For each trial of the black fixation flash paradigm (full-screen flashes), we measured average eye position in the final 50 ms before stimulus onset. Across trials within a given session, this resulted in a population of starting eye positions. We split the trials within a session based on the median value of starting vertical eye position. We then measured the ocular position drift response for all sessions when only taking the ‘above’ median group of the sessions (purple) and again when taking only the ‘below’ median group of all sessions (bluish). (A, B) Position measurements in two monkeys. We vertically aligned traces (based on starting position at stimulus onset) in order to highlight the relative change in eye position associated with the ocular position drift response (similar to how we analyzed data in all other figures; e.g. Figure 1). (C, D) Eye velocity measurements. In both eye position and eye velocity, the ocular drift response measurements were statistically very similar; there were no time intervals in which the 95% confidence intervals between the ‘above’ and ‘below’ median groups did not overlap with each other. We also confirmed that in the interval of 50–140 ms after stimulus onset, peak vertical eye velocity was not different between the two groups (p=0.63, Z = 0.48 for monkey A and p=0.46, Z = 0.74 for monkey M; Wilcoxon rank sum test). All other conventions are similar to our other analyses of the ocular position drift response (e.g. Figure 1). Similar results were obtained with the split view stimulus and white fixation flash paradigms (e.g. p=0.32, Z = −1.01, and p=0.62, Z = −0.5 for monkeys A and M, respectively, in the white fixation flash paradigm; similarly, p>0.15, Z = −1.46,–0.82, and p>0.2, Z = −0.55, 1.27, and p>0.85, Z = 0.01, 0.19 for monkeys A, M, and N, respectively, in the split view stimulus paradigm; Z-values are reported for right and left darkness for each monkey in the split view stimulus paradigm).