Identifying distinct neural features between the initial and corrective phases of precise reaching using AutoLFADS


Many initial movements require subsequent corrective movements but how motor cortex transitions to make corrections and how similar the encoding is to initial movements is unclear. We recorded a large population of neurons during a precision reaching task across multiple sessions to examine the neural space during not only initial movements but also subsequent corrective movements. AutoLFADS, an auto-encoder based deep-learning model, was applied to analyze individual corrective movements unique to any given trial and to stitch across sessions into a single, common neural space. Several locations in the neural space where corrective submovements originated after the initial reaches were identified and were different than the baseline firing rate before initial movements. The neural trajectories for corrective submovements were organized according to both their starting position as well as reach direction but varied with several local domains rather than a single global space. Decoding of reach velocity generalized poorly from initial to corrective submovements and corrective decoding was relatively poor. To improve corrective movement decoding, we demonstrate that a state-dependent decoder incorporating where in the neural space a correction was initiated improved performance, highlighting the diverse neural features of corrective movements during a precision reaching task.