Beatrice de Gelder

Social species, and specifically primates, rely heavily on conspecifics for survival. While higher cognitive abilities play a role in understanding action, affect, and decision-making aspects of social behavior, it is clear that across the primate order nonverbal communication predominantly starts with visual perception of other agents. Our current understanding of the visual basis of these processes is still poor, in part because it is not clear how a wealth of insights into object perception would carry over to perception of social agents.

A core aspect of social communication concerns the perception of whole-body movements and their synchronization between agents. This research project aims at discovering the midlevel visual features of human whole-body movement perception that are the core drivers of affect perception. For this purpose, our study will range from behavioral and neural measures of body shape and movement encoding to online social interaction. Most experiments include Virtual Reality designs whereby VR, used in the manner of classical psychophysics, systematically alters the dimension of interest in body movement features per experiment. The research will devise a neurobiologically and ethologically realistic and highly performant model of body perception. This model will combine elements from deep learning suitable for coping with large stimulus sets, with physiologically plausible neurodynamic approaches appropriate for modeling the flexible organization of body processing. The novel questions and methodologies of this research will transcend the traditional debates in the neurobiology of emotion on hardwired vs. subjective constructions and orient theory towards a neuroethological perspective.