Optimization and Variability of Motor Behavior in Multifinger Tasks: What Variables Does the Brain Use?

We address the nature of unintentional changes in performance in two papers. This first paper tested a hypothesis that unintentional changes in performance variables during continuous tasks without visual feedback are due to two processes. First, there is a drift of the referent coordinate for the salient performance variable toward the actual coordinate of the effector. Second, there is a drift toward minimum of a cost function. We tested this hypothesis in four-finger isometric pressing tasks that required the accurate production of a combination of total moment and total force with natural and modified finger involvement. Subjects performed accurate force-moment production tasks under visual feedback, and then visual feedback was removed for some or all of the salient variables. Analytical inverse optimization was used to compute a cost function. Without visual feedback, both force and moment drifted slowly toward lower absolute magnitudes. Over 15s, the force drop could reach 20% of its initial magnitude while moment drop could reach 30% of its initial magnitude. Individual finger forces could show drifts toward both higher and lower forces. The cost function estimated using the analytical inverse optimization reduced its value as a consequence of the drift. We interpret the results within the framework of hierarchical control with referent spatial coordinates for salient variables at each level of the hierarchy combined with synergic control of salient variables. The force drift is discussed as a natural relaxation process toward states with lower potential energy in the physical (physiological) system involved in the task.

This brief review addresses two major aspects of the neural control of multi-element systems. First, the principle of abundance suggests that the central nervous system unites elements into synergies (co-variation of elemental variables across trials quantified within the framework of the uncontrolled manifold hypothesis) that stabilize important performance variables. Second, a novel method, analytical inverse optimization, has been introduced to compute cost functions that define averaged across trials involvement of individual elements over a range of values of task-specific performance variables. The two aspects reflect two features of motor coordination: (1) using variable solutions that allow performing secondary tasks and stabilizing performance variables; and (2) selecting combinations of elemental variables that follow an optimization principle. We suggest that the conflict between the two approaches (a single solution vs. families of solutions) is apparent, not real. Natural motor variability may be due to using the same cost function across slightly different initial states; on the other hand, there may be variability in the cost function itself leading to variable solutions that are all optimal with respect to slightly different cost functions. The analysis of motor synergies has revealed specific changes associated with atypical development, healthy aging, neurological disorders, and practice. These have allowed formulating hypotheses on the neurophysiological mechanisms involved in the synergic control of actions.

In this study, we investigated deficits in coordination of trunk muscle modes involved in the stabilization of the trunk's trajectory for reaching upward and downward beyond functional arm length. Trunk muscle activity from 10 stroke survivors (8 men, 2 women; 64.1 ± 10.5 years old) and 9 healthy control subjects (7 men, 2 women; 59.3 ± 9.3 years old) was analyzed. Coordination of trunk muscle modes to stabilize the trunk trajectory was investigated using the uncontrolled manifold (UCM) analysis. The UCM analysis decomposes the variability of muscle modes into good and bad variability. The good variability does not affect the control of trunk motion, whereas the bad variability does. In stroke survivors, deficits in the ability to flexibly combine trunk muscle modes was associated with reduced ability to minimize those combinations of trunk muscle modes that led to an error in trunk trajectory (bad variability), and this had a greater effect on reaching upward. This reduced coordination of trunk muscle modes during reaching was correlated with a clinical measure of trunk impairment.

This study involved an optimization, numerical analysis of a network for two-hand multi-finger force production, analogous in its structure to the double-representation mirror image (DoReMi) network suggested earlier based on neurophysiological data on cortical finger representations. The network accounts for phenomena of enslaving (unintended finger force production), force deficit (smaller force produced by a finger in multi-finger tasks as compared to its single-finger task), and bilateral deficit (smaller forces produced in two-hand tasks as compared to one-hand tasks). Matrices of connection weights were computed, and the results of optimization were compared to the experimental data on finger forces during one- and two-hand maximal force production (MVC) tasks. The network was able to reproduce the experimental data in two-hand experiments with high accuracy (average error was 1.2 N); it was also able to reproduce findings in one-hand multi-finger MVC tasks, which were not used during the optimization procedure, although with a somewhat higher error (2.8 N). Our analysis supports the feasibility of the DoReMi network. It suggests that within-a-hand force deficit and bilateral force deficit are phenomena of different origins whose effects add up. Is also supports a hypothesis that force deficit and enslaving have different neural origins.

A formal and data-based comparison of measures of motor-equivalent covariation

Motor abundance and control structure in the golf swing

We investigated the muscle activation patterns and the center of pressure (COP) displacement in stepping behavior to determine the relations between anticipatory synergy adjustments (ASAs) and anticipatory postural adjustments (APAs) during support surface translation. Surface muscle activity of eleven leg and trunk muscles was analyzed to identify sets of four muscle modes (M-modes). Linear combination of M-modes and their relationship to changes in the COP shift in the anterior–posterior (AP) direction were then determined. Uncontrolled manifold (UCM) analysis was performed to determine variance components in the M-mode space and indices of M-mode synergy stabilizing the COP shift. Prior to the step initiation, synergies stabilizing COP were seen in both conditions. The synergy index started to drop before a change in the averaged activation levels across trials in postural muscles. The magnitude of synergy index was significantly larger under the perturbation condition. Results of the study have revealed that the central nervous system is able to prepare multi-muscle synergies when a step is performed during support surface translation. Prior to APAs, ASAs reduce stability of COPAP coordinate that is to be adjusted during the APAs. These findings may help get closer to understanding of physiological mechanism of postural preparation to external perturbation.

