Ph.D Thesis

Ph.D StudentHendel Tal
SubjectPhysiological Control of Binocular Eye Movements
DepartmentDepartment of Biomedical Engineering
Supervisor ASSOCIATE PROF. Moshe Gur


In primates, the high-acuity region of the retina is confined to a small circular region (~1º diameter), called the fovea. To properly interact with the dynamic environment, primates constantly move their eyes so that the fovea is directed at the current object of attention. However, since primates' eyes are front-facing and thus have a large region of binocular overlap, the oculomotor system is confronted with the challenge of controlling both eyes together and ensuring that both reach a single target accurately and reliably. This work investigates the physiological mechanisms that allow this exquisite control of binocular eye movements. Our results are the following. (1) There has been a continuing controversy as to whether binocular eye movements are controlled by mechanisms that are inherently binocular or inherently monocular. To resolve this dichotomy, we analyzed the scheme that suggests an inherently-binocular control of binocular eye movements and derived a prediction that is amenable to experimental testing. To examine whether this prediction is confirmed, we measured binocular eye movements in human subjects. Our results are inconsistent with an inherently-binocular control of binocular eye movements. This provides indirect support for the alternative scheme which suggests an inherently-monocular control of binocular eye movements. (2) Behaviorally, binocular eye movements exhibit a superposition of a slow component with an extremely rapid component, called a saccade. We used three well-known behavioral results of saccadic eye movements to show that the magnitudes of the saccadic commands to the two eyes must be given by weighted averages of the two eyes' angular demands to the target. To check this theoretical result, we conducted another set of experiments on human subjects. The results confirm our theoretical analysis; further, they strongly suggest that the weights used in the weighted averaging are constant. (3) We developed a physiologically-plausible model of the saccadic mechanism that explains how the weighted averaging of angular demands in saccadic commands comes about. (4) The superior colliculus is a midbrain structure whose deeper layers contain a motor map that place-codes saccade metrics through the population activity of a large number of neurons. Several decoding schemes have been proposed to explain how brainstem mechanisms downstream from the colliculus translate this place-coded neural activity into the rate code used by neurons in the brainstem saccadic generator. Importantly, however, these decoding schemes have only been shown to work heuristically. We therefore provide a comprehensive theoretical analysis of three prominent collicular decoding schemes.