On biological principles of motor control

We review some general problems underlying sensorimotor transformations for biological control of limb movements. We present evidence that limb kinematics can be controlled independently of kinetics. Ill-defined inverse transformations from endpoint to joint coordinates are solved by means of kinematic constraints, such as a law of planar inter-segmental co-ordination, or by means of optimization principles. Hybrid feedback/feedforward control schemes are used whenever possible. Finally internal models mapping motor commands onto their sensory consequences and viceversa are used to improve estimates and to learn new tasks.