Predictive processing in the cortical network of biological motion perception: An fMRI-DCM study

Abstract

Biological motion perception (BMP) is fundamental for interpreting and predicting human actions, playing a crucial role in social cognition. While traditionally viewed as a bottom-up process, recent evidence suggests that top-down mechanisms, particularly expectations, influence BMP. This fMRI study investigates how prior expectations shape BMP by employing a predictive cueing paradigm. Participants discriminated between biological and scrambled motion while presented with congruent, incongruent, or uninformative cues. Behavioral results revealed that correct expectations facilitated performance, while incorrect expectations impaired it. Dynamic Causal Modeling (DCM) analysis demonstrated bidirectional connectivity within the action observation network (AON), particularly between the posterior superior temporal sulcus (pSTS), posterior parietal cortex (PPC), and inferior frontal gyrus (IFG) under all expectation conditions. Additionally, Multivariate Pattern Analysis (MVPA) confirmed the involvement of these regions in BMP. These findings challenge traditional bottom-up models by highlighting the role of predictive processes in action perception and provide evidence for expectation-dependent modulation of AON connectivity.