Towards a Single-Neuronal Basis for Theory of Mind in a Primate Model

Abstract

Central to human social cognition is the ability to attribute mental states (e.g. beliefs and intentions) to another individual with the understanding that those states may differ from one’s own. This ability, termed “theory of mind” (ToM), is disrupted in a number of neurocognitive disorders, most notably in autism spectrum disorder. Here, we studied the neuronal correlates of ToM by obtaining single-neuronal recordings in the dorsal anterior cingulate cortex (dACC) and frontal pole cortex (FPC) of rhesus macaques as they engaged in a canonical test for ToM: the false belief task. The false belief task assesses a subject’s ability to predict another agent’s behaviors based on the other’s beliefs, which may be incongruent with reality. We trained a pair of macaques to perform such a task. In each session, the animals were designated either “actor” or “observer” roles. The actor is presented with a food item in one of two wells on a platform. The experimenter then maintains or alters the position of the item while the actor’s view is obscured, thus altering the veracity of the actor’s belief about the item location. The platform is then covered, the actor’s vision unblocked, and his arm released, allowing him to select one of the two well locations. Critically, the observer is able to watch these events, and therefore, can attribute false beliefs to the actor. During the task, we simultaneously tracked the observer’s eye movements and recorded neuronal activity from the observer’s dACC and FPC. We found that the observing primates behaved differently when their partners held a false belief. The observer looked at the initial food location for a significantly longer duration in the moments after the actor’s arm was released, prior to the actor’s decision timepoint. This change in behavior was not seen when the actor was aware of a food item switch, nor when the actor was replaced with a non-conspecific actor. We discovered neuronal populations in the dACC and FPC that increase their activity during the prediction of another agent’s incorrect decision based on a false belief. These findings suggest the presence of primitive ToM cognitive skills in rhesus macaques, which correlate with the activity of populations of neurons in both the dACC and FPC. Our results lay the groundwork for identifying specific, neurobiologically guided targets for the treatment of social cognitive disorders.