Shaping the perceptual representation of observed human action through prediction.

Abstract

Understanding the actions of others is crucial for all social interactions. Despite a dynamic and complicated social world, humans can derive the goals, attitudes and beliefs that drive others’ actions, imbuing them with meaning and understanding. While such abilities were traditionally accounted for by a direct matching of observed actions to actions within the observer’s motor system, contemporary theories of social perception explain them within a predictive processing framework. They argue that perception of others’ actions is shaped by prior assumptions about their goals and intentions and the behaviours that these mental states predict. This thesis aimed to resolve whether people make such predictions, whether they are represented perceptually, and on which information they rely. Ten experiments utilized a variant of the classical Representational Momentum paradigm. They presented participants with the initial stages of a goal-directed action and asked them to make spatial judgments of its last seen position prior to sudden offset. As expected, the results revealed the top-down expectations that guide action perception. The findings revealed (1) that social predictions follow the principle of efficient action, biasing perception towards efficient action expectations, such that hands seen to reach straight towards an obstacle were perceptually lifted over it. These predictions were (2) derived spontaneously, were (3) perceptually represented, and emerged (4) from attributions of intentionality to the observed actor, even (5) when the action was already underway, based on the match between action kinematics and available target objects. The current findings provide direct evidence for predictive models of social perception. They confirm that the perceptual representation of others’ actions is biased by the intentions we assign to them and our predictions of how these intentions will be fulfilled, therefore providing new avenues to understand how action expectations can shape our understanding of other people’s actions.