The human ability to effortlessly understand the actions of other people has been the focus of intense research in cognitive neuroscience for decades. What have we learned about this ability, and what open questions remain? We address these questions by first considering the kinds of information an observer may gain when viewing an action. A “what, how, and why”
framing organises evidence and theories about: 1) the representations that support classifying an action; 2) how the way an action is performed supports observational learning, and inferences about other people; and 3) how an actor’s intentions are inferred from her actions. We then consider key evidence about how brain systems support action understanding, by illustrating several lines of research inspired by “mirror neurons” and related concepts. Understanding actions from vision is a multi-faceted process that serves many behavioural goals, and accordingly is served by diverse mechanisms and brain systems.