The Social Brain: Motor Similarity Sharpens Affective Intention Decoding

Humans readily infer others’ feelings from how they move, yet the computational principlesthat support this ability remain debated. Two prominent accounts propose either direct motormatching or predictive coding mechanisms. Here we combine motion-capture, behavior asses-sments, and fMRI multivoxel pattern analysis to test whether an observer’s own motor reper-toire operates as an embodied prior during affective action perception. We quantified motorsimilarity between each observer and stimulus kinematics and examined its impact on neuralrepresentations within the action observation network. Motor similarity sharpened affectiverepresentations in inferior frontal gyrus: cross-validated classifiers trained on high-similaritytrials decoded valence reliably above chance and outperformed classifiers trained on low-simi-larity trials. In contrast, motor similarity attenuated univariate BOLD responses in inferior pa-rietal lobule, consistent with reduced prediction error under more precise priors. Behaviorally,observers judged motorically dissimilar interactions as more intense, indicating greater per-ceptual salience when observed kinematics diverge from one’s own repertoire. Affective inten-sity further modulated neural precision in inferior frontal gyrus, revealing an interactionbetween social salience and embodied priors. Together, these converging neural and behavio-ral results adjudicate between competing accounts by demonstrating a principled double dis-sociation: representational sharpening in inferior frontal gyrus without increased activation,alongside attenuation in inferior parietal lobule under higher similarity, supporting a hierar-chical predictive coding framework for social perception. Beyond action observation, our re-sults show that a person’s own movement repertoire calibrates neural representations of affect,offering a general principle for efficient embodied inference.