Neurological basis for eye movements of the blind

When normal subjects fix their eyes upon a stationary target, their gaze is not perfectly still, due to small movements that prevent visual fading. Visual loss is known to cause greater instability of gaze, but reported comparisons with normal subjects using reliable measurement techniques are few. We measured binocular gaze using the magnetic search coil technique during attempted fixation (monocular or binocular viewing) of 4 individuals with childhood-onset of monocular visual loss, 2 individuals with late-onset monocular visual loss due to age-related macular degeneration, 2 individuals with bilateral visual loss, and 20 healthy control subjects. We also measured saccades to visual or somatosensory cues. We tested the hypothesis that gaze instability following visual impairment is caused by loss of inputs that normally optimize the performance of the neural network (integrator), which ensures both monocular and conjugate gaze stability. During binocular viewing, patients with early-onset monocular loss of vision showed greater instability of vertical gaze in the eye with visual loss and, to a lesser extent, in the normal eye, compared with control subjects. These vertical eye drifts were much more disjunctive than upward saccades. In individuals with late monocular visual loss, gaze stability was more similar to control subjects. Bilateral visual loss caused eye drifts that were larger than following monocular visual loss or in control subjects. Accurate saccades could be made to somatosensory cues by an individual with acquired blindness, but voluntary saccades were absent in an individual with congenital blindness. We conclude that the neural gaze-stabilizing network, which contains neurons with both binocular and monocular discharge preferences, is under adaptive visual control. Whereas monocular visual loss causes disjunctive gaze instability, binocular blindness causes both disjunctive and conjugate gaze instability (drifts and nystagmus). Inputs that bypass this neural network, such as projections to motoneurons for upward saccades, remain conjugate.