Affiliations: Department of Physiology and Northwestern University Institute for Neuroscience, Northwestern University, Chicago, Illinois USA
Note: [*] Reprint address: K.D. Powell, Laboratory of Sensorimotor Research, National Eye Institute, NIH - Building 49, Room 2A50, Bethesda, MD 20892.
Abstract: The ability of the vestibulo-ocular reflex (VOR) to alter the phase of the motor output relative to the sensory input is examined. Alert cats were trained for 2 h with 0.25 Hz sinusoidal horizontal vestibular and vertical optokinetic rotational stimuli. In each experiment the optokinetic training stimulus was phase shifted by 0°, +45°, −45°, or 90° from the vestibular stimulus. Vertical and horizontal eye movements were measured during horizontal rotations in darkness before and after the training procedure. Phase-advance experiments (+45°) produced an adaptive vertical VOR with a mean phase of +28°. After phase-delay experiments (−45°), the adapted VOR had a mean phase of −19°. The peak adaptive change in VOR gain was at or near the 0.25 Hz training frequency in each experimental group, but the gain depended in a complex manner on the testing frequency and the degree of phase shift of the training stimulus. Training with a 45° phase-delayed optokinetic stimulus produced an adaptive vertical VOR with a gain that was relatively higher at frequencies below the training stimulus than at those that were above. Training with a 45° phase-advanced optokinetic stimulus produced an adaptive vertical VOR with a gain that was higher at frequencies above the training frequency than at those that were below. During training with a phase-shifted optokinetic stimulus, adjustment of the relative efficacies of two neural pathways, a velocity pathway and an integrating pathway, could account for gain dependence on testing frequency and phase shift. This was corroborated by a model of the VOR that incorporates parallel velocity and integrating pathways. Data from 45° phase advances were fit by increasing the gain of the velocity versus integrating pathway, whereas 45° phase delay data were fit by decreasing the gain of the direct versus integrating pathway. The models altered the time constants of either the common oculomotor integrator or the velocity storage mechanism.
