Conventional models of motor control exploit the spatial representation of the controlled system to generate control commands.\nTypically, the control command is gained with the feedback state of a specific instant in time, which behaves like an optimal\nregulator or spatial filter to the feedback state. Yet, recent neuroscience studies found that the motor network may constitute an\nautonomous dynamical system and the temporal patterns of the control command can be contained in the dynamics of the motor\nnetwork, that is, the dynamical system hypothesis (DSH). Inspired by these findings, here we propose a computational model that\nincorporates this neural mechanism, in which the control command could be unfolded from a dynamical controller whose initial\nstate is specified with the task parameters. The model is trained in a trial-and-error manner in the framework of deep deterministic\npolicy gradient (DDPG). The experimental results show that the dynamical controller successfully learns the control policy for arm\nreaching movements, while the analysis of the internal activities of the dynamical controller provides the computational evidence\nto the DSH of the neural coding in motor cortices.
Loading....