Predictive coding for the actions and emotions of others and its deficits in autism spectrum disorders

Traditionally, social perception has been studied by presenting brief actions or emotions of others in randomized order, thereby preventing participants from anticipating the actions and emotions of others. Here we review efforts inspired by the notion that the brain functions as a hierarchical Bayesian predictive coding system that suggest that when participants can predict the actions and emotions of others, feedback information flow from higher levels of the hierarchy can generate predictive signals about the upcoming actions and emotions of others and attenuate expected sensory responses in earlier levels of processing to generate prediction errors. We show how the combination of animal and human studies using a variety of methods is necessary to unravel these predictive coding architectures together with a shift towards paradigms enabling predictions. We further discuss how the notion of predicting coding is influencing how we think of autism spectrum disorder.