CNS Seminar

It has long been believed that voluntary movements are prepared before they are executed.  Yet it has remained unclear what this computation might be.  Furthermore, recently it has been observed that voluntary movements can be executed with surprisingly low latency, calling into question the hypothesis of a time-consuming preparatory computation. Neural recordings have not yet been able to settle this debate, and have yielded evidence both for and against the idea of a distinct preparatory stage.  A key hurdle has been that putatively preparatory responses and movement-related responses are mixed in the same neural populations.  We exploit a method that segregates preparatory and movement-related response components, allowing them to be studied independently across different contexts.  We find that a preparatory component is indeed present in motor and premotor cortex, regardless of whether movements are initiated following an instructed delay (the standard task in our field), are initiated internally at a time of the subject's choosing, or are generated rapidly and almost reflexively.  Indeed, these three contexts involved all involved the same sequence: a preparatory stage with expansive dynamics followed by a movement-generating stage with rotational dynamics.  In the standard instructed-delay context, preparatory activity was present throughout the delay.  When initiating a movement at a time of the subject's choosing, preparatory activity was deferred until movement neared.  When a movement had to be executed rapidly, preparation consumed only a few tens of milliseconds.  These results argue that voluntary movements, although they can be initiated in different ways, are subserved by a unified mechanism in which a preparatory stage seeds subsequent movement-generating dynamics.  However, the timecourse of preparation and its transition to movement appears to be extraordinarily flexible.