Strategies used by the central nervous system (CNS) for muscle recruitment and solving the question of abundancy are not still fully understood. Many observations suggest that the CNS overcomes the complexity of abundant number of muscles to be controlled using a dimension reduction policy based on developing muscle synergy groups. This will result in a modular control strategy, which is assumed to make the controlling task easier for the CNS. An important question in this field is how the synergy patterns may change during learning a new task. In this work, we assessed the effectiveness of modularity in describing muscle activity changes during learning. For this purpose, we designed a set of experiments comprising of two drawing tasks of spiral and circle based on tracking predefined patterns, on horizontal plane, by non-dominant arm. The drawing tasks were repeated in 5 different sessions (each session on a separate day) to observe the effect of training on learning. EMG signals from eight muscles of the non-dominant upper limb and the actual trajectory of the pen attached to the hand during drawing were collected. Data were recorded from six healthy participants. For kinematics evaluation of motor learning, the Inverse Efficiency Score (IES) was used in a different way compared to its original defined context, and it's decreasing trend indicated that learning has occurred. In addition, for evaluating the effect of motor learning on muscle activities, space-by-time decomposition model (unified method [1]) was applied to extract spatial and temporal synergies at the same time. Using the Variance Accounted For (VAF) criteria, four spatial and temporal synergies were the minimum necessary number of synergies necessary to re-generate the the EMG's. To study the effect of practice/learning, changes in synergy components over training sessions were evaluated. For all participants, in the last session, spatial synergy modules have become more similar(increasing trend with r2 = 0.9125). On the other hand, the temporal synergy modules, which represent the pattern of time in the EMG data, indicates more rhythmicity of the movement in the last session. The learning effect on the coefficient matrix was measured by the Pearson Correlation (PC) index; increasing trend of this index indicates that the coefficient matrix is converging to a constant matrix by passage of time.
展开▼