Figures and data
Framework and experiment setup.
(A) A patient exhibiting a typical flexor posture at rest. Dashed arrows indicate elements of the posture: shoulder depression, arm adduction/internal rotation, elbow flexion. If one were to physically constrain the hand in a position away from the resting posture, the torques involved in each component of the abnormal resting posture translate to a force on the hand (sketched as a blue arrow); we thus designed an experiment to measure the resting force bias on the hand, as a marker of the overall postural abnormality. The goal was to compare resting postural force biases to active movement control in the same area (B). (C) Experiment setup. The participant holds the handle of the robotic arm; reach targets and cursor position are projected on a screen on top; for arm support, the participant’s arm is strapped on an armrest (c) connected to an air sled (a) which rests on the table. Air is provided through tube labeled (b). (D) Top-down view of setup, illustrating the different hand positions where resting postural forces were measured in Experiment 1 (open circles). Also shown are the five target positions used in the reaching and holding task for Experiment 2 (filled red circles). The gray box indicates the workspace depicted in Figure 2. (E) Examples of measurements of resting force biases. Each panel shows the evolution of resting forces during the 5s holding period for one participant (same participants as in Figure 2), taken at positions close to midline and distal from the body, under the same condition (paretic arm, arm support given). Solid lines indicate the force on the x-axis (positive values indicate forces towards the left), whereas dashed lines indicate the force on the y-axis (positive values indicate forces towards the body). The shaded area indicates the time window over which forces were averaged to estimate the resting bias, illustrating how resting biases were relatively stable by the 2s mark. Note that the third panel includes one trial (blue) which was rejected following visual inspection as described in Materials and Methods – Data Exclusion Criteria, due to instability in force production and movement during the hold period.
Patient characteristics.
FM-UE: Fugl-Meyer Assessment for the Upper Extremity; ARAT: Action Research Arm Test.
Summary of patient and control characteristics.
FM-UE: Fugl-Meyer Assessment for the Upper Extremity (/66); ARAT: Action Research Arm Test (/57). MoCA: Montreal Cognitive Assessment (/30).). Here, ± indicates standard deviation.
Examples of resting postural force biases.
Shown are three stroke patients and one healthy control. Arrows indicate magnitude and direction of abnormal resting postural forces as measured at the hand at each location. Isoclines and the corresponding color levels provide a visual representation of where these biases would tend to direct the hand (from red towards blue). These isoclines represent different levels of the (spatial) integral of the posture bias-field (with zero referring to the isocline passing through the center position). Postural bias force vectors cross these isoclines perpendicularly. Please see Supplementary materials on how this visual aid was constructed. The red dots are the reach targets, with the center location circled (used in Experiment 2). Note how abnormalities in the paretic side are considerably stronger when arm support is removed. FM-UE: Fugl-Meyer score for the Upper Extremity (0-66).
Average resting postural force biases.
(A) Average resting postural forces for the paretic (red) and non-paretic (cyan) arms of patients, as well as control participants (gray), illustrating how abnormal forces in the paretic arm are stronger in more distal targets and attenuated when arm support is provided (lighter shades). To average across left- and right-hemiparetic patients, left-arm forces were flipped left to right. (B) corresponding average resting postural force magnitudes and resting postural force directions (0 indicating the 3 o’clock direction, with negative values indicating clockwise directions). Errorbars indicate mean±SEM (circular mean±SEM in the case of movement directions).
Three aspects of active motor control that we tested in Experiment 2.
We separately examined the early part of the reaching movement (blue) and the late part, when the arm was coming to a stop (red). This was done by studying both unperturbed movements at different stages and movements that were perturbed with brief force pulses. In addition, we examined active holding control after the movement was over (black), using perturbations that tried to move the arm away from the held point. Shown is an example of trajectory and (absolute) velocity profiles from the reaching and coming-to-a-stop parts of a trial (left) and active holding against a perturbation after the trial was over (right).
Abnormal resting postural force biases do not interact with active reaching.
