Neural Dynamics of Reaching Following Incorrect or Absent Motor Planning

Neural prostheses are biomedical devices for restoring movement to individuals with paralysis or lost limbs. Signals from the patient’s motor cortex are decoded to infer intended movements, allowing the user to naturally control a prosthetic arm. Current neural prostheses assume that the observed neural activity relates only to the immediate movement being controlled. This assumption precludes doing more complex tasks such as planning out a movement or simultaneously executing one action while planning a subsequent one. To enable next-generation neural prostheses, we set out to understand how movement planning relates to movement execution. When given time to plan a reach, neural activity achieves and holds a distinct pattern of firing rates, called the plan state, which is specific to the upcoming movement. The plan is beneficial for initiating the reach quickly. It remains unknown, however, whether entering this plan state is necessary for successfully generating the reach. Our experiments answer this fundamental question by testing whether the neural state must pass through the plan state before a movement is generated in two cases: 1) there is no time to plan, and 2) a different movement has already been planned. We recorded from premotor and primary motor cortices of macaques making reaches in these two situations. In both cases, we found that although the monkey’s reaction time was slower, the movement was the same even though neural activity did not pass through the plan state. Thus, motor planning is a beneficial, but not strictly necessary, step of movement generation.