Separating Predicted and Perceived Sensory Consequences of Motor Learning

Automatic learning mechanisms for flexible human locomotion

Movement flexibility and automaticity are necessary to successfully navigate different environments. When encountering difficult terrains such as a muddy trail, we can change how we step almost immediately so that we can continue walking. This flexibility comes at a cost since we initially must pay deliberate attention to how we are moving. Gradually, after a few minutes on the trail, stepping becomes automatic so that we do not need to think about our movements. Canonical theory indicates that different adaptive motor learning mechanisms confer these essential properties to movement: explicit control confers rapid flexibility, while forward model recalibration confers automaticity. Here we uncover a distinct mechanism of treadmill walking adaptation - an automatic stimulus-response mapping - that confers both properties to movement. The mechanism is flexible as it learns stepping patterns that can be rapidly changed to suit a range of treadmill configurations. It is also automatic as it can operate without deliberate control or explicit awareness by the participants. Our findings reveal a tandem architecture of forward model recalibration and automatic stimulus-response mapping mechanisms for walking, reconciling different findings of motor adaptation and perceptual realignment.

Perceptual error based on Bayesian cue combination drives implicit motor adaptation

The sensorimotor system can recalibrate itself without our conscious awareness, a type of procedural learning whose computational mechanism remains undefined. Recent findings on implicit motor adaptation, such as over-learning from small perturbations and fast saturation for increasing perturbation size, challenge existing theories based on sensory errors. We argue that perceptual error, arising from the optimal combination of movement-related cues, is the primary driver of implicit adaptation. Central to our theory is the increasing sensory uncertainty of visual cues with increasing perturbations, which was validated through perceptual psychophysics (Experiment 1). Our theory predicts the learning dynamics of implicit adaptation across a spectrum of perturbation sizes on a trial-by-trial basis (Experiment 2). It explains proprioception changes and their relation to visual perturbation (Experiment 3). By modulating visual uncertainty in perturbation, we induced unique adaptation responses in line with our model predictions (Experiment 4). Overall, our perceptual error framework outperforms existing models based on sensory errors, suggesting that perceptual error in locating one's effector, supported by Bayesian cue integration, underpins the sensorimotor system's implicit adaptation.

Distinct learning, retention, and generalization patterns in de novo learning versus motor adaptation

People correct for movement errors when acquiring new motor skills (de novo learning) or adapting well-known movements (motor adaptation). While de novo learning establishes new control policies, adaptation modifies existing ones, and previous work have distinguished behavioral and underlying brain mechanisms for each motor learning type. However, it is still unclear whether learning in each type interferes with the other. In study 1, we use a within-subjects design where participants train with both 30° visuomotor rotation and mirror reversal perturbations, to compare adaptation and de novo learning respectively. We find no perturbation order effects, and find no evidence for differences in learning rates and asymptotes for both perturbations. Explicit instructions also provide an advantage during early learning in both perturbations. However, mirror reversal learning shows larger inter-participant variability and slower movement initiation. Furthermore, we only observe reach aftereffects following rotation training. In study 2, we incorporate the mirror reversal in a browser-based task, to investigate under-studied de novo learning mechanisms like retention and generalization. Learning persists across three or more days, substantially transfers to the untrained hand, and to targets on both sides of the mirror axis. Our results extend insights for distinguishing motor skill acquisition from adapting well-known movements.

Understanding implicit sensorimotor adaptation as a process of proprioceptive re-alignment

Multiple learning processes contribute to successful goal-directed actions in the face of changing physiological states, biomechanical constraints, and environmental contexts. Amongst these processes, implicit sensorimotor adaptation is of primary importance, ensuring that movements remain well-calibrated and accurate. A large body of work on reaching movements has emphasized how adaptation centers on an iterative process designed to minimize visual errors. The role of proprioception has been largely neglected, thought to play a passive role in which proprioception is affected by the visual error but does not directly contribute to adaptation. Here, we present an alternative to this visuo-centric framework, outlining a model in which implicit adaptation acts to minimize a proprioceptive error, the distance between the perceived hand position and its intended goal. This proprioceptive re-alignment model (PReMo) is consistent with many phenomena that have previously been interpreted in terms of learning from visual errors, and offers a parsimonious account of numerous unexplained phenomena. Cognizant that the evidence for PReMo rests on correlational studies, we highlight core predictions to be tested in future experiments, as well as note potential challenges for a proprioceptive-based perspective on implicit adaptation.

