Integration of Motor Planning and Execution through Latent Structure Reorganization in the Posterior Parietal Cortex
The posterior parietal cortex (PPC) plays a central role in sensorimotor control, performing visuomotor transformations, supporting planning, and providing visual feedback. However, it is unknown how the neural populations in different PPC areas organize their activity during this process. It has been proposed that PPC activity reflects population-level dynamics rather than distinct subpopulations, raising the question of how the population flexibly reorganize between the two main phases of motor control, planning and execution. To address this question, we analyzed neural dynamics in three PPC areas (PE, PEc, V6A) in the context of a delayed reaching task, applying dimensionality reduction techniques. This approach allows identifying whether activity in each area is organized into independent or partially overlapping dynamics across task phases. We found evidence of area-specific population subspaces, distinct for movement planning and execution. Specifically, the analysis revealed that in PE, which is a predominantly somatomotor area, neural activity occupied nearly orthogonal subspaces between the two phases, suggesting independent dynamics for movement planning and execution. In contrast, in V6A and PEc, which are involved in visuomotor transformations, we identified both shared and exclusive subspaces, indicating a more flexible representation of motor information in these areas. Overall, our findings suggest that parietal circuits combine both separation and sharing of neural representation to support computations during the different movement stages, providing new insights into the role of the PPC in generating flexible motor behavior . Significance StatementThe posterior parietal cortex (PPC) plays a central role in sensorimotor control. How neural ensembles in the PPC transition from planning to executing goal-directed movements remains poorly understood. Using dimensionality reduction on macaque electrophysiological data from delayed reach tasks, we identify an anteroposterior gradient: posterior visuomotor areas exhibit both overlapping and segregated subspaces for planning and execution, whereas anterior somatomotor regions show predominantly orthogonal dynamics. This organizational gradient reflects persistent posterior encoding of action goals and posture, and more distinct anterior representations of body state across planning and execution. The topographical and functional shift from visuomotor to somatic coding aligns with hierarchical predictive coding accounts of sensorimotor control and advances our understanding of sensorimotor transformations in the PPC.