The project will develop real-time neural network models for the adaptive control and planning of complex sensory-motor skills. Such models are equally important for understanding human movement planning and the design of flexible characterized circuits for generating synchronous movements of multi-joint motor systems, such as arms and speech articulators, at variable speeds. The current research will investigate the design of supplementary circuits that are capable of rapidly adapting to the unexpected loads and inertias that arise during novel movements, and of slowly learning to predict the correct movement parameters when these novel movements become familiar through practice. In addition, the project will attemp to analyze how elementary, or atomic, movements are organized into complex movement sequences, and how the controls for these sequences become automated through learning. Also, eye-hand coordination will be analyzed by showing how a representation of where the eye-head system is looking is associatively transformed into where the hand-arm system is commanded to move. Issues concerning the learned timing of actions and self-organization of a body-centered coordinated system will also be analyzed.
