Systems Neuroscience of Primate Social Cognition

The brains of humans and other primates are tuned for social information. From the moment of birth, our social partners capture our interest. Observing others not only helps us optimise our own decisions ('social learning'), but also helps us understand our partners' minds and predict their intentions and behaviour ('mentalizing'). These social-cognitive abilities are profoundly impaired in autism, social anxiety, schizophrenia and related human conditions, due to dysfunction of specific brain circuits. Yet, we know very little about how neurons in these brain circuits process information to enable social cognition and social interactions. This is an important question because neurons are the basic computing elements of the brain that exchange information.

To understand human-like social cognition and its disorders, we need to know how neurons and distributed brain systems process information during social interactions. We address this issue by recording the activity of individual neurons in specific parts of the brain while monkeys socially interact with each other to solve reward-guided and rule-based decision problems.

Recording neuronal activity while the animals perform these decision tasks in a social, interactive context allows us to address specific questions: When monkeys observe and learn from each other's choices, how do neurons process the value of the social partner's choice options? Specifically, do neurons process such reward values from the partner's unique subjective perspective, even when this perspective differs from one's own? This process, known as mentalizing, is profoundly impaired in autism. Further, when monkeys interact to learn about abstract rules in a cognitive problem-solving task, do neurons update expectations about which rule is currently valid based on the observed partner's choices? And do they track the partner's belief about which rule is currently valid, even when this belief is different to one's own belief?

After studying the neuronal activity in these tasks, we also reversibly interfere with specific brain areas during task performance, which will help us understand which brain areas make essential contributions to social cognition. Finally, based on the obtained data, we will build computer models of the brain circuits we study, to better understand the information-processing mechanisms and computations performed in these brain areas. We focus on brain areas such as the amygdala and parts of prefrontal cortex, which we already know play a role in human social-cognition disorders.

We use monkeys in this research because they have sophisticated social abilities close to those of humans, and because they are suitable for recording the activity of individual neurons. Although human brain imaging can help identify which areas of the brain participate in social cognition, the imaging signal is too slow and spatially imprecise to reveal the dynamic, millisecond-precise information flow in single neurons. Likewise, although studies in rodents can uncover neuronal mechanisms for basic social behaviours such as aggression and mating, they cannot investigate advanced, human-typical social cognition such as social rule-learning, perspective-taking and mentalizing.

Primate social cognition involves sophisticated decision-making, observational learning and perspective-taking. To study the underlying neuronal mechanisms, we combine multi-area multi-electrode electrophysiological recordings with reversible interventions and computational modelling while monkeys perform well-established decision tasks in a social context.

In a value-guided decision task, two monkeys observe and learn from each other's choices how to obtain subjectively preferred rewards. In a social rule-based decision task, two monkeys observe and learn from each other which of several abstract rules is currently valid and leads to reward.

Neuronal recordings in both tasks allow us to identify neuronal signals related to shared, task-general social processes (self-other distinction, observational learning, social choice prediction, perspective-taking) and task-specific signals (subjective valuation and reward processing vs. rule following and conflict processing).

We use chronic electrode arrays and semi-chronic multi-electrode drives to target areas previously implicated in social behaviour, including anterior cingulate cortex, amygdala and connected temporal-lobe areas, and prefrontal areas implicated in decision-making and advanced cognition, including frontal pole and dorsolateral prefrontal cortex.

Using electrical microstimulation and reversible pharmacological interventions, we disrupt local activities and network interactions before and during task performance to establish which of these brain areas make essential contributions to social task performance. We integrate the neural and behavioural data using biophysically realistic neural-network modelling to build task-specific and task-general computational models, and thereby uncover the mechanisms and circuit architectures underlying primate social cognition.