Context compensation in the vestibuloocular reflex during active head rotations

The vestibuloocular reflex (VOR) needs to modulate its gain depending on target distance to prevent retinal slip during head movements. We investigated gain modulation (context compensation) for binocular gaze stabilization in human subjects during voluntary yaw and pitch head rotations. Movements of each eye were recorded, both when attempting to maintain gaze on a small visual target at straight-ahead in a darkened room and after its disappearance (remembered target). In the analysis, we relied on a binocular coordinate system yielding a version and a vergence component. We examined how frequency and target distance, approached here by using vergence angle, affected the gain and phase of the version component of the VOR and compared the results to the requirements for ideal performance. Linear regression analysis on the version gain-vergence relationship yielded a slope representing the influence of target proximity and an intercept corresponding to the response at zero vergence ("default gain"). The slope of the fitted relationship, divided by the geometrically required slope, provided a measure for the quality of version context compensation ("context gain"). In both yaw and pitch experiments, we found default version gains close to one even for the remembered target condition, indicating that the active VOR for far targets is already close to ideal without visual support. In near target experiments, the presence of visual feedback yielded near unity context gains, indicating close to optimal performance (retinal slip <0.4 degrees /s). For remembered targets, the context gain deteriorated but was still superior to performance in corresponding passive studies reported in the literature. In general, context compensation in the remembered target paradigm was better for vertical than for horizontal head rotations. The phase delay of version eye velocity relative to head velocity was small (approximately 2 degrees) for both horizontal and vertical head movements. Analysis of the vergence data from the near target experiments showed that context compensation took into account that the two eyes require slightly different VORs. In the DISCUSSION, comparison of the present default VOR gains and context gains with data from earlier passive studies has led us to propose a limited role for efference copies during self-generated movements. We also discuss how our analysis can provide a framework for evaluating two different hypotheses for the generation of binocular VOR eye movements.