| In human motor control, feedforward and feedback neural circuits integrate in cortical and subcortical networks of the central nervous system (CNS) to generate movements. For the human hand, this involves controlling over 21 degrees of freedom (DoF). It is hypothesized that rather than directing individual DoFs at each point in time, the CNS may work in a lower dimensional subspace. This subspace is made of a minimal set of synergies, or movement primitives, that describe coordination across multiple DoFs. While it is generally hypothesized that the motor control system works in a lower dimensional synergy subspace, the structure of these synergies and how they may combine is still unknown. This dissertation first explored methods to improve an existing synergy-based motor control model. Linear and nonlinear dimensionality reduction methods were used to derive synergies from grasping tasks. Principal component analysis was found to be an optimal method for extracting meaningful spatiotemporal kinematic patterns, or synergies. Deprivation of sensory feedback (visual and tactile) showed that certain synergies contributed more towards finer movements and later in time. Additionally, modifying synergy duration allowed for better approximations when reconstructing grasps. Electroencephalography (EEG) was then used to record neural signals during grasping. Using a multivariate regression analysis, neural correlates of kinematic synergies were determined. The neural correlates were then used to decode grasp kinematics from EEG, resulting in decoding accuracy above chance level. The use of synergy-based movement models has applications across multiple fields. In motor learning, continuous repetition of a task is considered a gold standard. Training with movement synergies, however, was found to improve task proficiency and translational ability, compared to repetition training and control groups. Such a training paradigm may be valuable to individuals undergoing rehabilitation. While the general synergy patterns derived in these studies were similar across multiple subjects, unique subject-specific properties were still observable. These unique characteristics, at the kinematic synergy level and neural level, showed potential as biometric markers for identity verification systems. The conclusions drawn from this dissertation further the understanding of mechanisms by which synergies may be used in human motor control as well as their potential applications. |