(A) Target array for Experiment 2 (movement task), illustrating the 5 start/end points of reaches and the 8 movement directions. (B) Example outwards trajectories (unperturbed trials) for a patient (cyan: non-paretic side; red: paretic side) and a healthy control (gray). (C) Subject-averaged reach performance based on either time (top) or path length to target (bottom) indicates impaired reaching control in patients’ paretic side. (D-F) Within-subject analysis of whether resting postural forces at movement start bias early movement towards their direction. (D) Sketch illustrating the concept behind this analysis. Assuming a movement from a start position subject to a strong rightwards resting bias (Fstart), will that translate to a corresponding rightwards movement bias which can be expressed as the early reach angle θstart,? (E) For each individual, we selected the target direction for which the counter-clockwise (CCW) component of Fstart was the strongest (red) vs. the target direction for which the clockwise (CW) component was the strongest (blue). The left panel shows this selection for an example participant: postural forces at start position were projected lateral to the movement direction, allowing us to select movement directions for which the lateral component was directed the strongest CCW or CW. The right panel shows the magnitude of these selected components across all patients. (F) Left: Corresponding movement trajectories (rotated so start position is at the bottom and end position at the top) for the directions selected for the same example participant. Right: Average initial angular deviations, θstart, across the selected directions for each participant. Note the lack of difference between the instances for which the CCW vs. the CW component of Fstart was the strongest, and thus no evidence that Fstart impinges upon the movement. (G-I) same as D-F but for endpoint resting postural forces, Fend and endpoint deviations, θend. Errorbars indicate SEM.
Responses to pulse perturbations during movement are not affected by resting postural force biases.
(A) Examples of perturbed (red: perturbed with CCW pulse; blue: perturbed with CW pulse) and unperturbed (gray) outward trajectories - same individuals as in Figure 3B. (B) Lateral velocity (positive: CCW to movement) before and after pulse onset, and corresponding responses from controls (gray), illustrating how patients, in response to the pulse, take longer time to settle and tend to experience larger lateral deviations compared to controls. (C) Summary performance measures for patients and controls, indicating impaired performance with the paretic side: settling time (left) and maximum lateral deviation on pulse direction (right). (D) Within-individual analysis: here, for each individual, we selected the movements for which the starting-position resting postural force would be either the strongest CCW or CW (left); we then examined the corresponding settling time (middle) and maximum lateral deviations (right). We find no effects of the most CCW vs. most CW resting postural forces in either case: there is no evidence for either reduced settling time or increased maximum lateral deviation for instances where pulse and resting bias are most opposing (open circles) compared to the instances where pulse and resting bias are most aligned (filled circles).
Responses to release perturbations during holding and their relationship to resting postural forces.
(A) Time course of the perturbation. (B) Example responses (all for the same position in the workspace) from two patients (top row) and two controls (bottom row). (C) Corresponding imposed force directions, the abrupt removal of which perturbs the movement in the opposite direction (compare with B). (D) Examples of tangential velocity profiles after the sudden release to the imposed hand force, averaged for all trials at the same position for each participant. Dashed line indicates the 2cm/s threshold used to assess time to stabilize. Left; example patient (paretic side); Right; example control. Colors correspond to different directions of the imposed hand force. (E) Summary of performance metrics after the perturbation for the paretic and non-paretic side of patients and healthy controls. (F) Within-subject analysis of the relationship between resting postural forces in the direction of the perturbation vs. performance against the perturbation. For each individual, we selected the two position/perturbation direction combinations for which resting postural forces were either the most opposed (green) to the perturbation or the most aligned (blue) with it. From left to right: forces in selected position/perturbation direction combinations; corresponding path traveled to stabilization; corresponding time to stabilization; corresponding maximum deviation. Note how the most-opposed resting bias for each patient is equal and opposite to the their most-aligned resting bias. This is because the same resting bias, when projected along the direction of two oppositely-directed perturbations (illustrated in C), it would oppose one with the same magnitude it would align with the other. This analysis suggests that resisting postural perturbations and restoring hand position after the perturbation was indeed easier when resting postural forces opposed, rather than were aligned with, the perturbation. Gray dots indicate individual data; colored dots and errorbars indicate mean±SEM.
Relationship between resting force biases and abnormal synergies.
Across-patient relationships of FM-UE (/66, higher scores indicating lower impairment) and resting postural force magnitudes, for distal (green) and proximal (blue) target positions, with (left) and without support (right). Note the strong effects of arm support, proximity, and FM-UE. Lines indicate linear fits; shading indicates 95% confidence interval for each fit.
