Bio-mechanical characterisation of synergistic muscle control in goal directed movements

Abstract

The generation of skilled movement requires the coordinated activation of muscles to produce precise muscle forces and joint torques. The co-ordination of muscles, acting on multi-joint limbs and characterised by highly non-linear dynamics, requires significant computational effort. The synergistic control hypothesis suggests that the nervous system may simplify the required computations by producing movements through the coordinated activation of a limited number of movement building blocks, termed muscle synergies. To evaluate how synergistic control may be used to produce the variety of kinematic behaviours observed in vivo, we developed two algorithms to identify sets of muscles whose linear combinations can be used to reproduce the range of joint-torque profiles recorded in vivo. Sets of two to three synergies accounted for 71.9% to 89% of the variance in the torque data--depending on the time-structure imposed on the synergies. Small numbers of synergies were sufficient to reconstruct the data because of the high degree of correlation between the hip and knee joint-torques required to produce both natural and artificial behaviours. Dynamic properties of the limb act to increase the correlation between hip and knee extension torques, and between hip and knee flexion torques. The structure of the musculoskeletal system of the limb and the structure of the extracted muscle synergies both act to produce muscle-torque combinations that counter- act the dynamic properties of the limb.