Handwriting: Hand–pen contact force synergies in circle drawing tasks

The use of flexible arm muscle synergies to perform an isometric stabilization task

A core feature of joint action is task-sharing-the dynamic coordination of individual efforts to produce and stabilize a shared outcome. Yet, it remains unclear which features of task-sharing dynamics relate to stable performance. In this study, we address this gap by identifying behavioral markers that capture the organization and robustness of task-sharing in a joint force-production task. Thirty dyads performed a continuous force-production task in which each partner independently applied force to jointly match a target total force. Real-time visual feedback was provided on their combined output. The task was completed under low-force (10% maximum voluntary contraction) and high-force (30% maximum voluntary contraction) conditions to manipulate difficulty. Trials revealed qualitatively distinct patterns of force contribution, including cases in which a single partner carried most of the output, patterns in which partners contributed with distinct temporal roles, and fully synergistic task-sharing in which both partners continuously shaped total force fluctuations. We used within-trial uncontrolled manifold (UCM) analysis to characterize the dominant, synergistic mode along two dimensions. The first is the degree of variation in task-sharing patterns that leaves performance unchanged (VUCM). The second is the degree of irregularity-computed as sample entropy-of these patterns (SampEnUCM). Dyads with lower VUCM and lower SampEnUCM demonstrated greater performance stability. VUCM and SampEnUCM thus serve as indices of robust task-sharing patterns, offering behavioral access to the control mechanisms that enable stable joint performance, particularly under more challenging conditions.

The link between visual information and postural control was investigated based on a multi-degree-of-freedom model using the framework of the uncontrolled manifold (UCM) hypothesis. The hypothesis was that because visual information specifies the position of the body in space, it would couple preferentially into those combinations of degrees of freedom (DOFs) that move the body in space and not into combinations of DOFs that do not move the body in space. Subjects stood quietly in a virtual reality cave for 4-min trials with or without a 0.2, 2.0Hz, or combined 0.2 and 2.0Hz visual field perturbation that was below perceptual threshold. Motion analysis was used to compute six sagittal plane joint angles. Variance across time of the angular motion was partitioned into (1) variance associated with motion of the body and (2) variance reflecting the use of flexible joint combinations that keep the anterior-posterior positions of the head (HD(POS)) and center of mass (CM(POS)) invariant. UCM analysis was performed in the frequency domain in order to link the sensory perturbation to each variance component at different frequencies. As predicted, variance related to motion of the body was selectively increased at the 0.2-Hz drive frequency but not at other frequencies of sway for both CM(POS) and HD(POS). The dominant effect with the 2.0-Hz visual drive also was limited largely to variance related to motion of the body.

Exploring the Concept of Iso-perceptual Manifold (IPM): A Study of Finger Force-Matching Tasks

Intra-limb joint coordination measures of upper limb and hand movements: A systematic review

Utilizing hierarchical redundancy for accurate throwing movement.

Community dwelling stroke survivors most often fall while walking. Understanding how post-stroke individuals control mediolateral footpath during walking may help elucidate the mechanisms that contribute to walking instability. By applying the Uncontrolled Manifold (UCM) approach, we investigated (1) how post-stroke individuals coordinate lower-extremity joint motions to stabilize mediolateral footpath of the swing leg, and (2) how the inter-joint coordination in footpath stabilization correlates to their walking stability. Nine stroke subjects and nine healthy controls walked on a treadmill at four different speeds. UCM analysis partitions the variance of kinematic configurations across gait cycles into “good variance” (i.e., the variance component leading to a consistent footpath) or “bad variance” (i.e., the variance component leading to an inconsistent footpath). We found that both groups had a significantly greater “good” than “bad” variance (p<0.05) for most of the swing phase, suggesting that mediolateral footpath is an important variable stabilized by the central nervous system during walking. Stroke subjects had significantly greater relative variance difference (ΔV) (i.e. normalized difference between “good” and “bad” variance) (p<0.05), indicating a stronger kinematic synergy in footpath stabilization, than the controls. In addition, the kinematic synergy in mediolateral footpath stabilization is strongest during mid-swing but weakest during late swing in healthy gait. However, this phase-dependent strategy is preserved for mid-swing but not for late swing in stroke gait. Moreover, stroke and healthy subjects demonstrated different relationships between UCM and walking stability measures. A stronger kinematic synergy in healthy gait is associated with better walking stability whereas having more “good variance” or stronger kinematic synergy in stroke gait is associated with less walking stability. The current findings suggest that walking with too much “good variance” in people following stroke, despite no effect on the footpath, may adversely affect their walking stability to some extent.