PTJ: Physical Therapy & Rehabilitation Journal | Physical Therapy, 2022;102:1–9

[Link]
Advance access publication date October 25, 2021
Perspective

Updates in Motor Learning: Implications for Physical

Therapist Practice and Education
Kristan A. Leech, PT, DPT, PhD1 ,* , Ryan T. Roemmich, PhD2 ,3 , James Gordon, PT, EdD, FAPTA1 ,
Darcy S. Reisman, PT, PhD, FAPTA4 , Kendra M. Cherry-Allen, PT, DPT, PhD2
1 Divisionof Biokinesiology and Physical Therapy, University of Southern California, Los Angeles, California, USA
2 Department of Physical Medicine and Rehabilitation, Johns Hopkins University, Baltimore, Maryland, USA
3 Center for Movement Studies, Kennedy Krieger Institute, Baltimore, Maryland, USA

Downloaded from [Link] by guest on 01 July 2023

4 Physical Therapy Department, University of Delaware, Newark, Delaware, USA

*Address all correspondence to Dr Leech at: kleech@[Link]

Over the past 3 decades, the volume of human motor learning research has grown enormously. As such, the understanding of
motor learning (ie, sustained change in motor behavior) has evolved. It has been learned that there are multiple mechanisms
through which motor learning occurs, each with distinctive features. These mechanisms include use-dependent, instructive,
reinforcement, and sensorimotor adaptation-based motor learning. It is now understood that these different motor learning
mechanisms contribute in parallel or in isolation to drive desired changes in movement, and each mechanism is thought
to be governed by distinct neural substrates. This expanded understanding of motor learning mechanisms has important
implications for physical therapy. It has the potential to facilitate the development of new, more precise treatment approaches
that physical therapists can leverage to improve human movement. This Perspective describes scientific advancements
related to human motor learning mechanisms and discusses the practical implications of this work for physical therapist
practice and education.
Keywords: Instructive, Motor Learning, Rehabilitation, Reinforcement, Sensorimotor Adaptation, Use-Dependent

Received: April 16, 2021. Revised: August 12, 2021. Accepted: October 1, 2021
© The Author(s) 2022. Published by Oxford University Press on behalf of the American Physical Therapy Association.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License ([Link]
by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
2 Updates in Motor Learning

Introduction mechanisms through which it occurs will ultimately guide

Promoting motor learning (ie, sustained change in a motor and sharpen physical therapist practice and promote future
behavior) is a fundamental objective of many interventions clinical and research innovation.
within neurological physical therapy. Though this may be true
now, it has not always been. Prevailing treatment philosophies Four Mechanisms of Motor Learning
(or “best practices”) in neurological physical therapy have
evolved over time. Over the last 60 years, the predominant Through significant research efforts, we have come to under-
approach to treating patients with neurological damage or stand that motor learning does not occur through a single pro-
disease has shifted from mainly a neurophysiology-based cess. Rather, there are multiple mechanisms of motor learning
approach to a more pragmatic and eclectic approach that that can occur in parallel or in isolation to ultimately lead to
emphasizes motor learning principles along with exercise sustained changes in motor behaviors.19,21 In this section, we
science and biomechanics.1,2 provide an overview of 4 widely studied motor learning mech-
This evolution was prompted by discoveries in the fields of anisms: use-dependent, instructive, reinforcement, and senso-
behavioral motor learning and neuroscience and an increased rimotor adaptation-based motor learning. We will describe