Individual differences in proprioception predict the extent of implicit sensorimotor adaptation

Recent studies have revealed an upper bound in motor adaptation, beyond which other learning systems may be recruited. The factors determining this upper bound are poorly understood. The multisensory integration hypothesis states that this limit arises from opposing responses to visual and proprioceptive feedback. As individuals adapt to a visual perturbation, they experience an increasing proprioceptive error in the opposite direction, and the upper bound is the point where these two error signals reach an equilibrium. Assuming that visual and proprioceptive feedback are weighted according to their variability, there should be a correlation between proprioceptive variability and the limits of adaptation. Alternatively, the proprioceptive realignment hypothesis states that the upper bound arises when the (visually biased) sensed hand position realigns with the expected sensed position (target). When a visuo-proprioceptive discrepancy is introduced, the sensed hand position is biased toward the visual cursor, and the adaptive system counteracts this discrepancy by driving the hand away from the target. This hypothesis predicts a correlation between the size of the proprioceptive shift and the upper bound of adaptation. We tested these two hypotheses by considering natural variation in proprioception and motor adaptation across individuals. We observed a modest, yet reliable correlation between the upper bound of adaptation with both proprioceptive measures (variability and shift). Although the results do not clearly favor one hypothesis over the other, they underscore the critical role of proprioception in sensorimotor adaptation.NEW & NOTEWORTHY Although the sensorimotor system uses sensory feedback to remain calibrated, this learning process is constrained, limited by the maximum degree of plasticity. The factors determining this limit remain elusive. Guided by two hypotheses, we show that individual differences in the upper bound of adaptation in response to a visual perturbation can be predicted by the bias and variability in proprioception. These results underscore the critical, but often neglected role of proprioception in human motor learning.

Implicit motor learning within three trials

In motor learning, the slow development of implicit learning is traditionally taken for granted. While much is known about training performance during adaptation to a perturbation in reaches, saccades and locomotion, little is known about the time course of the underlying implicit processes during normal motor adaptation. Implicit learning is characterized by both changes in internal models and state estimates of limb position. Here, we measure both as reach aftereffects and shifts in hand localization in our participants, after every training trial. The observed implicit changes were near asymptote after only one to three perturbed training trials and were not predicted by a two-rate model's slow process that is supposed to capture implicit learning. Hence, we show that implicit learning is much faster than conventionally believed, which has implications for rehabilitation and skills training.

Adaptation of reach action to a novel force-field is not predicted by acuity of dynamic proprioception in either older or younger adults

Healthy ageing involves degeneration of the neuromuscular system which impacts movement control and proprioception. Yet the relationship between these sensory and motor deficits in upper limb reaching has not been examined in detail. Recently, we reported that age-related proprioceptive deficits were unrelated to accuracy in rapid arm movements, but whether this applied in motor tasks more heavily dependent on proprioceptive feedback was not clear. To address this, we have tested groups of younger and older adults on a force-field adaptation task under either full or limited visual feedback conditions and examined how performance was related to dynamic proprioceptive acuity. Adaptive performance was similar between the age groups, regardless of visual feedback condition, although older adults showed increased after-effects. Physically inactive individuals made larger systematic (but not variable) proprioceptive errors, irrespective of age. However, dynamic proprioceptive acuity was unrelated to adaptation and there was no consistent evidence of proprioceptive recalibration with adaptation to the force-field for any group. Finally, in spite of clear age-dependent loss of spatial working memory capacity, we found no relationship between memory capacity and adaptive performance or proprioceptive acuity. Thus, non-clinical levels of deficit in dynamic proprioception, due to age or physical inactivity, do not affect force-field adaptation, even under conditions of limited visual feedback that might require greater proprioceptive control.

External error attribution dampens efferent-based predictions but not proprioceptive changes in hand localization

In learning and adapting movements in changing conditions, people attribute the errors they experience to a combined weighting of internal or external sources. As such, error attribution that places more weight on external sources should lead to decreased updates in our internal models for movement of the limb or estimating the position of the effector, i.e. there should be reduced implicit learning. However, measures of implicit learning are the same whether or not we induce explicit adaptation with instructions about the nature of the perturbation. Here we evoke clearly external errors by either demonstrating the rotation on every trial, or showing the hand itself throughout training. Implicit reach aftereffects persist, but are reduced in both groups. Only for the group viewing the hand, changes in hand position estimates suggest that predicted sensory consequences are not updated, but only rely on recalibrated proprioception. Our results show that estimating the position of the hand incorporates source attribution during motor learning, but recalibrated proprioception is an implicit process unaffected by external error attribution.

