Incorporating gravity into synergistic control of upper limb movements using phasic synergies with positive and negative weights.

Two models have been proposed to describe how motor control is affected by gravity. According to the gravity-compensation model, accelerating and decelerating the limb through phasic muscle activations is independent of the control of gravity forces, with tonic muscle activations counteracting gravity force. The effort-optimization model, instead, hypothesizes that muscles exploit gravity, decreasing tonic activity to minimize effort using negative phasic EMG components. Muscle synergies have been used for assessing motor control in neurophysiological studies, but synergistic models so far have neglected explicit representations of gravity forces. Therefore, we aimed at incorporating the pervasive presence of gravity into muscle synergies by extracting synergies with negative weights to capture negative phasic EMG components. Muscle synergies with positive and negative weights were extracted using the mixed-matrix factorization (MMF) algorithm on a set of upper limb reaching movements performed by 15 healthy participants across targets in different planes designed to elicit positive and negative phasic activations. Movements were grouped depending on the tonic components at movement onset, needed for gravity exploitation, and identified as "increasing tonic EMG" (ITE) and "decreasing tonic EMG" (DTE). ITE showed better reconstruction accuracy than DTE when extracting five or fewer synergies. DTE exhibited more negative phasic activations and synergy weights showed more negative values. A bootstrap procedure showed that synergies extracted from ITE and DTE are different in structure, and cluster analysis found nine clusters for ITE and ten for DTE. These results indicate that compensation and effort minimization models can coexist within the muscle synergy framework. For the first time, a novel approach based on muscle synergies with positive and negative weights allows to account for the exploitation of gravity into synergistic models. This is achieved by a synergistic controller that incorporates both simplicity, as a reduced set of synergies underlying movement and static gravity compensation (phasic and tonic synergies), and effort optimization, based on the exploitation of gravity through negative phasic components.