Control

Our understanding of motor control has been limited by models that oversimplify the complexity of arm and hand control, which requires modeling approaches that take into account the complexity of muscle dynamics, delays and noise inherent in biological systems, sensory feedback arising during movement, and the complementary roles of different brain regions.

We are developing new neural network models of closed-loop control that integrate all of these crucial factors, which will help us understand how brain regions coordinate to support reaching and grasping movements. The next step of this work will link the control strategies of these artificial networks to neural populations recorded from non-human primates performing skilled behaviour.

Neural network model trained perform sequential reaches by controlling simulated muscles (related paper)