The effect of age on visuomotor learning processes

Knowing where our limbs are in space is essential for moving and for adapting movements to various changes in our environments and bodies. The ability to adapt movements declines with age, and age-related cognitive decline can explain a decreased ability to adopt and deploy explicit, cognitive strategies in motor learning. Age-related sensory decline could also lead to a reduced fidelity of sensory position signals and error signals, each of which can affect implicit motor adaptation. Here we investigate two estimates of limb position; one based on proprioception, the other on predicted sensory consequences of movements. Each is considered a measure of an implicit adaptation process and may be affected by both age and cognitive strategies. Both older (n = 38) and younger (n = 42) adults adapted to a 30° visuomotor rotation in a centre-out reaching task. We make an explicit, cognitive strategy available to half of participants in each age group with a detailed instruction. After training, we first quantify the explicit learning elicited by instruction. Instructed older adults initially use the provided strategy slightly less than younger adults but show a similar ability to evoke it after training. This indicates that cognitive explanations for age-related decline in motor learning are limited. In contrast, training induced much larger shifts of state estimates of hand location in older adults compared to younger adults. This is not modulated by strategy instructions, and appears driven by recalibrated proprioception, which is almost twice as large in older adults, while predictions might not be updated in older adults. This means that in healthy aging, some implicit processes may be compensating for other changes to maintain motor capabilities, while others also show age-related decline (data: https://osf.io/qzhmy).

Split-Belt walking induces changes in active, but not passive, perception of step length

Successful motor control requires accurate estimation of our body in space for planning, executing, and evaluating the outcome of our actions. It has been shown that the estimation of limb position is susceptible to motor adaptation. However, a similar effect has not been found in locomotion, possibly due to how it was tested. We hypothesized that split-belt walking with the legs moving at different speeds changes the estimation of the legs' position when taking a step. Thus, we assessed young subjects' perception of step length (i.e., inter-feet distance at foot landing) when they moved their legs (active perception) or when the legs were moved by the experimenter (passive perception). We found that the active perception of step length was substantially altered following split-belt walking, whereas passive perception exhibited minor changes. This suggests that split-belt walking induced the adaptation of efferent signals, without altering sensory signals. We also found that active perceptual shifts were sensitive to how they were tested: they were most salient in the trailing leg and at short step lengths. Our results suggest that split-belt walking could modulate the deficient perception of step length post-stroke, which may contribute to gait asymmetries impairing patients' mobility.

Adaptation to Laterally Asymmetrical Visuomotor Delay Has an Effect on Action But Not on Perception

When interacting with the environment, the sensorimotor system faces temporal and spatial discrepancies between sensory inputs, such as delay in sensory information transmission, and asymmetrical visual inputs across space. These discrepancies can affect motor control and the representation of space. We recently showed that adaptation to a laterally asymmetric delay in the visual feedback induces neglect-like effects in blind drawing movements, expressed by asymmetrical elongation of circles that are drawn in different workspaces and directions; this establishes a possible connection between delayed feedback and asymmetrical spatial processing in the control of action. In the current study, we investigate whether such adaptation also influences visual perception. In addition, we examined transfer to another motor task – a line bisection task that is commonly used to detect spatial disorders, and extend these results to examine the mapping of these neglect-like effects. We performed two sets of experiments in which participants executed lateral reaching movements, and were exposed to visual feedback delay only in the left workspace. We examined transfer of adaptation to a perceptual line bisection task – answers about the perceived midline of lines that were presented in different directions and workspaces, and to a blind motor line bisection task – reaching movements toward the centers of similar lines. We found that the adaptation to the asymmetrical delay transferred to the control of lateral movements, but did not affect the perceived location of the midlines. Our results clarify the effect of asymmetrical delayed visual feedback on perception and action, and provide potential insights on the link between visuomotor delay and neurological disorders such as the hemispatial neglect syndrome.

