LEARNING AND MAINTAINING SACCADIC ACCURACY - A MODEL OF BRAIN-STEM-CEREBELLAR INTERACTIONS

Saccadic accuracy requires that the control signal sent to the motor n eurons must be the right size to bring the fovea to the target, whatev er the initial position of the eyes (and corresponding state of the ey e muscles). Clinical and experimental evidence indicates that the basi c machinery for generating saccadic eye movements, located in the brai nstem, is not accurate: learning to make accurate saccades requires ce rebellar circuitry located in the posterior vermis and fastigial nucle us. How do these two circuits interact to achieve adaptive control of saccades? A model of this interaction is described, based on Kawato's principle of feedback-error-learning. Its three components were (1) a simple controller with no knowledge of initial eye position, correspon ding to the superior colliculus; (2) Robinson's internal feedback mode l of the saccadic burst generator, corresponding to preoculomotor area s in the brainstem; and (3) Albus's Cerebellar Model Arithmetic Comput er (CMAC), a neural net model of the cerebellum. The connections betwe en these components were (1) the simple feedback controller passed a ( usually inaccurate) command to the pulse generator, and (2) a copy of this command to the CMAC; (3) the CMAC combined the copy with informat ion about initial eye position to (4) alter the gain on the pulse gene rator's internal feedback loop, thereby adjusting the size of burst se nt to the motor neurons. (5) If the saccade were inaccurate, an error signal from the feedback controller adjusted the weights in the CMAC. It was proposed that connection (2) corresponds to the messy fiber pro jection from superior colliculus to oculomotor vermis via the nucleus reticularis tegmenti pontis, and connection (5) to the climbing fiber projection from superior colliculus to the oculomotor vermis via the i nferior olive. Plausible initialization values were chosen so that the system produced hypometric saccades (as do human infants) at the star t of learning, and position-dependent hypermetric saccades when the ce rebellum was removed. Simulations for horizontal eye movements showed that accurate saccades from any starting position could be learned rap idly, even if the error signal conveyed only whether the initial sacca de were too large or too small. In subsequent tests the model adapted realistically both to simulated weakening of the eye muscles, and to i ntrasaccadic displacement of the target, thereby mimicking saccadic pl asticity in adults. The architecture of the model may therefore offer a functional explanation of hitherto mysterious tectocerebellar projec tions, and a framework for investigating in greater detail how the cer ebellum adaptively controls saccadic accuracy.