Object-centered sensorimotor bias of torque control in the chronic stage following stroke

Abstract

When lifting objects whose center of mass (CoM) are not centered below the handle one must compensate for arising external torques already at lift-off to avoid object tilt. Previous studies showed that finger force scaling during object lifting may be impaired at both hands following stroke. However, torque control in object manipulation has not yet been studied in patients with stroke. In this pilot study, thirteen patients with chronic stage left hemispheric stroke (SL), nine patients with right hemispheric stroke (SR) and hand-matched controls had to grasp and lift an object with the fingertips of their ipsilesional hand at a handle while preventing object tilt. Object CoM and therewith the external torque was varied by either relocating a covert weight or the handle. The compensatory torque at lift-off (Tcom) is the sum of the torque resulting from (1) grip force being produced at different vertical finger positions (∆CoP × GF) and (2) different vertical load forces on both sides of the handle (∆Fy × w/2). When having to rely on sensorimotor memories, ∆CoP × GF was elevated when the object CoM was on the ipsilesional-, but decreased when CoM was on the contralesional side in SL, whereas ∆Fy × w/2 was biased in the opposite direction, resulting in normal Tcom. SR patients applied a smaller ∆CoP × GF when the CoM was on the contralesional side. Torques were not altered when geometric cues were available. Our findings provide evidence for an object-centered spatial bias of manual sensorimotor torque control with the ipsilesional hand following stroke reminiscent of premotor neglect. Both intact finger force-to-position coordination and visuomotor control may compensate for the spatial sensorimotor bias in most stroke patients. Future studies will have to confirm the found bias and evaluate the association with premotor neglect.

Figure 1

Experimental apparatus, variables and design. (A) The custom-built grip-device consists of a handle element mounted centrally on a horizontal bar (frontal view). The handle element allowed subjects to freely choose digit placement on the grip surfaces (40 × 120 mm) covered with sandpaper. Two 6-axis-force/torque sensors were mounted under the grasp surfaces. In the ‘no cues’ condition a hidden weight was either placed in the left or right cavity resulting in an external torque after lift-off. The exerted total torque is the sum of the torque components ∆CoP × GF and ∆Fy × w/2 and must compensate for the external torque to prevent object tilt. (B) The recorded experimental variables are illustrated for an exemplary trial, the torque variables at lift off were considered to be indicators of anticipatory torque control. (C) The experimental protocol comprised the two cue- conditions ‘no cues’ in which the center of mass (CoM) was changed by placing a hidden weight either on the left or the right (with the handle being positioned above the middle cavity), resulting in external torque of ± 0.21 Nm after liftoff, and the ‘geometric cues’ condition in which the handle was either mounted above the left or right cavity (with the hidden weight inserted in the central cavity) , resulting in external torque of ± 0.46 Nm after liftoff. The order of the conditions and first CoM side was randomly assigned to participants (see Supplementary Table 1). For each cue-conditions participants first completed a pseudorandom sequence of 37 trials in which the CoM could change from trial to trial and 33 trials in which the CoM stayed constant for 8 trials before it was inverted.

Figure 2

Sensorimotor learning of anticipatory torque compensation. Box and whiskers plots in the style of Tukey (central horizontal line: median, lower, and upper hinges: 25th and 75th percentiles, upper and lower whiskers extend up to 1.5 interquartile ranges) as well as the mean and standard deviation of the ratios of anticipatory torque anticipation success Tcom/external torque (A), ΔCoP * GF/External Torque (B), and ΔFy * 0.5*w/External Torque (C) for trials 4–8 of blocks in the ‘no cues’ condition are depicted for each group together with Holm-adjusted p-values of post-hoc t-tests of pairwise differences between controls and left- respectively right-hemispheric stroke patients.

Figure 3

Transfer of sensorimotor learning of anticipatory torque compensation to explicit CoM changes. (A) Tcom/external torque, (B) ΔCoP * GF/External Torque, and (C) ΔFy * 0.5*w/External Torque of the first trial of a block after the CoM has changed in the ‘no cues, blocked’ condition (first trial of first block excluded) of all groups.

Figure 4

Sensorimotor torque control in uncertainty. (A) Tcom/external torque, (B) ΔCoP * GF/External Torque, and (C) ΔFy * 0.5*w/External Torque of all groups averaged for trials in which the CoM has changed and trials in which it remained constant for both possible CoMs in the ‘no cues, pseudorandom’ condition.

Figure 5

Learning of anticipatory torque compensation according to both geometric cues and sensorimotor memories. (A) Tcom/external torque, (B) ΔCoP * GF/External Torque, and (C) ΔFy * 0.5*w/External Torque for trials 4–8 of blocks in the ‘geometric cues’ condition of all groups.

Figure 6

Interaction of visuomotor transformations and the transfer of sensorimotor learning of anticipatory torque compensation after CoM change in the blocked condition. (A) Tcom/external torque, (B) ΔCoP * GF/External Torque, and (C) ΔFy * 0.5*w/External Torque of the first trial of a block after the CoM has changed in the ‘geometric cues’ condition (first trial of first block excluded) of all groups.

Figure 7

Interaction of visuomotor transformations and the transfer of sensorimotor learning of anticipatory torque compensation after CoM change in the pseudorandom condition. (A) Tcom/external torque, (B) ΔCoP * GF/External Torque, and (C) ΔFy * 0.5*w/External Torque of all groups averaged for trials in which the CoM has changed and trials in which it remained constant for both possible CoMs in the ‘geometric cues, pseudorandom’ condition.

Figure 8

∆CoP at lift-off. (A) the blocked, no-cues condition (trials 4–8), (B) the blocked, visual-cues condition (trials 4–8), (C) the pseudorandom, no-cues condition and (D) the pseudorandom, visual-cues condition.

Figure 9

GF at lift-off was similar between stroke and control groups in all experimental conditions. (A) the blocked, no-cues condition (trials 4–8), (B) the blocked, visual-cues condition (trials 4–8), (C) the pseudorandom, no-cues condition and (D) the pseudorandom, visual-cues condition.