The temporal representation of computational features from body expressions

Body expression is an important channel for human communication. To understand the underlying perceptual mechanisms, the present study used electroencephalography and time-resolved multivariate analyses to track the temporal emergence of computationally defined body features and their contribution to emotion representations. Six body postural features (i.e., limb angles, symmetry, shoulder ratio, surface, potential energy, and limb contraction) were computed based on fourteen body joints. Representational similarity analysis revealed that the body surface and potential energy were significantly encoded at an early stage, persisting for an extended duration over parietal- posterior electrodes. In contrast, limb contraction was encoded at a later stage and exhibited a shorter duration. Furthermore, commonality analysis demonstrated that the postural features of surface and potential energy explained the neural variance of emotion categories, with body surface accounting for the largest proportion of the shared neural variance. These findings provide empirical evidence for the temporal dynamics of computationally defined body-posture representations during emotional body perception and may inform future computational models of body expression perception.