Better on the Second Encounter: Separating Components of Motor Relearning

You move your hand toward a target, but the cursor on the screen travels in a different direction. With experience, you can learn to compensate, perhaps sending your hand somewhere else so the cursor reaches the target. When the same arrangement returns, your early performance may improve. Does that improvement mean every process contributing to the movement has become stronger? Research published in 2021 by Guy Avraham, J. Ryan Morehead, Hyosub E. Kim, and Richard B. Ivry shows why the overall result needs to be separated into components. [1]

The study examined what happens when people encounter the same visuomotor perturbation twice. Faster relearning is often called savings. A benefit of that kind can arise from recalling a successful strategy, however, rather than from every part of the sensorimotor system learning more rapidly. Deliberately changing where to aim and adapting without an explicit decision can both contribute to a similar hand movement. The researchers used different instructions and feedback conditions to investigate those contributions instead of treating total performance as a sufficient explanation. [1]

Making the cursor succeed or sending the hand directly

In the first experiment, twenty-four participants made center-out movements using a tablet apparatus while their hands were hidden from direct view. A screen supplied visual feedback. On one type of learning trial, cursor direction was rotated forty-five degrees relative to hand direction. Participants were asked to make the cursor reach the target. Deliberately aiming the hand away from the target could help compensate for that rotation. [1]

On interleaved probe trials, the cursor was not visible. Participants were instructed to stop using any compensatory strategy and move their hand directly to the target. These trials were intended to estimate the adaptation remaining apart from deliberate re-aiming. The estimate did not come from asking people to report their intended aim verbally before every movement. It came from changing the task instructions and observing what behavior persisted. The implicit component was therefore operationally estimated, not read directly from the mind. [1]

After learning, a separate no-feedback block again instructed participants to move directly to the target. A continuing deviation of the hand provided an aftereffect measure. An extended washout phase with accurate visual feedback followed before the same rotation returned. Washout was not simply sitting still and waiting: participants continued to make movements. Reducing the observable effects of earlier adaptation mattered for interpreting what happened when learning began again. [1]

The overall benefit concealed a mixed pattern

On trials requiring participants to bring the rotated cursor to the target, performance improved more quickly early in the second learning block. By late learning, the difference between blocks was not significant. An early advantage and a higher final level are different outcomes. Starting to solve a task more effectively does not necessarily mean eventually reaching a higher endpoint. [1]

The probe trials required a more qualified interpretation. Their overall difference between learning blocks was not significant. A small early increase appeared in one analysis, but the cluster analysis of the time course did not identify a significant between-block difference. The authors also considered contamination from incomplete strategy switching. If someone failed to abandon re-aiming fully on a probe trial, the remaining strategy could make an estimate of implicit adaptation look larger. That possibility was especially relevant when deliberate compensation changed strongly between blocks. [1]

The later no-feedback aftereffect, in contrast, was reliably smaller after the second learning block. Describing the whole motor system as simply speeding up would conceal that result. Overall early performance improved while a separate measure of adaptation decreased. It would be equally misleading to say that every implicit measure decreased at every point in time. The trial type and the stage of learning remained important to the interpretation. [1]

An angular error the participant could not control

The second experiment used another method to reduce the strategy-switching problem. Cursor direction was clamped to a path fifteen degrees away from the target, regardless of the hand's movement direction. The cursor was not completely stationary or entirely independent of the hand: its radial distance from the start followed the hand's progress. Its angular direction, however, could not be corrected by aiming the hand somewhere else. Participants saw demonstrations of this arrangement and were told to ignore the cursor and reach directly to the target. [1]

Despite those instructions, their hands gradually shifted in the direction opposite to the displayed angular error. In the debriefing, participants reported planning their movements toward the target and did not report awareness of the systematic change in hand direction. Those observations supported interpreting the drift as implicit adaptation. They did not provide an exhaustive measurement of every aspect of conscious experience. Behavior, instructions, and retrospective reports were considered together. [1]

Two groups of sixteen participants took part in this experiment. The test group encountered two learning blocks separated by washout. That washout first reversed the direction of the clamped feedback and then supplied accurate feedback. A control group instead received one extended block of the same clamped feedback. The comparison helped examine whether a reduction later in the session could simply reflect fatigue or habituation from prolonged exposure. [1]

For the test group, late adaptation and the subsequent aftereffect were smaller during relearning. The numerical reduction in early learning was not significant. The extended-exposure control group did not show a comparable late decline. This pattern weakened a simple account based solely on spending longer in the task. Nevertheless, not every comparison between the groups was significant. The direct group comparison based on the first aftereffect cycle was not significant, while the time-course cluster analysis identified more selective differences. Different analyses should not be compressed into an assertion that all measures agreed without qualification. [1]

Repetition did not have one uniform effect

The researchers also tested a related design online with forty-eight participants using their own computers. Late-learning and aftereffect comparisons supported attenuation, but the overall block effect by itself was not significant. Separating stages was important in that replication as well. Participants used their own mice or trackpads rather than an identical laboratory apparatus, so the online study provided another setting without reproducing every equipment detail. [1]

The mechanisms behind the attenuation were not settled. One possibility involved a changed response to familiar errors. Another involved interference from memories formed during washout. These are explanatory candidates, not demonstrated neural mechanisms. The experiments did not directly measure exhausted synapses or identify a brain process as the unique cause of the behavioral pattern. They also did not show that all forms of implicit learning must weaken when repeated. Their scope concerned the visuomotor transformations, instructions, and feedback arrangements used here. [1]

That boundary matters for practical interpretation. The findings are not a prescription for skill practice or a test of a clinical rehabilitation program. Nor do they imply that improvement is somehow unreal because several processes contribute to it. Rather, improvement needs a sufficiently specific description. A recalled strategy can improve task performance even when an independently assessed adaptation is smaller. Opposing changes can coexist within behavior that appears, at first glance, simply better on its second attempt.

The central lesson is therefore methodological as much as psychological. An overall performance benefit does not explain which component changed. By altering instructions, withholding feedback, and clamping its angular direction, the study asked different questions of apparently similar movements. Separating those questions made it possible to see that relearning was not a single process moving uniformly in one direction.

Sources

[1] Avraham, G., Morehead, J. R., Kim, H. E., and Ivry, R. B. (2021). Reexposure to a sensorimotor perturbation produces opposite effects on explicit and implicit learning processes. PLOS Biology, 19(3), e3001147. https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3001147 . DOI: 10.1371/journal.pbio.3001147. Original research under a Creative Commons Attribution license; this article is an independently written explanation and assessment.