Moving fast : neural constraints in closed loop

Abstract

The generation of fast movements during sensorimotor control is fundamentally limited by the biophysics of neural activity and the physiological dynamics of the muscles involved. Yet, the limiting factors and the corresponding tradeoffs have not been rigorously quantified. We use feedback control principles to identify limitations in the ability of the sensorimotor control system to track intended fast periodic movements. We show that (i) a linear model for movement generation fails to predict known undesirable phenomena encountered in the regime of fast movements, and (ii) the theory of pulsatile control of movement generation allows us to correctly characterize fundamental limitations in this regime. This thesis identifies the fastest periodic movement possible for given musculoskeletal and neuronal dynamics, which has far-reaching implications in sensorimotor control. The use of neuronal decoders in the Brain Machine Interface setting is discussed; we introduce a real-time decoder of neuronal activity, and derive conditions for its stability in the presence of feedback. The framework developed in this thesis allows us to characterize the effect of compromised neural and physiological activity on movement, and guide the design of corresponding therapeutic measures.