Howard Hughes Medical Institute, University of California, San Francisco
Distributed coding of commands for smooth pursuit eye movements in visual area MT
How are visual inputs converted into commands for accurate movement? For eye movements, smooth target motion is transduced by the visual system, represented in the responses of a large population of neurons in cortical area MT, and then converted into commands for smooth tracking eye movements in the cerebellum and brainstem. Previous models of pursuit have demonstrated that it is possible to reproduce the eye movements of human and monkey subjects on a millisecond time scale if the visual inputs reflect the dynamics of target motion. Specifically, these models used visual inputs related to instantaneous image velocity and image acceleration. MT is a likely candidate to provide these visual inputs for pursuit because almost all neurons in MT are selective for the direction of target motion and lesions of MT disrupt the initiation of pursuit eye movements. We have conducted a series of neural and computational analyses that demonstrate how image velocity and acceleration are represented in cortical area MT and how to reconstruct these variables from the population response in MT by a neurally-plausible computation. First, we demonstrated that MT neurons have transient responses that provide information about target acceleration, along with the well-known sustained responses that represent target velocity. Cells with large transient responses for steps of target speed also had larger responses for smooth accelerations than for decelerations through the same range of target speeds. Second, we created a model that reproduced the transient responses of MT neurons using divisive gain control, and we used the model to visualize the response of the population of MT neurons to individual stimuli. Finally, we devised weighted-averaging computations to reconstruct target speed and acceleration from the population response. Target speed could be reconstructed if each neuron's output was weighted according to its preferred speed. Target acceleration could be reconstructed if each neuron's output was weighted according to the product of preferred speed and a measure of the size of its transient response.