Downloaded from [Link] by guest on 01 July 2023

awareness of the relevance of this work to the field of physical the key features of these mechanisms including the primary
therapy.3,4 A detailed historical overview of the integration behavioral drivers, neural substrates, cognitive involvement,
of motor learning research into neurological physical ther- and timescales. Here we define the primary behavioral driver
apist practice is provided by Winstein et al.5 Recently, it as the task demands that will increase the relative contribution
has become evident that neuroscience and behavioral motor of each learning mechanism to the overall change in behavior.
learning research are also relevant to other areas of physical This is perhaps the most clinically relevant feature of these
therapist practice (eg, orthopedics6–9 and pelvic health10,11 ). motor learning mechanisms, because it is the element that
Accordingly, motor learning has become a fundamental physical therapists can manipulate through the structure of
component of entry-level physical therapy education. This their interventions. The distinct features of each mechanism
includes content related to the inherent plasticity of the ner- are summarized in Figure 1. In alignment with the motor
vous system that can be driven through repeated practice learning literature, we use the term “mechanism” generally to
of a new movement. Physical therapy curricula also com- refer to the processes through which motor learning occurs.
monly include behavioral motor learning research findings This differs from its use in rehabilitation literature, in which
that provide guidelines on how to structure interventions to it often describes a neurophysiological process underlying a
promote motor learning (eg, variable versus constant prac- clinical presentation or observed behavior. Of note, to help
tice,12,13 feedback schedules,14,15 principles of behavioral clarify how the concepts described here are related to existing
economics16,17 ). Although this approach to teaching motor literature, commonly used synonyms for various motor learn-
learning concepts is useful and an improvement over what ing terms are provided in Figure 1.
was taught in previous years, it does not reflect more recent The current state of science related to these motor learn-
advances in motor learning research. ing mechanisms is still largely based on laboratory-based
As physical therapists have begun to integrate principles of research. There have been few intervention studies (eg, Reis-
motor learning and neuroplasticity into practice and educa- man et al22 and Wolf et al23 ) to inform the design of robust
tion, motor learning research has also continued to advance. treatment protocols or the development of guidelines for
We now have a more sophisticated and comprehensive under- clinical implementation. Therefore, we will primarily focus on
standing of how motor learning occurs. Specifically, we have the potential practical implications of these advances in motor
learned that (1) there are multiple mechanisms of motor learning research versus direct clinical application.
learning, (2) these mechanisms have distinct features, and (3)
they primarily engage different areas of the nervous system. Use-Dependent Motor Learning
For an overview of these scientific developments, please see Use-dependent motor learning (Fig. 1; blue) is defined as a
reviews by Krakauer and Mazzoni18 in 2011, and Kitago and change in motor behavior that is driven by repeated task-
Krakauer19 in 2013. Expanding our motor learning knowl- specific practice. In use-dependent learning, repeated practice
edge base to include these findings will allow physical ther- of a new movement causes future repetitions to be more simi-
apists to diversify the means through which they can influ- lar to the practiced behavior. For instance, a golfer repeatedly
ence human movement. This is critical to the advancement practices a new swing technique so that future attempts to
of movement rehabilitation. Unfortunately, these discoveries produce the skilled behavior will result in the expression of the
have remained largely confined within the research commu- well-practiced movement pattern. This form of motor learning
nity, despite their relevance to physical therapy (as discussed has been well studied and is widely accepted as a powerful
in a recent review by Roemmich and Bastian20 ). treatment tool to promote changes in motor behavior. A large
To translate these scientific advances into evidence-based body of work has demonstrated that the primary behavioral
practice, we must first integrate them into our current driver of use-dependent motor learning is the amount of
understanding of motor learning. In this Perspective, we practice completed by the learner.24 Importantly, researchers
describe 4 well-studied mechanisms of human motor learning: have also identified a number of practice parameters that
use-dependent, instructive, reinforcement, and sensorimotor can be manipulated to optimize use-dependent learning (task
adaptation. Specifically, we outline the current explanatory specificity, aerobic intensity, etc.).25
framework for each of these motor learning mechanisms Improvements in motor behavior resulting from repeated
and we discuss the practical implications of each mechanism practice are mediated by structural and functional changes
for physical therapist practice and education. Broadening throughout the central nervous system26–29 (referred to as
our understanding of motor learning to include the multiple experience-dependent neuroplasticity). Thus, there is a natural
Leech et al 3

Downloaded from [Link] by guest on 01 July 2023

Figure 1. Key features of 4 distinct motor learning mechanisms. This represents a simplified summary of a large body of research aimed to understand
how changes in human motor behavior occur. Alternative terms commonly used in the literature for these mechanisms and their features are provided in
parentheses.

