Neural Control Of Gaze

National Eye Institute

Investigators

Linked publications & trials

Abstract

One of the most important brain functions supporting vision is the ability to coordinate the movements of the eyes and head to control gaze, which is where we look in space. This project is devoted to understanding the nature of neuronal interactions underlying the neural control of gaze. Neurophysiological techniques can observe the responses of some neurons, but they can not reveal directly the nature of their functional interactions. This project constructs and tests mathematical models of sensory and motor functions involved in the control of gaze based on experimentally observed neuronal activity. In prior models of saccades (rapid, voluntary eye movements), the key role of controlling the movement's goal and speed was given to the superior colliculus (SC). The role of the cerebellum (C), in contrast, was assumed to involve the long-term regulation of saccadic accuracy. Analysis of neuronal responses from the SC and the C has led us to postulate a new model of how the brain controls visually guided saccades. The new model has two branches, one through the SC and one through the C, operating in parallel. This helps explain one of the earliest lesion studies in SC: even after bilateral SC ablations, the brain can still make saccades. Under normal conditions, the model uses the SC to control saccade beginnings, and the C to control saccade endings. More importantly, it lets us form a new interpretation of the role of each area. Thus, the SC is now believed to be generating a signal, in retinotopic coordinates, that represents where the selected target is. Thus, it is a sensory, and not a motor signal (as has been previously thought). Several recent studies have lent further support to this reinterpretation. The model also suggests a new, dual role for the cerebellum. The first, and more basic, role is to update the distribution of output activity during a movement. This is the role that corresponds to feedback in a classical model. The second, and more subtle, role is to recognize the constellation of inputs (i.e., sensory, motor, behavioral), or context, before the movement. This context causes the C to inhibit activity at a specific locus in the cerebellar vermis. This in turn disinhibits activity on specific fastigial nuclear cells, which in combination with the first role of the cerebellum drives and steers the eye to its final position. These two roles lead us to describe the cerebellum as having a "pilot map", which controls the eye movement. Although considerable evidence supports the spatial integration part of this theory, the context-dependent aspect of the theory has only recently begun to be studied experimentally. Current work in our laboratory is using functionally realistic models of microcircuitry in the C to evaluate the plausibility of these recognition and integration functions. Additionally, this theoretical approach is also being extended to the study of vergence eye movements. Vergence movements are the disconjugate movements made when the eyes look at objects at different distances. Closer objects require convergence of the two eyes, and farther objects require divergence of the eyes. Importantly, failures of the vergence mechanism may be related to the inappropriate ocular alignment seen in infantile strabismus (esotropias result from too much convergence, and exotropias result from too much divergence). Surprisingly, how the brain maintains the alignment of the two eyes is still unknown. A theoretical study may elucidate much of the existing clinical data on strabismus and its surgical treatment, and may suggest improved methods for characterizing and treating strabismus. Gaze saccades (coordinated eye and head movements) could be controlled by a single gaze controller, with eye and head movements sharing a common motor drive. However, experimental evidence has shown that eye and head trajectories can be decoupled by changes in initial eye and head position. This evidence has led to gaze models based on independent eye and head controllers. However, they cannot determine the end of the gaze shift if the onset and speed of the head movement vary. We propose a model of gaze saccades based on two separate controllers, one for gaze and one for head, but no separate controller for eye. This model extends that of Lefevre et al. (1998) by introducing a duplicate structure in the cerebellum (CB) for controlling the head. The core of the model is a CB controller that monitors movement progress by integrating velocity feedback signals. Eye velocity feedback comes from an efference copy of the eye motor command, and head velocity feedback comes from semicircular canal afference. Gaze velocity (the sum of eye and head velocity) feeds back to the gaze-CB. Head velocity feeds back to the head-CB. The gaze-CB drives the eye and the head-CB drives the head. These drives complement the common gaze drive signal, shared by eye and head, which comes from the superior colliculus (SC). The SC and the gaze-CB receive an input related to the desired change in gaze. The head-CB receives a desired change in head position signal that depends on movement context. During a gaze shift, the VOR is turned off and the gaze-CB provides a drive compensating for the variability in the head trajectory. When gaze reaches the target, the VOR is restored and stabilizes gaze despite any remaining head movement. Although the head contributes to gaze in this model, the head can have an independent goal, because any variability in its trajectory is treated as a perturbation by the gaze controller. We studied a mother and daughter with virtually continuous oscillations (tiny saccades with no intersaccadic interval) of both eyes. 3D coil recordings of the mother found largely conjugate horizontal, vertical and torsional components. When looking straight-ahead (even in darkness) she had tiny saccadic oscillations (20 Hz, amplitude: 0.05-0.1?, speed: 7.6?/s horizontal (H), and 10.3?/s vertical (V); torsional (T) was even smaller). During fixation, eye closure and blinks evoked larger oscillations (~18 Hz, ~95?/s H, ~40?/s V, and ~50?/s T). During 20? vertical saccades, H oscillations increased more than ten fold (1.2?, at 156?/s). During 20? horizontal saccades, V oscillations increased five fold (0.6? at 69?/s). Frequencies remained ~20 Hz. Vertical oscillations were often skewed (one eye up, the other down). Overall, the frequency of the microsaccadic oscillations changed little, despite large variations in amplitude. We previously explained episodic horizontal microsaccadic oscillations in normal subjects with a model of the reciprocal feedback connections between excitatory and inhibitory premotor burst neurons (EBN and IBN) with post-inhibitory rebound. That model can also explain our patient?s disorder if two parameters are modified: 1) weaken the inhibitory projection of the omnipause neurons (OPN) onto IBN, and 2) strengthen the inhibitory projection of the IBN onto EBN. During fixation, OPN are on but do not completely inhibit IBN. The reciprocal inhibition of left and right side IBN form a positive feedback loop and produce microsaccadic oscillations. During OPN inhibition, with no saccade command to the EBN (as during blinks or orthogonal saccades), oscillations become larger because the EBN join the IBN in the positive feedback loop and also drive the motor neurons. The familial nature of this disorder suggests an isolated hereditable change to premotor saccadic neurons, primarily localized to the IBN.