Abstract
INTRODUCTION: Smooth pursuit eye movements are slow rotations that allow the eye to track a moving visual stimulus. In contrast, the vestibulo-ocular reflex (VOR) is a low-latency reflex in which the eye counter-rotates to compensate for head movements and stabilizes gaze relative to spatial targets. These movements are generally believed to have little to do with each other, and are usually studied independently. Here, we present an integrated computational model that can generate both VOR and pursuit, depending on the stimulus. METHODS: The model is organized into several feedback loops, with each component representing a neural structure known to participate in eye movements. First, we show that vestibular input may be used to convert a desired 3D gaze velocity signal into the velocity of the eye relative to the head. We propose that this conversion occurs in the floccular target neurons or eye-head neurons in the brain stem. Second, we model the motor neuron pool and the cerebellum as an adaptive controller that learns to control eye velocity relative to the head, with efference copy from PMT cell groups and a teaching signal from retinal slip. Third, pursuit is achieved by providing a descending desired gaze velocity to the above structures. The motion of a target in retinal coordinates (estimated in areas MT and MSTl) may be converted to desired gaze velocity (possibly in FEF or the rostral pole of the superior colliculus). RESULTS: The implemented model produces VOR movements when only the head is moved, and pursuit eye movements when a designated visual target is moved. We have also implemented this model with a robotic eye [Lesmana, et al. 2014]. CONCLUSION: Our findings suggest that smooth pursuit is an integrated part of the gaze stabilization system and utilizes the phylogenetically older mechanisms related to VOR.
Meeting abstract presented at VSS 2016