Effective connectivity in human parieto-frontal networks during grasping with and without visual feedback

Reach-to-grasp actions require continuous transformation of visuomotor signals within specialized parieto-frontal circuits. Neuroimaging can identify these networks, but dissociating motor execution from visual feedback of one’s own movements is challenging. Here, we addressed this limitation using MOTUM (Motion Online Tracking Under MRI), a system combining real-time kinematic tracking with virtual reality during fMRI, enabling independent manipulation of motor execution and visual feedback. Twenty-four right-handed human participants performed and/or observed reach-to-grasp actions during fMRI, in a 2×2 factorial design with movement and visual feedback present or absent. In visual-only trials, they observed a replay of their own previous movement. The sensorimotor cortex (M1-S1) was activated when moving and suppressed (vs. baseline) in visual-only trials. Dorsal extrastriate regions (hMT+, V3A) were activated by observation and suppressed when moving without visual feedback. The supplementary motor area was active only during movement. The premotor ventral and dorsal areas, the anterior intraparietal sulcus (aIPS) and the anterior superior parietal lobule (aSPL) responded to both movement execution and visual feedback. Critically, aIPS and aSPL showed a sub-additive response when movement and visual feedback co-occurred. Dynamic causal modelling with parametric empirical Bayes revealed a driving input on aSPL when moving with visual feedback and a aIPS excitation on PMv when watching hand visual feedback. Crucially, we observed a backward excitation and forward inhibition in aSPL-PMd when moving without visual feedback, which was reversed in a forward excitation and backward inhibition in when moving with visual feedback, suggesting that this pathway conveys motor-based predictions of sensory consequences.