What is computational social cognition?

Computational social cognition is a new approach to studying the mechanisms of social cognition—that is, for understanding how we form social impressions and attitudes, represent them in the mind, and use them to guide our choices and actions (Hackel & Amodio, 2018).

Our lab combines computational modeling with behavioral and neuroimaging methods to study the formation and expression of prejudice, intergroup bias, attitudes, and trait impressions. This method begins by developing a theoretical model about the cognitive processes that produce a response—for example, how a preference for one social group over another forms across repeated experiences with group members. This model is then expressed formally, as a computational model that generates expected patterns of outcomes. By testing the degree to which participants’ experimental task behavior fits the model, we can identify specific processes (e.g., forms of learning, decision biases, and interactions among processes) that drive aspects of social cognition.

In much of our work, we pair this computational approach with a Memory Systems Model of learning and memory, which identifies distinct systems of learning that operate, often in concert, to support social learning and behavior (Amodio, 2019). For example, our recent work examines the role of instrumental learning in the formation of attitudes, impressions, and prejudices, and our models allow us to compare these instrumental processes with other potential mechanisms, such as Pavlovian or episodic learning.

Instrumental social learning

The computational approach in our lab has allowed to examine an under-studied aspect of social cognition: how we form impressions and attitudes through direct social interaction. In a series of studies, we show that impressions may be formed through reward reinforcement learning in repeated interactions—that is, the process of acting toward a person and learning from their feedback. We find that human perceivers naturally infer not just the reward value of an interaction partner (e.g., how often a person shares money), but also personality traits (e.g., their degree of generosity).

Using a computational neuroimaging approach, we have shown that instrumental learning about rewards and traits involves distinct computations in the striatum—a neural structure that computes prediction errors—and that representations of reward and trait values are combined to guide future social decisions (Hackel et al., 2015). We also find that the way this information is combined changes adaptively depending on the context (Hackel et al., 2022), and that trait-based reinforcement learning is enhanced for social agents but not unique to them (Hackel et al., 2020). Finally, in a set of behavioral experiments, we show that instrumental learning from individual members of a group is represented as a group-level preference (Hackel et al., 2022).

Instrumental learning of prejudice

How do explicitly prejudiced messages lead to deeply-­‐engrained implicit racial biases? When Geert Wilders calls Moroccans “scum,” people may dismiss it as political rhetoric. Yet somehow this information, conveyed explicitly, works its way into our implicit (unconscious) thoughts and feelings. Once implicit, this bias can influence judgements and actions toward racial minorities without our awareness.

My NWO-funded VICI project tests a computational social neuroscience model of this process, identifying for the first time how explicit prejudices become implicit. By doing so, we can better understand how implicit biases are formed and expressed and how they may be changed.