Representing Reach-to-Grasp Trajectories Using Perturbed Goal Motor States

In the biological system which controls movements of the hand and arm, there is no clear distinction between movement planning and movement execution: the details of the hand’s trajectory towards a target are computed ‘online’, while the movement is under way. At the same time, human agents can reach for a target object in several discretely different ways, which have their own distinctive trajectories. In this paper we present a method for representing different reach movements to a target without reference to full trajectories: movements are defined through learned perturbations of the hand’s ultimate goal motor state, creating distinctive deviations in the hand’s trajectory when the movement is under way. We implement the method in a newly developed computational platform for simulating hand/arm actions.