overlap between the drivers of use-dependent motor learning an intentional movement strategy.36,37 This mechanism of
and the canonical principles of experience-dependent neu- motor learning is elicited when a learner is provided specific
roplasticity as described by Kleim and Jones30 in 2008— external feedback about a movement error or performance
commonly taught as the guiding principles to optimize motor relative to a task goal (ie, knowledge of performance) that
behavior change in many current doctor of physical therapy prompts the development of an intentional error-reducing
curricula. movement strategy. An important component of instructive
Another important feature of use-dependent motor learning motor learning is that the error-reducing movement strategy
is the degree of cognitive processing that is required. Studies can be explicitly described by the learner38 and reproduced
have demonstrated that passive or overly guided movement in the future.39 In the context of physical therapy, the error-
practice (eg, assisted passive range of motion, manual facil- reducing movement strategy is often instructed or cued by a
itation of a movement pattern) does not lead to changes in physical therapist (further described below). Therefore, here
motor behavior31 or neuroplasticity.32 As Bernstein described, we refer to this mechanism as instructive motor learning, but
“practice . . . does not consist of repeating the means of it should be noted that it is also commonly termed explicit or
solution of a motor problem time after time, but in the process strategy-based motor learning.40
of solving this problem again and again by techniques we Though not widely referred to as “instructive motor learn-
have changed and perfected from repetition to repetition”; ing” in the clinic, this learning mechanism is a staple in rehabil-
rather, “practice is a particular type of repetition without repe- itation interventions. Physical therapists often provide exter-
tition.”33 For motor practice to drive use-dependent learning, nal feedback about undesirable movements (eg, verbal and/or
the learner must be exerting motor effort to actively prac- visual cues about uneven step lengths during walking) and
tice the task. In addition to active movement, use-dependent explicitly instruct corrective movement strategies (eg, “take
learning also requires a degree of cognitive engagement in the a bigger step with your left leg”). Moreover, a majority of
task—that is, the participant must understand the task goal the investigators conducting early behavioral motor learning
and make intentional changes to their movements to solve research related to physical therapy focused on understanding
the movement problem and achieve that goal. Ultimately, instructive learning. Many studies were conducted to explore
extensive task practice of this sort will automatize behaviors the most effective modality (eg, auditory, visual, tactile) and
and reduce the cognitive load required to complete them.34 timing of external feedback to provide to the learner.41 As
One of the disadvantages of use-dependent motor learning such, instructive learning is a major focus of motor learning–
is that it occurs over a long timescale. This means that lasting related content in physical therapy curricula.
improvements in motor behavior are slow to accumulate The prevailing theory is that instructive motor learning is
(ie, over weeks or months)23 and the generalization from largely mediated by structures in the prefrontal cortex40,42
training to real-world improvement can be quite small35 with and involves multiple cognitive processes (eg, comprehension
interventions that primarily depend on use-dependent motor of instructions, performance monitoring).43,44 This is sup-
learning. One approach to elicit more immediate changes ported by evidence that individuals with frontal lobe lesions
in motor behavior is to combine use-dependent learning in and neuropsychological impairments have difficultly develop-
parallel with instructive motor learning, which occurs on a ing38 and retaining36 explicit strategies to correct undesirable
much faster timescale. This is discussed in more detail in the movements. A recent study in individuals poststroke also
section below. demonstrates a strong relationship between cognition and the
capacity for instructive motor learning.45 The high cognitive
burden of instructive learning is likely intuitive to many
Instructive Motor Learning clinicians; an external feedback-heavy treatment approach is
Instructive motor learning (Fig. 1; green) is defined as a often ineffective for improving motor behaviors in patients
change in motor behavior achieved through the use of with cognitive deficits.
4 Updates in Motor Learning

One key advantage of instructive motor learning is that basal ganglia and primary motor cortex within the basal
it occurs on a relatively fast timescale when cognition is ganglia-thalamo-cortical circuits.
intact. When the correct movement strategy is executed, motor The involvement of cognitive processes in reinforcement
behavior can improve markedly within a single session or motor learning is unclear. Traditionally, it was thought that
even from one repetition to the next. Importantly, move- reinforcement learning was a more implicit or automatic
ments acquired through instructive learning can be voluntarily process. One theory is that a successful movement would
recalled when the learner is prompted to use the newly learned spontaneously occur during practice due to the inherent vari-
movement strategy in the future.39 Retention, defined as the ability in the individual’s movement, and this would lead to
persistence of a behavior beyond the training period, is a an automatic biasing of future movement selection toward the
hallmark of motor learning. This differentiates instructive successful or rewarding outcome.46 However, it now appears
learning-dependent improvements in motor behavior from that intentional exploration of different movements in search
transient changes in motor performance.41 of the successful behavior may be critical for reinforcement
Consider a patient poststroke who walks with reduced learning.51 Additionally, explicit control or development of
stance time on their weaker leg. A therapy session structured an intentional strategy may also play an important role in
reinforcement motor learning.52 From this emerging work,

