The compositional encoding of hand-eye coordinated movements for single neurons in the posterior parietal cortex

Human posterior parietal cortex (PPC) is known to play a role in hand-eye (H-E) coordination, yet the underlying encoding mechanisms remain uncertain. We recorded 412 single neurons across 11 sessions from motor cortex (MC; n = 251) and PPC (n = 161) in a single human participant performing an H-E coordinated center-out task. While MC neurons showed negligible modulation by eye movements, 79% of PPC neurons had neural representations that were additively separable into independent hand- and eye-movement tuning curves. Due to this separability, representations could be separated and additively recomposed while maintaining structural similarity. Consequently, compositional decoders trained solely on single-effector movements could match the performance of decoders trained on coordinated H-E movements (hand: 66% vs. 69%; eye: 34% vs. 36%). These results show that, during simple center-out tasks, MC hand-movement codes are unaffected by eye movements and that compositionality can be used to modularly decode H-E coordinated movements in PPC.