Abducens nerve multiunit responses were recorded in darkness from decerebrated frogs during steps of angular velocity about an axis tilted with respect to the earth vertical (off-vertical axis rotation, OVAR). Thereby, a rotating gravity vector activated utricular hair cells and modulated the abducens nerve discharge sinusoidally as a function of head position in space. As expected, a bias velocity response component and nystagmus-related changes in neural activity were absent, since frogs do not possess a functioning velocity storage mechanism. Responses increased as a function of the tilt angle and of the velocity and direction of the platform rotation. OVAR in the direction of the recorded abducens nerve (clockwise for the right and counterclockwise for the left abducens nerve) evoked significantly smaller responses than rotation in the opposite direction. The possible origin of these direction-specific response properties was further studied after lesioning various structures assumed to modify utriculo-ocular reflexes. Each of these lesions (ipsilateral hemilabyrinthectomy, cerebellectomy, contralateral canal nerve sections) had a specific effect on the recorded response properties, but none of them, nor combinations thereof, abolished the direction-specific characteristics of the responses as long as the contralateral utricular nerve branch remained intact. Our results demonstrate that direction-specificity is a property of the basic utriculo-ocular reflex that is independent of the velocity storage mechanism in the brainstem, of the intervestibular commissural system, of the inhibitory control by the cerebellum and of the central convergence of utricular and horizontal canal inputs. A simple, unidirectional interaction between central utricular neurons with adjacent functional polarization vectors is suggested as the basic element for the observed direction specificity.

Abducens nerve responses of frogs were evoked by sinusoidal oscillations on a horizontal linear sled. The depth of modulation of these responses and their phases depended on the orientation of the head with respect to the direction of linear acceleration. Longitudinal acceleration evoked abducens responses that consisted of two discharge maxima per stimulus cycle. At consecutively more oblique head orientations, one of these two discharge maxima increased and the other decreased. Transverse accelerations evoked abducens responses that consisted of only one discharge maximum per stimulus cycle. Removal of the labyrinthine organs on one side abolished these responses in the contralateral abducens nerve but did not affect the responses in the ipsilateral abducens nerve. The latter result indicates that the responses in each abducens nerve originate from hair cells on the contralateral utricle. The experimentally determined modulation and phase values and their dependence on the orientation angle of the acceleration vector were used to characterize a functional cluster of hair cells located medially with respect to the striola in a fan-like sector on the utricle ('lateral rectus fan'). Parameters of this fan were used to develop a model that satisfactorily simulates the recorded abducens responses. This model predicts a majority of afferents with excitatory and a few afferents with inhibitory contributions to the abducens nerve responses. The phasic response components of about 90% of these afferents are larger than their tonic response components.