Neural control of eye movement

Duke University, Durham NC

Investigators

Linked publications & trials

Abstract

One of the important functions of visual inputs is to guide motor activity, especially eye movements. Primates can used smooth pursuit eye movements to rotate their eyes smoothly to keep them pointed at small objects that are moving slowly and smoothly, a function that is much more poorly expressed in most other species. Moving objects cause neural activity in a part of the visual cortex called the "middle temporal visual area" or MT, where neurons respond only to moving objects and encode the direction and speed of object motion. MT provides the visual inputs for smooth pursuit eye movements, but speed and direction are encoded only in the response of a large number of MT neurons, and not in the responses of any individual neuron. Therefore, the population response in MT needs to be "decoded" to provide motor commands that indicate the required direction and speed of smooth eye velocity. This proposal uses awake, behaviorally-trained rhesus monkeys to ask how visual population decoding is done by neural circuits. First, it proposes to use a combination of recordings of the first 100 ms of pursuit eye movements and the electrical activity of MT neurons to ask whether certain sub-populations of MT neurons are selectively involved in pursuit. The approach is to characterize the effect of changes in stimulus form on the variation in eye movement, and then to determine whether the parallel effects on the responses of certain sub-populations of MT neurons vary in a way that is appropriate to cause the effects on eye movement. Second, the proposal will use a combination of experiment and computation to characterize how visual population responses are decoded using biologically-plausible models of neural circuits. The approach will be to create a library of different circuit models that make different predictions for the responses of doable experiments. The experiments consist mainly of assessing the correlations between the trial-by-trial variations of MT responses and pursuit eye movements. Special attention will be paid to how the MT-pursuit correlations vary as a function of the preferred speed of an MT neuron, a feature that is particularly sensitive to the details of how the neural circuit decoding model is implemented. An interplay between computation and experiment will constrain severely the neural mechanisms that are used for visual population decoding in the brain.