Action observation in autism: preserved kinematic-driven sensorimotor resonance, distinct adult neural strategies

Abstract Background Difficulties understanding and predicting others’ actions are frequently reported in autism spectrum disorder (ASD). Whether they arise from atypicalities in early sensorimotor processing or at higher-order cognitive levels remains unclear. Methods We indexed kinematics-driven sensorimotor resonance by assessing cortico-kinematic coupling (CKC) during executed and observed movements in autistic and non-autistic participants. MEG signals were recorded while children (7–13 y.o; 17 autistic, 20 non-autistic) and adults (18–35 y.o; 22 autistic, 23 non-autistic) executed goal-directed hand-grasping movements or observed an actor performing the same task. Subjects' and actor's forefinger movement kinematics were monitored with an accelerometer. CKC was assessed as the coherence between movement acceleration and MEG signals at both sensor and source levels and compared between groups. Results Across participants, action execution and observation led to significant CKC within an overlapping visual, parietal and sensorimotor network. Within that network, CKC was stronger in adults than children, with no significant difference between autistic and non-autistic participants. In adults, autistic participants showed enhanced CKC at the right primary motor cortex during action observation, together with trait-dependent scaling of CKC in the somatosensory cortex. Conclusion Our results show that kinematics-driven sensorimotor resonance is preserved in autism. Age- and trait-dependent modulations in adulthood suggest a greater reliance on predictable observed action kinematics. These findings challenge accounts positing a primary disruption of embodied kinematic-driven sensorimotor mechanisms in autism. The root of altered action understanding in ASD may thus be located at higher-order cognitive levels. Limitations : The cross-sectional design limits developmental inferences, while the selected sample restricts generalizability and calls for broader inclusion criteria in future studies. Differences in autistic trait assessment and gender distribution across age groups, together with potential age-related MEG confounds, should also be considered.