Latency of voluntary cancellation of the human vestibule-ocular reflex during transient yaw rotation

Volitional suppression of the initial vestibuloocular reflex (VOR) was studied in ten normal humans, aged 29+/-8 years (mean+/-standard deviation, SD), who were rotated about a vertical axis centered between the otoliths. Rotations consisted of steps in acceleration of 2800, 1600, 1000, or 500 degrees/s(2) delivered at unpredictable times in unpredictable directions in the horizontal plane. As a test of the VOR, subjects were asked to attend to an earth-fixed target located 500 cm away that was either continuously visible or extinguished immediately before rotation. The gain of the VOR (angular eye velocity/angular head velocity) was 0.78+/-0.01 (mean+/-standard error of the mean, SE) during the period 35-45 ms after the onset of head rotation and 0.952+/-0.005 during the period 125-135 ms after the onset of head rotation. Subsequent rotations were performed during viewing of a target that moved with the head (cancellation). Cancellation was studied under three conditions of target visibility: (1) with the target always visible; (2) with the target always extinguished immediately prior to head rotation; or (3) with the target unpredictably extinguished half of the time immediately before each rotation. Cancellation responses showed individual idiosyncrasies, but certain features were common to most subjects. During cancellation, the VOR response generally differed from the earth-fixed target condition in that there was usually a small decrease in slow-phase VOR gain followed by an oppositely directed saccade. During the highest acceleration (2800 degrees/s(2)), the latency of the earliest statistically significant gain decrease for cancellation, as compared with the earth-fixed target condition, averaged 48+/-5 ms (mean+/-SE) from the onset of head rotation, although it was significantly shorter in one subject who had an onset at 14+/-2 ms. The latency of cancellation increased as head acceleration decreased such that, for each stimulus, cancellation began when the head was displaced an average of 1.4+/-0.1 degrees (+/-SD). Because VOR cancellation generally occurred before the availability of visual feedback or under conditions when vision was never permitted, it is inferred that cancellation is triggered by a threshold eye position or an estimate of head rotation based on integration of vestibular afferents. Cancellation occurred significantly earlier with a visible target than with an extinguished target only at the lowest peak head acceleration of 500 degrees/s(2). Corrective saccades with a visible target occurred later for head accelerations of 500 and 1000 degrees/s(2) than for greater head accelerations. Significant effects of target illumination on the latencies of both saccades and cancellation occurred at least 80-90 ms after the onset of head rotation, consistent with the earliest available visual feedback. This longer latency of saccades for visible as compared with extinguished targets may be analogous to a release of fixation, as occurs with express saccades. The latency difference due to target visibility was not related to prediction, since it was unchanged under conditions of random target illumination.