We showed previously that smooth pursuit training combined with whole-body
rotation in the orthogonal plane induces adaptive cross-axis vestibulo-ocul
ar reflex (VOR). To gain an insight into the possible pathways and the natu
re of error signals for cross-axis VOR adaptation, we examined further prop
erties of adaptive responses. In the first series, we trained monkeys for v
ertical pursuit during sinusoidal yaw rotation at 0.5 Hz ( +/- 10 degrees)
by presenting a target spot either in phase with, or with phase shifts (lea
d or lag) of 90 degrees to, the chair for I h. After training, sinusoidal o
r trapezoidal yaw rotation was tested in complete darkness without a target
. Different training conditions resulted in different amounts of phase shif
t in cross-axis VOR. Trapezoidal yaw rotation (peak acceleration, approxima
te to 780 degrees /s(2)) revealed further differences in the direction, lat
ency and time course of the adaptive responses depending on the conditions
of the pursuit task. At least two (fast and slow) components with different
latencies were induced in the cross-axis VOR by trapezoidal rotation after
in-phase and phase-shift training. Adaptive responses were accurately simu
lated by the weighted sum of these two components. In the second series, we
examined the effects of sequentially flashed (10 mus) targets in the horiz
ontal plane during pitch rotation. The monkeys learned to track such target
s by smooth pursuit, and cross-axis VOR was also induced after such apparen
t motion stimuli without retinal slip of the target image. These results in
dicate the importance of eye velocity for cross-axis VOR and suggest that t
his adaptation occurs most probably in the smooth pursuit pathways.