SACCADE-RELATED ACTIVITY IN MONKEY SUPERIOR COLLICULUS .2. SPREAD OF ACTIVITY DURING SACCADES

1. In the companion paper we described two classes of cells in the monkey superior colliculus (SC) that were related to saccade generation, buildup cells and burst cells, which fell into two functional sublayers within the intermediate layers of the SC. Fixation cells in the rostral SC were deemed to be part of the buildup cell layer. The buildup cells had several characteristics in common with cells in the cat described as having a ''hill of activity'' moving across the SC, but the burst cells had no such characteristics. In this paper we further investigate whether there is evidence for such a moving hill of activity in the monkey by analyzing the spatial and temporal activity of cells across the SC during the generation of visually guided saccades. 2. We recorded the activity of single cells while the monkey made saccades of different amplitudes (0.5-60 degrees). We recorded cells from locations extending from the rostral to caudal SC in order to sample cells whose optimal amplitudes ranged from small to large saccades. This allowed us to see any shift of activity across the SC before, during, and after saccades. It also allowed us to determine the fraction of the SC that was active during the successive phases of saccade generation. 3. During active visual fixation, the fixation cells in the rostral pole of the buildup layer showed an increased discharge rate. From the population reconstruction, we estimate that the zone of active cells spanned the most rostral 0.72 mm in each SC. Assuming the SC is 5 mm in length, similar to 15% of the cells lying along the horizontal meridian in the buildup layer would be active during fixation. 4. At least 100 ms before the initiation of a saccade, long-lead activity began to appear in the buildup layer at the site on the SC motor map related to the next saccade. Fixation activity in the rostral poles simultaneously began to diminish, but the cells in the burst layer remained relatively silent. 5. Approximately 25 ms before saccade onset, the fixation cells ceased firing and both burst and buildup cells began to burst. The active zone in the burst layer was estimated to be similar to 1.4 mm diam, occupying roughly 28% of the SC along a line sunning from the rostral pole through the center of the initially active zone. The size of this active area among the burst cells was independent of saccade amplitude. The size of the initially active zone in the buildup layer was larger than in the burst layer and was dependent on saccade amplitude; it was larger for larger saccades. 6. During the saccade, all cells in the buildup layer lying rostral to the initially active zone became active, and their peak discharge occurred later in the saccade as the cells were located more rostrally. Cells lying caudal to the initially active buildup cells were not activated. During the saccade, activity in the burst cell layer collapsed, but there was no shift in the locus of this activity in the SC. 7. We interpret the sequential activation of the buildup cells during a saccade as a spread of activity rostrally across the buildup layer of the SC. We saw no evidence for a spread of activity in the burst layer. 8. These experiments allow us to propose the following sequence of activity among the SC cells during generation of a saccade. During fixation, activity is confined to the fixation cells in the rostral SC, and we hypothesize that these cells suppress saccades via inhibitory connections directly onto the saccade cells in the caudal SC and excitatory connections onto the omnipause neurons in the pens. The buildup cells show the earliest activity preceding a saccade, and we suggest that this activity is related to preparation to make a saccade, including selection of target amplitude and direction. The burst cells are active just before saccade onset and could provide input to the pens for the amplitude and direction of the saccade. The pause in activity of the fixation cells is critical for the timing of the saccade. We think that the rostral spread of activity in the buildup cells, and the sharp reduction in burst cell discharge, are consistent with a feedback signal to the SC from the pens. We conclude that these changes in the spatiotemporal distribution of activity in the monkey SC are critical for controlling when a saccade occurs, its amplitude and direction, and its trajectory.