1. This study was designed to determine whether the supplementary motor area (SMA), the primary motor cortex (MC), and the putamen, all of which are components of the basal ganglia-thalamocortical 'motor circuit', contain neural representations of the target or goal of a movement, independent of specific features of the movement itself. Four rhesus monkeys were trained to perform two visuomotor delayed step-tracking tasks in which the subject used a cursor to track targets on a display screen by making flexion and extension movements of the elbow. Single-cell activity was recorded from the SMA, MC, and putamen while the monkeys performed the two tasks. In the Standard task, the cursor and the forearm moved in the same direction. The Cursor/Limb Inversion task was identical to the Standard task except that there was an inverse relationship between the directions of movement of the forearm and cursor. Together, these tasks dissociated the spatial features of the target or goal of the movement from those of the movement itself. Both tasks also included features that made it possible to distinguish neuronal activity related to the preparation for movement from that related to movement execution. A total of 554 directionally selective, task-related neurons were tested with both tasks (SMA, 207; MC, 198; putamen, 149). 2. Two types of directionally selective preparatory activity were seen in each motor area. Cells with target-dependent preparatory activity showed selective discharge prior to all preplanned movements of the cursor toward one of the side targets (right or left), irrespective of whether the limb movement involved extension or flexion of the elbow. Comparable proportions of target-dependent preparatory cells were seen in the SMA (36%), MC (40%), and putamen (38%). Cells with limb-dependent preparatory activity showed selective discharge prior to all preplanned elbow movements in a particular direction (extension or flexion), irrespective of whether the target to which the cursor was moved was located on the right or left side of the display. The SMA contained a higher proportion of limb-dependent preparatory cells (40%) than either MC (15%) or putamen (9%). 3. Two types of directionally selective movement-related activity were also seen in each motor area. For cells that showed limb-dependent activity, the movement-related discharge was associated with elbow movements in a particular direction (extension or flexion). In contrast, for the cells that showed target-dependent activity, the movement-related discharge was associated with all movements of the cursor toward one of the side targets (right or left) regardless of whether the elbow was being extended or flexed. Most movememt-related neurons in all three motor areas were limb dependent (SMA, 65%; MC, 71%; putamen, 63%). However, some movement-related cells showed target dependence, with the percentages of such cells in the SMA (16%) and MC (14%) being twice as high as that in the putamen (6%). 4. These results indicate that the SMA, MC, and putamen each contain neurons that represent the target or goal of a limb movement as well as neurons that represent the direction of the limb movement itself. Moreover, both aspects of motor control appear to be represented in all three areas during the preparation for movement as well as during movement execution. The fact that target-dependent representations were more common during the preparation for movement, whereas limb-dependent representations were predominant during movement execution, suggests that relatively 'high' levels of motor processing are emphasized during the preparation for movement, with an incomplete shift toward 'lower' levels during movement execution. Nevertheless, the observation that both target-dependent and limb-dependent variables were represented simultaneously in all three motor areas (SMA, MC, and putamen) indicates that multipel levels of motor processing progress largely in parallel during both the preparation and execution of visually guided limb movements. 5. Within the MC, many of the target-dependent cells, both preparatory and movement-related, showed 1) discrete sensorimotor fields restricted to the elbow or shoulder, 2) short-latency 'proprioceptive' responses to torque application, and/or 3) 'muscle-like' responses to loads that opposed or assisted the task-related limb movements. A small number of such cells were also seen in the SMA and putamen. These findings appear to be at variance with a strictly hierarchical, serial/analytical model of motor processing.
