Joint angle-torque relations at different levels of muscle activity in maintaining constant position of an upper extremity

While, in general, spring-like property of a muscle has been assumed to account for the position control of extremity, the property has not been firmly proved in human muscles; in particular the static relation between angle and torque under the constant muscle activity that is one of the most fundamental characteristics has not been observed in human voluntary movements yet. The purpose of the preset study is to determine this static relation of human elbow joint with an aid of a neural network model. Three normal volunteers were instructed to maintain their upper extremity at target angle on the horizontal plane by balancing against an external load applied to their wrist. Both joint angle and torque around their elbow and shoulder as well as integrated electromyogram (IEMGs) of five muscles (caput longum and brave bicipitis brachii, triceps brachii, pectralis major and deltoideus) were recorded for five seconds and then time-averaged. The three-layer neural network model was constructed; the inputs were five channels of IEMGs, elbow joint angle, shoulder joint angle, and the outputs were torques of the elbow and shoulder joints. After learning, the elbow joint torque was estimated at various angles of the elbow joint while both IEMGs of five muscles and the shoulder angle were kept constant. Estimated result were as follows. The torque of elbow extensors monotonically increased as flexing the elbow joint, i.e., as elongating extensor muscles. However, the torque of elbow flexors decreased as extending the elbow joint, i.e., as elongating flexor muscles, over an angle range of largely flexed elbow joint. The property of the extensors was in agreement with that of the equilibrium hypothesis, and the property of the flexors was apparently opposite to it.