Motor learning without moving: Proprioceptive and predictive hand localization after passive visuoproprioceptive discrepancy training

An accurate estimate of limb position is necessary for movement planning, before and after motor learning. Where we localize our unseen hand after a reach depends on felt hand position, or proprioception, but in studies and theories on motor adaptation this is quite often neglected in favour of predicted sensory consequences based on efference copies of motor commands. Both sources of information should contribute, so here we set out to further investigate how much of hand localization depends on proprioception and how much on predicted sensory consequences. We use a training paradigm combining robot controlled hand movements with rotated visual feedback that eliminates the possibility to update predicted sensory consequences (‘exposure training’), but still recalibrates proprioception, as well as a classic training paradigm with self-generated movements in another set of participants. After each kind of training we measure participants’ hand location estimates based on both efference-based predictions and afferent proprioceptive signals with self-generated hand movements (‘active localization’) as well as based on proprioception only with robot-generated movements (‘passive localization’). In the exposure training group, we find indistinguishable shifts in passive and active hand localization, but after classic training, active localization shifts more than passive, indicating a contribution from updated predicted sensory consequences. Both changes in open-loop reaches and hand localization are only slightly smaller after exposure training as compared to after classic training, confirming that proprioception plays a large role in estimating limb position and in planning movements, even after adaptation. (data: https://doi.org/10.17605/osf.io/zfdth, preprint: https://doi.org/10.1101/384941)

The effects of awareness of the perturbation during motor adaptation on hand localization

Awareness of task demands is often used during rehabilitation and sports training by providing instructions which appears to accelerate learning and improve performance through explicit motor learning. However, the effects of awareness of perturbations on the changes in estimates of hand position resulting from motor learning are not well understood. In this study, people adapted their reaches to a visuomotor rotation while either receiving instructions on the nature of the perturbation, experiencing a large rotation, or both to generate awareness of the perturbation and increase the contribution of explicit learning. We found that instructions and/or larger rotations allowed people to activate or deactivate part of the learned strategy at will and elicited explicit changes in open-loop reaches, while a small rotation without instructions did not. However, these differences in awareness, and even manipulations of awareness and perturbation size, did not appear to affect learning-induced changes in hand-localization estimates. This was true when estimates of the adapted hand location reflected changes in proprioception, produced when the hand was displaced by a robot, and also when hand location estimates were based on efferent-based predictions of self-generated hand movements. In other words, visuomotor adaptation led to significant shifts in predicted and perceived hand location that were not modulated by either instruction or perturbation size. Our results indicate that not all outcomes of motor learning benefit from an explicit awareness of the task. Particularly, proprioceptive recalibration and the updating of predicted sensory consequences appear to be largely implicit. (data: https://doi.org/10.17605/osf.io/mx5u2, preprint: https://doi.org/10.31234/osf.io/y53c2)

The capacity to learn new motor and perceptual calibrations develops concurrently in childhood

Learning new movements through an error-based process called motor adaptation is thought to involve multiple mechanisms which are still largely not understood. Previous studies have shown that young children adapt movement more slowly than adults, perhaps supporting the involvement of distinct neural circuits that come online at different stages of development. Recent studies in adults have shown that in addition to recalibrating a movement, motor adaptation also leads to changes in the perception of that movement. However, we do not yet understand the relationship between the processes that underlie motor and perceptual recalibration. Here we studied motor and perceptual recalibration with split-belt walking adaptation in adults and children aged 6-8 years. Consistent with previous work, we found that this group of children adapted their walking patterns more slowly than adults, though individual children ranged from slow to adult-like in their adaptation rates. Perceptual recalibration was also reduced in the same group of children compared to adults, with individual children ranging from having no recalibration to having adult-like recalibration. In sum, faster motor adaptation and the ability to recalibrate movement perception both come online within a similar age-range, raising the possibility that the same sensorimotor mechanisms underlie these processes.

