Observer-agent kinematic similarity modulates neural activity in regions of the action observation network

Body movement conveys important information about agents’ internal states, such as their emotions or intentions. But how do humans use movement cues to understand others? A growing body of behavioural and neuroimaging work suggests that we map observed actions onto our own motor system to successfully interpret others’ movements, and that movement similarity between an agent and observer facilitates accurate inferences about the agent’s internal state. A network of frontal, parietal and occipitotemporal regions, termed the action-observation network (AON), is presumed to subserve this coupling between action observation and -execution. However, it is currently unclear how exactly the AON supports action understanding and how movement similarity modulates putative motor mapping processes.We used a well-established Theory of Mind task and fMRI to investigate whether regional blood flow in AON regions during mentalising is modulated by movement similarity. 31 participants first created their own animations of interacting triangles, depicting three mental state words, while their finger movements were recorded. Subsequently, individuals viewed animations created by an independent sample, which varied in the degree of kinematic similarity between animator and observer, while undergoing fMRI scanning. ROI analysis revealed a parametric decrease of neural activity within the right angular gyrus with increasing movement similarity. FWE-corrected whole-brain analyses additionally revealed a decrease in bilateral insular activation along with movement similarity. Results are in support of a predictive processing view of action understanding, wherein observers invert a generative model linking their own actions to affective outcomes to infer the internal states underlying others’ actions.