A Neural Network Model for Cortical Representation of Movement

Bagrat Amirikian
Department of Physiology
University of Minnesota

The direction and the velocity of an intended limb movement are represented in the motor cortex as the direction and the length of the neuronal population vector that is a vector sum of activities of derectional tuned cortical neurons. The speaker proposed a new physiologically relevant neural network model that describes directional operations in the motor cortex. The massive and fully recurrent interconnections between network elements ensured a complex dynamics of the system. The dynamic evolution toward an attractor of the system modeled motor cortical processing that controlled real upcoming movement. The spatio-temporal features of neural-vector trajectories revealed by the model were in agreement with neurophysiological data.