Downloaded from [Link] by guest on 01 July 2023

to make use of instructive motor learning (eg, external feed-
back about stance time symmetry) would enable this patient we suggest reinforcement motor learning may require more
to learn a new movement strategy to walk with a longer stance cognitive processing than originally proposed.
time on their weaker leg by the end of the session. Experienced Reinforcement learning can lead to sustained behav-
clinicians are aware that this patient will most likely return ioral improvements within a session. However, relative
for the next visit exhibiting the same stance time asymmetry to sensorimotor-based adaptation (discussed below), these
as they did prior to the last treatment session. From this, one improvements take longer to develop.53 Despite the apparent
may assume that the previously employed movement strategy slower rate of learning, reinforcement is associated with
to reduce stance time asymmetry was not retained. However, longer retention of acquired movements.53,54 This feature of
to assess retention of a learned movement strategy, the learner reinforcement learning is particularly attractive for designing
must be prompted or instructed to reproduce the learned interventions with long-lasting effects. Recently, researchers
behavior. If the patient can voluntarily produce a desired have started to study how reinforcement-driven retention can
motor behavior after initial learning, then the instructive be coupled with learning mechanisms that drive faster initial
motor learning was retained. Importantly, whereas movement changes in motor behavior to be both quickly acquired and
strategies can be retained and explicitly recalled in the future, long-lasting. For instance, research groups that add success-
the extensive practice that drives use-dependent learning is based feedback to sensorimotor adaptation55,56 or protocols
likely necessary for the new movement to become automatic that require practice of a novel task54 have found improved
or habitual.20,34 Development of a new movement habit may retention relative to execution of those protocols without
also require practicing within multiple contexts that require engaging reinforcement learning.
the patient to employ that movement strategy in order to be
successful. However, identification of the key ingredients that
drive a patient to transition from voluntary recall of a learned Sensorimotor Adaptation-Based Motor Learning
movement strategy to habitual use of that behavior remains a Sensorimotor adaptation-based motor learning (also termed
critically important goal of ongoing investigation. sensorimotor adaptation; Fig. 1; purple) is defined as a
change in motor behavior that is driven by sensory prediction
errors.57 Sensory prediction errors are perceived in the
Reinforcement Motor Learning nervous system when the actual sensory consequence of a
Reinforcement motor learning (Fig. 1; red) is defined as an movement (most commonly detected by visual, auditory, or
improvement in motor behavior that is driven by binary proprioceptive feedback) differs from the predicted sensory
outcome-based feedback. That is, reinforcement learning consequence of that movement. This most often occurs
depends on external feedback about the success or failure when individuals encounter unexpected task demands or
of the movement relative to a task goal (ie, knowledge of changes in the environment (ie, perturbations) that require
results). After a movement success or failure, the learner does modifications to the executed motor program. When these
not receive information about how the movement needs to errors are detected, the motor command is automatically
be modified (or not) in order to be successful.21 This type updated to adapt the movement and reduce the magnitude
of nondirectional external feedback prompts the learner to of the prediction error (see Bastian58 for review). Imagine, for
explore different movements and select actions that have the example, the experience of driving a car in which the brake
highest probability of success, while avoiding actions with sensitivity is different than your own. Applying a certain
low probability of success (reviewed in Sutton and Barto46 amount of pressure to the brake pedal will cause this car
and Lee et al47 ). to stop more abruptly than anticipated. The perception of
The basal ganglia are thought to be the primary neural this sensory prediction error will prompt an adaption of the
substrates involved in selecting movements that result in movement to reduce the pressure applied to the brake in the
task success.48 This is potentially mediated by reward-based future, allowing you to decelerate the car smoothly. This is
dopamine signaling, where activity of dopaminergic neurons the same process that occurs, for instance, when a patient is
increases in response to task success during early stages of learning to navigate a manual wheelchair over carpet versus
practice.49 The primary motor cortex may also be involved tile or more generally when therapists vary the conditions
in reinforcement learning, because recent evidence shows that of movement practice according to Gentile’s taxonomy of
reinforcement learning leads to plasticity in the primary motor tasks.59 The phenomenon of sensorimotor adaptation aligns
cortex.50 This may be due to the connections between the with the recognition schema proposed by Schmidt60 in
Leech et al 5

