Coordinate Systems in the Motor System: Computational Modeling and EEG Experiment

Abstract

How neurons in the motor system control arm movements is not yet understood completely. Here we show the equations of motion (EOMs) governing reaching simplify when expressed in spatial coordinates, composed of vector cross products between limb position and movement vectors. If these cross products are identified with motor neuron activities, our model explains a wide range of properties such as cosine directional tuning, nonuniformly distributed preferred directions, coexistence of multiple reference frames, and spatiotemporal properties of the population vector. The cross-product basis also explains the generalization patterns in both dynamic and kinematic motor adaptations in human psychophysics. In addition to the modeling approach, we report a noninvasive EEG recording during a pointing experiment and demonstrate that EEG dipoles are directionally tuned. Our integrated approach should clarify the neural computation underlying the visuomotor transformation for upper limb movements.