A neuro-mechanical transducer model for controlling joint rotations and limb movements

Here we report on the development of an integrated general model for the control of limb movements. The model computes muscle forces and joint rotations as functions of activation signals from motoneuron pools. It models the relationship between neural signals, muscle forces and movement kinematics by taking into account how the discharge rates of motoneuron pools and the biomechanical characteristics of the musculoskeletal system affect the movement pattern that is produced. The lengths and inertial properties of limb segments, muscle attachment sites, the muscles' force-length, force-frequency and force-velocity (of contraction) relationships, as well as a load parameter that simulates the effect of body weight are considered. There are a large number of possible ways to generate a planned joint rotation with muscle activation. We approach this "overcompleteness problem" by considering each joint to be controlled by a single flexor/extensor muscle pair and that only one of the two muscles is activated at a given time. Using this assumption, we have developed an inverse model that provides discharge rates of motoneuron pools that can produce an intended angular change in each joint. We studied the sensitivity of this inverse model to the muscle force-length relationship and to limb posture. The model could compute possible firing rates of motoneuron pools that would produce joint angle changes observed in rats during walking. It could also compare motoneuron activity patterns received for two different hypothetical force-length relations and show how the motoneuron pool activity would change if joints would be more flexed or extended during the entire movement.