Figure 3.

General model structure. The focus of the model is on the network formed by the superior colliculus (SC) and the omnipause neurons (OPNs) and burst neurons (BNs) in the brain stem and how they generate microsaccades in the presence of noise. Thus, the SC and the brain stem reticular formation are the only elements modeled as a network of neurons. All the other elements are “lumped” elements that just attempt to simulate a certain transfer function. The SC receives two complementary inputs indicating the desired gaze location in retinotopic coordinates, one excitatory from frontal eye field (FEF) and one inhibitory from basal ganglia (BG). It also receives a noise input that represents all the possible sources of noise the SC may receive. The SC is a network of 101 interconnected neurons with short-range excitatory connections and long-range inhibitory connections. The output of the SC that drives the BNs includes a nonlinearity that gives more weight to neurons firing more. The SC also drives the OPNs, especially when activity is centered around the center of the SC (representing the rostral pole). The model also includes a simple model of a cerebellar circuit that would produce a stop signal to reset both the SC and the BNs after a saccade. The output of the brain stem drives the eye plant that accounts for the mechanical properties of the eye. The resulting changes in eye position are fed back to the input of the model with a delay of 50 ms.