Adult humans generally possess a ToM ability, demonstrated by reasoning like “he did not see the chocolate being switched from the red box to the blue one, so I predict he will choose the red box.” Note the distinction between what the speaker believes to be true, and what the speaker believes about the other agent’s belief states. ToM is of interest in developmental psychology (when and how do children acquire it?) and primatology (do our near relatives possess it?). In this project we ask: in an evolving population of social agents, under what circumstances would a ToM ability be selected for? Using simulation to identify the ecological niches that produce selection pressure for ToM should cast light on its origin in humans and on when we should expect to see it in other animals. We build on earlier work by Takano and Arita.