Relationship between mechanical cost functions and muscle oxygen consumption

In biomechanical studies inverse-dynamic models are often used to predict muscle forces. The distribution of those forces among the muscles is often calculated by optimising a cost function, assuming that movements are controlled in an optimal manner. Unfortunately the quantity optimised in real life is unknown and only assumptions can be made. Many different cost functions have been proposed (see Tsirakos et al. 1997 for an overview). Although a few are based on physiological reasons, most cost functions are chosen rather arbitrarily. Non-linear cost functions provide physiologically more realistic results than linear cost functions, since linear cost functions predict sequential recruitment of muscles instead of load sharing. (Tsirakos et al. 1997).