The human horizontal vestibulo-ocular reflex during combined linear and angular acceleration

We employed binocular magnetic search coils to study the vestibule-ocular reflex (VOR) and visually enhanced vestibulo-ocular reflex (VVOR) of 15 human subjects undergoing passive, whole-body rotations about a vertical (yaw) axis delivered as a series of pseudorandom transients and sinusoidal oscillations al frequencies from 0.8 to 2.0 Hz. Rotations were about a series of five axes ranging from 20 cm posterior to the eyes To IO cm anterior to the eyes. Subjects were asked to regard visible or remembered targets 10 cm, 25 cm, and 600 cm distant from the right eye. During sinusoidal rotations, the gain and phase of the VOR and VVOR were found to be highly dependent on target distance and eccentricity of the rotational axis. For axes midway between or anterior to the eyes, sinusoidal gain decreased progressively with increasing target proximity, while, for axes posterior to the otolith organs, gain increased progressively with target proximity. These effects were large and highly significant. When targets were remote, rotational axis eccentricity nevertheless had a small but significant effect on sinusoidal gain. For sinusoidal relational axes midway between or anterior to the eyes, a phase lead was present that increased with rotational frequency, while for axes posterior to the otolith organs phase lag increased with rotational frequency. Transient trials were analyzed during the first 25 ms and from 25 to SO ms after the onset of the head rotation. During the initial 25 ms of transient head rotations, VOR and VVOR gains were not significantly influenced by rotational eccentricity or target distance. Later in the transient responses, 25-80 ms from movement onset, both target distance and eccentricity significantly influenced gain in a manner similar to the behavior during sinusoidal rotation. Vergence angle generally remained near the theoretically ideal value during illuminated test conditions (VVOR), while in darkness vergence often varied modestly from the ideal value. Regression analysis of instantaneous VOR gain as a function of vergence demonstrated only a weak con-elation, indicating that instantaneous gain is not likely to be directly dependent on vergence. A model was proposed in which linear acceleration as sensed by the otoliths is scaled by target distance and summed with angular acceleration as sensed by the semicircular canals to control eye movements. The model was fit to the sinusoidal VOR data collected in darkness and was found to describe the major trends observed in the data. The results of the model suggest that a linear interaction exists between the canal and otolithic inputs to the VOR.