Impact of muscle redundancy and nonlinearities on the muscle synergy hypothesis : a computational investigation

Muscle synergies is an hypothesis in neuroscience trying to explain how the central nervous system (CNS) is able to efficiently control our body, despite its very high complexity. Muscle synergies are groups of muscle activation patterns which can be linearly combined in order to generate complex muscle actuations. This strategy may simplify motor control because the CNS only has to choose how to combine these predefined patterns instead of synchronizing the individual muscles independently. The classical approach in order to verify the hypothesis has been to record EMG (Electromyographic) signals during the execution of tasks, and try to extract components (i.e. synergies) able to reconstruct the EMG dataset. This approach leaves, however, many questions open. For example we don't know if it is theoretically possible to control a musculoskeletal system by linear combination of synergies, nor we know how the biomechanical properties of the system influence the synergy hypothesis. In this thesis we want to understand if the control by linear combinations of synergies is possible, and to investigate the impact of redundancy and muscle nonlinearities (two important biomechanical features of musculoskeletal systems) on the synergy hypothesis. We modeled the human arm as a 2-joint kinematic chain, we synthesized appropriate synergies, and we measured their performance in solving reaching tasks. We incrementally added the biomechanical features mentioned above: first we actuated the kinematic chain by means of torques applied to the joints (no redundancy); then we actuated the kinematic chain by means of 6 forces applied to the links (redundancy); finally we generated the 6 forces applied to the links by means of nonlinear muscles controlled in activation (redundancy and muscle nonlinearities). Our results suggest that redundancy does not necessarily increase the number of synergies required to execute a task, and therefore that the number of muscles may not affect the complexity of the control. Nonlinearities, coming for example from muscle dynamics, may instead make the job of the CNS harder, and require a higher number of synergies.