An implicit memory of errors limits human sensorimotor adaptation

Abstract

After extended practice, motor adaptation reaches a limit in which learning appears to stop, despite the fact that residual errors persist. What prevents the brain from eliminating the residual errors? Here we found that the adaptation limit was causally dependent on the second order statistics of the perturbation; when variance was high, learning was impaired and large residual errors persisted. However, when learning relied solely on explicit strategy, both the adaptation limit and its dependence on perturbation variability disappeared. In contrast, when learning depended entirely, or in part on implicit learning, residual errors developed. Residual errors in implicit performance were caused by variance-dependent modifications to error sensitivity, not forgetting. These observations are consisted with a model of learning in which the implicit system becomes more sensitive to error when errors are consistent, but forgets this memory of errors over time. Thus, residual errors in motor adaptation are a signature of the implicit learning system, caused by an error sensitivity that depends on the history of past errors. Introduction During motor adaptation, perturbations alter the sensory consequences of motor commands, yielding sensory prediction errors. In humans and other animals, the brain learns from these errors and adjusts its motor commands on subsequent attempts. Over many trials, the adjustments accumulate, but surprisingly, adaptation often remains incomplete; even after extended periods of practice, residual errors persist in many behaviors including reaching 1-4 , saccades 5,6 , and walking 7 . Why does learning appear to stop despite the fact that errors remain?Current models suggest that adaptation is supported by distinct learning systems: one implicit 8 , and the other explicit [9][10][11] . It is thought that the implicit system contributes little to modulation of asymptotic performance; when challenged with fixed errors, the implicit system appears to saturate at identical levels 12,13 . In contrast, explicit strategy provides greater flexibility; its asymptotic behavior is altered as people age [14][15][16] , under different types of feedback 17 , and with the time allotted for the preparation of a movement 18 . Therefore, current evidence suggests that the explicit system alone modifies the asymptotic state of learning.Here we tested this view using stochastic perturbations that affected reaching movements. We found that when perturbation variability was high, residual errors increased [19][20][21] . Furthermore, when perturbation variability was increased mid-experiment, the asymptotic performance decreased, causing participants to lose what they had already learned. Thus, the asymptote of adaptation was not a hard limit, but a dynamic variable that depended on the second order statistics of the perturbation. Which adaptive system was responsible for limiting the adaptation process?To answer this question we isolated implicit and explicit components of adaptation using several methodologies inc...

Residual errors in visuomotor adaptation persist despite extended motor preparation periods

“…Notwithstanding these studies, we found that on average participants' learning plateaued before eliminating error, with asymptotic learning levels statistically different from the degree of imposed rotation for all rotational groups (30°, 45°, 60°), regardless of the amount of preparation time enforced. These findings are consistent with previous reports which have suggested that sustained errors during visuomotor adaptation are a product of the implicit learning system; either suppressing alternative learning mechanisms which may be able to overcome persisting errors (Shmuelof et al, 2012;Vaswani et al, 2015) and/or modulating the system's sensitivity to errors dependent on their size, variability and history (Albert et al, 2021).…”

“…We found that motor adaptation was impaired in the 45° and 60° rotational groups, when planning time was restricted to 0.35 seconds. This result is in line with previous studies who report reduced error compensation when preparation times are limited (Fernandez-Ruiz et al, 2011;Leow, Gunn, et al, 2017;Leow, Marinovic, Riek, & Carroll, 2017;Albert et al, 2021) and may reflect the suppression of explicit re-aiming/cognitive strategies. Additionally, impaired adaptation associated with restricted preparation times was coupled with an increased retention during no-feedback trials, which is consistent with the idea that learning involved more implicit, procedural processes (Fernandez-Ruiz et al, 2011;Haith et al, 2015).…”

Sensitivity to Error During Visuomotor Adaptation is Similarly Modulated by Abrupt, Gradual and Random Perturbation Schedules

“…Recent work suggests that error sensitivity in sensorimotor adaptation is likely not constant, but rather can vary depending on prior experience (Albert et al, 2021;Herzfeld et al, 2014;Marko et al, 2012;Wei & Körding, 2009). We modelled movement angle across each session with the state-space equations proposed by Smith et al (2006), and focused on changes in the retention and error sensitivity parameters.…”

“…The copyright holder for this preprint this version posted June 15, 2021. ; https://doi.org/10.1101/2021.06.14.448375 doi: bioRxiv preprint perturbation schedules used during initial training was that the slow process accounted for a significant portion of the error reduction. Albert et al (2021) recently investigated the persistence of residual errors during motor adaptation in the context of implicit and explicit learning systems. Of importance to the present study, they propose that it is the implicit learning system which maintains a history for prior errors.…”

Pupil diameter tracked during motor adaptation in humans

“…As we did not employ the experimental design that explicitly quantified these components, this is still an open question. Notably, however, a very recent study (Albert et al, 2021) demonstrated that the learning rate for the "implicit" motor learning process is sensitive to environmental statistics (e.g., perturbation consistency), suggesting that the "implicit" motor learning process is likely modulated by environmental uncertainty/surprise.…”