Organization of the neural system for pursuit eye movements.
We are using multiple site, multiple single neuron recording to ask how the different nodes of the pursuit system work together to create a complex behavior. We are starting with simultaneous recordings from the sensory area MT and the motor area called the smooth eye movement region of the frontal eye fields. By deploying strategically-chosen behavioral paradigms and data analysis strategies, we can characterize the functional connectivity of the pursuit system and determine how functional connectivity changes as a function of the state of the system. Our long term goal is to extend this approach to the brainstem, the basal ganglia, and the cerebellum to determine the neural implementation of sensory decoding, control of the strength of visual-motor transmission, and target selection.
Neural implementation of an example of Bayesian inference.
We have developed behavioral paradigms to show that the initiation of pursuit is controlled by a Bayesian competition between sensory evidence and adaptable priors for target speed and direction. When the strength of visual motion is strong, sensory evidence defeats priors; when the strength of evidence is weak, the adaptable priors win. We also have found that the entire Bayesian computation is represented in the smooth eye movement region of the frontal eye field, and that adaptation of the prior could occur locally in the same region. Ongoing projects include (1) looking for further implementation of the Bayesian framework in the pontine relays from the cerebral cortex to the cerebellum, (2) analysis of neural circuit organization in the frontal eye fields, and (3) computational modeling to understand how the population response in sensory area MT could be transformed into the observed representation in the frontal eye fields.
Multiple components of learning in pursuit eye movements.
We have discovered that a single change in target direction causes a small change in eye movement in the subsequent trial. An important neural correlation of "single trial learning" occurs in the floccular complex of the cerebellum, where a single complex spike on one learning trial causes a properly-timed depression of simple spike response on the subsequent trial. We now are demonstrating the existence of multiple components of learning that have different time courses of acquisition. The next step will be neural recordings in the floccular complex to determine whether the different components of behavioral learning have different sites and mechanisms of neural learning.
Circuit mechanisms for cerebellar function and learning.
We will be using multiple-contact electrodes to record simultaneously from multiple neighboring single-units in the cerebellar cortex of the floccular complex. Through calculation of spike-timing cross-correlograms with Purkinje cells, we will attempt to determine the anatomical identify of interneuron recordings. If we are able to identify interneurons and record their activity during smooth pursuit eye movements and pursuit learning, then we will be in a position to construct a circuit model of the cerebellar cortex and explain how the circuit works and learns.