Heuristic versus optimization methods for solving muscle load-sharing problems

It is shown that the reductionist approach for solving the classical muscle load-sharing problem can be viewed as an inflexible heuristic optimization strategy created in the mind of the investigator based on limited information. In general, such techniques, even when carefully based on physiological and biomechanical conditions, are quite limiting. In contrast, optimization methods allow flexible solutions, take advantage of information on muscle dynamics and task history in making decisions on neuromotor strategy, and often provide illuminating solutions. However, these come at great computational cost. Also, the best performance criterion depends on the type of task and on the goal of the subject. A hybrid approach involving lower-level hard-wired heuristics driven by goal-specific optimization is proposed as a good approximation of the neuromotor system.<>