Toward an improved control theoretic metaphor for the saccadic system

Matthew James Nichols, University of Pennsylvania

Abstract

Saccades are very rapid, conjugate rotations of the eyes for the purpose of redirecting the line of sight. The high accuracy required to bring images onto the relatively tiny fovea depends on internal (non-sensory) feedback. In an early local feedback model of saccade control (eye position model), an outgoing saccadic velocity command is integrated mathematically by a neural integrator to provide a representation of the actual position of the eyes in the orbits (head-centered reference frame). This signal is fed back to a comparator where it is subtracted from a reference signal of desired orbital position. But because saccadic goals are initially specified in an eye-centered reference frame (retinal error), and because the superior colliculus (SC)--a primary structure for encoding the reference signal for the feedback controller--specifies desired change in eye position rather than desired orbital position, it was subsequently proposed that the downstream controller might operate in an eye-centered rather than head-centered reference frame. If the controller does operate in eye-centered coordinates, then the feedback integrator must be reset to zero after saccades. Unlike the eye position model, such a controller is non-stationary during the resetting time. Normally, saccades occur with sufficient temporal separation to mask this difference between the models. But we injected known reference signals during the resetting of the integrator and revealed the non-stationary behavior expected from an eye-centered controller with a resettable integrator. Our results are clearly inconsistent with the predictions of the eye-position model. It was possible to exploit this non-stationary behavior to test differential predictions of other classes of models for controlling oblique saccades in two-dimensions. Because the non-stationary behavior can be elicited in either the horizontal or vertical component of oblique saccades without influencing the other component, the component amplitudes of oblique saccades must be under independent feedback control. Finally, because these redirected saccades display appropriate component stretching despite the fixed locus of SC stimulation, component stretching must be achieved by mechanisms downstream of the SC.

Subject Area

Neurology

Recommended Citation

Nichols, Matthew James, "Toward an improved control theoretic metaphor for the saccadic system" (1996). Dissertations available from ProQuest. AAI9627974.
https://repository.upenn.edu/dissertations/AAI9627974

Share

COinS