Visual and proprioceptive recalibrations after exposure to a visuomotor rotation

Adaptation to a visuomotor rotation in a cursor-control task is accompanied by proprioceptive recalibration, whereas the existence of visual recalibration is uncertain and has even been doubted. In the present study, we tested both visual and proprioceptive recalibration; proprioceptive recalibration was not only assessed by means of psychophysical judgments of the perceived position of the hand, but also by an indirect procedure based on movement characteristics. Participants adapted to a gradually introduced visuomotor rotation of 30° by making center-out movements to remembered targets. In subsequent test trials, they made center-out movements without visual feedback or observed center-out motions of a cursor without moving the hand. In each test trial, they judged the endpoint of hand or cursor by matching the position of the hand or of a visual marker, respectively, moving along a semicircular path. This path ran through all possible endpoints of the center-out movements. We observed proprioceptive recalibration of 7.3° (3.1° with the indirect procedure) and a smaller, but significant, visual recalibration of 1.3°. Total recalibration of 8.6° was about half as strong as motor adaptation, the adaptive shift of the movement direction. The evidence of both proprioceptive and visual recalibration was obtained with a judgment procedure that suggests that recalibration is restricted to the type of movement performed during exposure to a visuomotor rotation. Consequently, identical physical positions of the hand can be perceived differently depending on how they have been reached, and similarly identical positions of a cursor on a monitor can be perceived differently.

Movements following force-field adaptation are aligned with altered sense of limb position

Previous work has shown that motor learning is associated with changes to both movements and to the somatosensory perception of limb position. In an earlier study that motivates the current work, it appeared that following washout trials, movements did not return to baseline but rather were aligned with associated changes to sensed limb position. Here, we provide a systematic test of this relationship, examining the idea that adaptation-related changes to sensed limb position and to the path of the limb are linked, not only after washout trials but at all stages of the adaptation process. We used a force-field adaptation paradigm followed by washout trials in which subjects performed movements without visual feedback of the limb. Tests of sensed limb position were conducted at each phase of adaptation, specifically before and after baseline movements in a null field, after force-field adaptation, and following washout trials in a null field. As in previous work, sensed limb position changed in association with force-field adaptation. At each stage of adaptation, we observed a correlation between the sensed limb position and associated path of the limb. At a group level, there were differences between the clockwise and counter-clockwise conditions. However, whenever there were changes in sensed limb position, movements following washout did not return to baseline. This suggests that adaptation in sensory and motor systems is not independent processes but rather sensorimotor adaptation is linked to sensory change. Sensory change and limb movement remain in alignment throughout adaptation such that the path of the limb is aligned with the altered sense of limb position.

The fast contribution of visual-proprioceptive discrepancy to reach aftereffects and proprioceptive recalibration

Adapting reaches to altered visual feedback not only leads to motor changes, but also to shifts in perceived hand location; "proprioceptive recalibration". These changes are robust to many task variations and can occur quite rapidly. For instance, our previous study found both motor and sensory shifts arise in as few as 6 rotated-cursor training trials. The aim of this study is to investigate one of the training signals that contribute to these rapid sensory and motor changes. We do this by removing the visuomotor error signals associated with classic visuomotor rotation training; and provide only experience with a visual-proprioceptive discrepancy for training. While a force channel constrains reach direction 30o away from the target, the cursor representing the hand unerringly moves straight to the target. The resulting visual-proprioceptive discrepancy drives significant and rapid changes in no-cursor reaches and felt hand position, again within only 6 training trials. The extent of the sensory change is unexpectedly larger following the visual-proprioceptive discrepancy training. Not surprisingly the size of the reach aftereffects is substantially smaller than following classic visuomotor rotation training. However, the time course by which both changes emerge is similar in the two training types. These results suggest that even the mere exposure to a discrepancy between felt and seen hand location is a sufficient training signal to drive robust motor and sensory plasticity.

Recalibration of hand position sense during unconscious active and passive movement

Precise knowledge of one's limbs' position in space is fundamental for goal-directed action. The brain's representation of the body in space is thought to be generated through a process of multisensory integration of visual, tactile and proprioceptive signals. In this study, we devised a setup that allowed us to displace participants' right hand without their subjective awareness. We accomplished this task by instructing the participants to view a live video feed of their hand from the first-person perspective. In the active condition, we used a sensorimotor illusion that caused the participants to actively but unknowingly displace their unseen right hand to a location 8 cm lateral to the image of their hand. In the passive condition, we mechanically displaced the participants' hand-at a slow, unnoticeable velocity-to the same location. We found that during active displacement, the participants indicated that the location of their hand was closer to the digital image of the hand rather than the veridical location of the hand, as compared with the passive condition, in which the participants indicated that the locations of their hand were closer to the actual location. These results indicated that, compared with passive displacement, active movements cause greater recalibration of the hand's spatial position and that the boosted spatial recalibration of hand position sense in the active task is driven by error-based sensorimotor corrections. These results have bearing on the perceptual mechanisms of recalibration of perceived limb location.