The nervous system uses nonspecific motor learning in response to random perturbations of varying nature

Abstract

We constantly make small errors during movement and use them to adapt our future movements. Movement experiments often probe this error-driven learning by perturbing movements and analyzing the after-effects. Past studies have applied perturbations of varying nature such as visual disturbances, position- or velocity-dependent forces and modified inertia properties of the limb. However, little is known about how the specific nature of a perturbation influences subsequent movements. For a single perturbation trial, the nature of a perturbation may be highly uncertain to the nervous system, given that it receives only noisy information. One hypothesis is that the nervous system can use this rough estimate to partially correct for the perturbation on the next trial. Alternatively, the nervous system could ignore uncertain information about the nature of the perturbation and resort to a nonspecific adaptation. To study how the brain estimates and responds to incomplete sensory information, we test these two hypotheses using a trial-by-trial adaptation experiment. On each trial, the nature of the perturbation was chosen from six distinct types, including a visuomotor rotation and different force fields. We observed that corrective forces aiming to oppose the perturbation in the following trial were independent of the nature of the perturbation. Our results suggest that the nervous system uses a nonspecific strategy when it has high uncertainty about the nature of perturbations during trial-by-trial learning.

Fig. 1.

A: sketch of the manipulandum and the experimental setup. B–G: 6 different perturbations are shown in both left and right directions. The target is 10 cm away in depth direction (Y direction) and all the perturbations are applied in the lateral direction (left and right, X direction). Note for the visuomotor rotation (shown in B), the X axis is for displacement and it is different from other panels where the X axis is for applied perturbation force.

Fig. 2.

Average of the hand trajectories from perturbation trials and subsequent catch trials are shown for different perturbation conditions separately. The perturbation trials are shown as dash lines and the catch trials as solid lines. Trials with left perturbations are shown in black and right in gray. The gray shades stand for the SE across subjects. Note in the 1st panel that shows the visual perturbation condition, the perturbation trajectories are the displayed visual feedback instead of the actual hand trajectories. Data are only from experiment 1; experiment 2 produced similar data for perturbation trials, but its catch trials have constrained the lateral deviation within 1 mm (data not shown).

Fig. 3.

A and C: the visual trajectories (in the lateral direction) in perturbation trials (A) and catch trials (C) are plotted for different perturbation conditions separately. The black lines are the subject average, and the gray shadowed areas (often tiny) indicate the SE across subjects. B and D: correlation coefficients from cross-correlation analysis of visual trajectories resulted from perturbed trials (B) and catch trials (D) are plotted as a matrix of gray shades. The correlation varies widely across pairwise comparisons for perturbed trials, but on average, they are low, indicating distinct perturbation effects. On the other hand, the correlation coefficients for catch trials are much higher, indicating similar 1-trial learning across conditions. Data are from experiment 1; experiment 2 does not permit analyses of trajectories because it constrained the lateral movements in catch trials by applying strong force channels.

Fig. 4.

A–F: the acceleration trajectories (in the lateral direction) in perturbation trials are plotted for different perturbation conditions separately. The black lines are the subject average, and the gray shades (not very visible) stand for the SE across subjects. G: correlation coefficients from cross-correlation analysis of acceleration trajectories resulted from different perturbations are plotted as a matrix of gray shades. The correlation varies widely across these pairwise comparisons, but on average, they are low, indicating distinct perturbation effects. Data are from experiment 1; experiment 2 used identical perturbations and thus produced similar data.

Fig. 5.

A: the acceleration trajectories (in the lateral direction) in catch trials from experiment 1 are plotted for different perturbation conditions separately. The black lines are the subject average, and the gray shades stand for the SE across subjects. B: the same acceleration trajectories as shown in A are scaled according to their own range and stacked together. Each line is from a single condition. C: the force trajectories (in the lateral direction) in catch trials from experiment 2 are plotted for different perturbation conditions separately. The black lines are the subject average, and the gray shades stand for the SE across subjects. The red lines denote the forces that are calculated by multiplying the raw force with decreasing gains. D: the force trajectories as shown in C are scaled according to their own range and stacked together. Each line is from a single condition.

Fig. 6.

A and B: correlation coefficients of pairwise correlation between perturbation trials and catch trials are displayed as gray shades for experiment 1 and experiment 2, respectively. The correlation analysis is performed on acceleration trajectories for experiment 1 and force trajectories for experiment 2. If the catch trial is specific to its preceding perturbation trial, the largest correlation should be found for their pairwise correlation, which lies on the diagonal that is displayed as a gray dash line. The largest correlation and the 2nd largest correlation on each row are shown, most of which are not on the diagonal line. This suggests there is no close correspondence between the perturbation and its corresponding catch trial. C: the influence (α) of the perturbation trial onto corrective responses in the catch trial is plotted as a function of conditions. None of the conditions has influence significantly different from 0 except the ramp condition for experiment 1 and the triangle condition for experiment 2. This indicates that, for most perturbations, the 1st-trial adaptation is not specific.

Fig. 7.

A and B: the 1st principal components of acceleration trajectories (A, experiment 1) and force trajectories (B, experiment 2) in catch trials are plotted against error measures in its preceding perturbation trials. These error measures, shown by different subplots, include the movement endpoint error, the maximal lateral deviation of the hand from the straight path, the integral of the deviation over the movement, and the deviation at the peak velocity of the reach, respectively. There is no visible dependence of between the trajectory and these error metrics.