CONTROL OF REMEMBERED REACHING SEQUENCES IN MONKEY .2. STORAGE AND PREPARATION BEFORE MOVEMENT IN MOTOR AND PROMOTOR CORTEX

Single-neuron responses in motor and premotor cortex were recorded dur ing a movement-sequence delay task. On each trial the monkey viewed a randomly selected sequence of target lights arrayed in two-dimensional space, remembered the sequence during a delay period, and then genera ted a coordinated sequence of movements to the remembered targets. Of 307 neurons studied, 25% were tuned specifically for either the first or the second target, but not both. Iq particular, for neurons tuned d uring both target presentations, tuned activity related to a particula r first target direction were maintained during the presentation of a second target in a different direction. During the delay period, 32% o f the neurons were tuned fbr upcoming movement in a single direction. These delay period responses often reflected activity patterns that fi rst developed during target presentations and may therefore act to mai ntain target period information during the delay. Neurons with tuned a ctivity during both the delay and movement periods exhibited two patte rns: the first exhibited tuned responses during the delay that were co rrelated with the tuning of first-movement responses, while the second pattern showed delay-period tuning that was better correlated with tu ned responses during second movements. This indicates that, before mov ement, distinct neural populations are correlated with specific moveme nts in a sequence. About half the neurons studied were not directional ly tuned during the initiation, target, or delay periods, but did show systematic changes in activity during task performance. Some (34%) we re exclusively tuned during movement and appear to be involved in the direct control of movement. Others (17%) showed changes in firing rate from period to period within a trial but showed no directional prefer ence for a particular direction of movement. Population analyses of tu ned activity during the target and delay periods indicated that accura te directional information about both first and second movements was a vailable in the neuronal ensemble well before reaching began. These re sults extend the idea that both motor and premotor cortex play a role in reaching behavior other than the direct control of muscles. While s ome early neural responses resembled muscle activation patterns involv ed in maintaining fixed postures before movement, others probably rela te to the sensory-to-motor transformations, information storage in sho rt-term memory, and movement preparation required to generate accurate reaching to remembered locations in space.