Sensorimotor Integration: Models

Our brain's remarkable ability to solve very complex mechanical problems is evident whenever we execute a movement; from picking up a cup of coffee to skating in an ice-rink. Several approaches have been suggested to address the complexities of biological motor-control systems. A prominent approach describes the representation of the body-environment dynamics in terms of transformations (internal models); both from muscle activities to movement ('forward model'), and from desired movement to muscle activities ('inverse model'). These internal models are equivalent to the computation of dynamics' equations but are based on elementary units, or primitives, that execute simple operations and that can be combined to generate a broader repertoire of behaviors. This article reviews some of the experimental evidence and theoretical arguments supporting the existence of such primitives and their role in the learning and generation of motor behavior.