A levels-of-analysis framework for studying social emotions

Abstract

Social emotions such as guilt and gratitude serve adaptive functions critical to social interactions and relationships. Therefore, an ecologically valid approach to studying the psychological and neural mechanisms of social emotions is to elicit and measure them in social interactive contexts, where relevant adaptive goals and functions are salient. However, multiple psychological and neurocognitive processes might be simultaneously activated during real-time social interactions: traditional observation-based tasks and self-report measures alone are not sufficient to capture and dissociate these processes. In this Perspective, we draw on Marr’s levels-of-analysis framework to argue that a holistic consideration of the goals and functions of a social emotion (computation level), formal modelling of its underlying cognitive operations (algorithm level), and neuroscientific measures of the biological bases of these cognitive operations (implementation level) will afford the theoretical frameworks and methodological tools necessary to advance understanding of social emotions. To support this argument, we describe research that showcases the utility of creative combinations of interactive tasks, neural and behavioural measures, and computational modelling for advancing understanding of how social emotions arise and achieve their adaptive goals and functions. Social emotions such as guilt and gratitude serve adaptive functions critical to social interactions and relationships. In this Perspective, Yu and colleagues argue that to advance a mechanistic understanding of social emotions, an integrative approach is needed that considers goals and functions, cognitive operations and biological implementation.