Learning through cooperation with conspecifics-'cooperative learning'-is critical to cultural evolution and survival. Recent progress has established that interbrain synchronization (IBS) between individuals predicts success in cooperative learning. However, the likely sources of IBS during learning interactions remain poorly understood. To address this dearth of knowledge, we tested whether movement synchrony serves as an exogenous factor that drives IBS, taking an embodiment perspective. We formed dyads of individuals with varying levels of prior knowledge (high-high (HH), high-low (HL), low-low (LL) dyads) and instructed them to collaboratively analyse an ancient Chinese poem. During the task, we simultaneously recorded their brain activity using functional near-infrared spectroscopy and filmed the entire experiment to parse interpersonal movement synchrony using the computer-vision motion energy analysis. Interestingly, the homogeneous groups (HH and/or LL) exhibited stronger movement synchrony and IBS compared with the heterogeneous group. Importantly, mediation analysis revealed that spontaneous and synchronized body movements between individuals contribute to IBS, hence facilitating learning. This study therefore fills a critical gap in our understanding of how interpersonal transmission of information between individual brains, associated with behavioural entrainment, shapes social learning. This article is part of the theme issue 'Minds in movement: embodied cognition in the age of artificial intelligence'.