1975, and allows for an impressive degree of movement Implications for Physical Therapist
flexibility. It has been demonstrated across many types of Practice and Education
movements including eye movements,61 arm movements,62 We now appreciate that there are multiple, distinct, co-
and walking.63 occurring motor learning mechanisms that subserve sustained
Our current understanding is that this mechanism of motor changes in movement. This more comprehensive understand-
learning is primarily dependent on the cerebellum.57 The ing of motor learning has powerful implications for physical
cerebellum is known to be involved in mapping outgoing therapy. Most importantly, it diversifies the methods that
motor commands to the predicted sensory feedback from clinicians have at their disposal to promote motor learning.
that movement.64 Moreover, the ability to update motor Physical therapist interventions across practice settings are
commands and adapt movements in response to sensorimotor currently structured to predominantly leverage use-dependent
prediction errors is impaired in individuals with cerebellar and instructive motor learning. Moving forward, physical
disorders.65–67 Importantly, the capacity for sensorimotor therapists must begin also to consider reinforcement and
adaptation-based learning may be related to the severity of sensorimotor adaptation as part of their arsenal. These 4
cerebellar degeneration.68 motor learning mechanisms can occur in parallel as a new

Downloaded from [Link] by guest on 01 July 2023

Sensorimotor adaptation is thought to occur automatically movement is being learned, and the relative contribution of
and implicitly, independent of intentional, voluntary modifi- each can be manipulated by incorporating their unique drivers
cations to movements. This is demonstrated by evidence that into movement practice. Physical therapist interventions can
sensorimotor adaptation still occurs even when people are be carefully designed to leverage any or all of the 4 motor
instructed how to correct their movements,69 provided exter- learning mechanisms described here.
nal error feedback,70 or prevented from correcting their move- This point is best illustrated with clinical examples. Con-
ment errors.71 Interestingly, mounting evidence from studies sider a patient with shoulder impingement who has a goal to
of upper limb movements suggests sensorimotor adaptation reach overhead without pain. To achieve this goal, a physical
often occurs in parallel with instructive or strategy-based therapist has the patient practice placing items of similar
motor learning.40 This indicates that the overall reduction in size but different weights onto an overhead shelf using a
movement errors in many of the studied tasks may be due movement pattern that does not cause pain. This intervention
to simultaneous sensorimotor adaptation and development of structure integrates 3 behavioral drivers that will promote dif-
explicit strategies to counteract an experimental perturbation. ferent motor learning mechanisms to contribute to the overall
Evidence for the involvement of strategy-based learning dur- change in reaching movement pattern (Fig. 2A): (1) repeated
ing the adaptation of walking (a more continuous, patterned, reaching practice will drive use-dependent motor learning,
automatic behavior) is emerging but less well established. In a (2) if the patient considers a pain-free reach a successful
study of locomotor adaptation the authors found that partic- outcome, using a movement pattern that does not cause pain
ipants, when instructed, could voluntarily reproduce part of would drive reinforcement motor learning, and (3) finally,
a new walking pattern that was learned through sensorimo- practicing lifting objects of the same size but different weights
tor adaptation.39 This suggests that cognitive strategies may will induce sensorimotor prediction errors that drive senso-
also contribute to locomotor adaptation, but more work is rimotor adaptation to promote flexibility of the pain-free
necessary to understand how these voluntary and involuntary reaching movement. With this intervention, the contribution
motor learning mechanisms interact to elicit an overall change of instructive motor learning is relatively small. If the patient
in behavior during walking. has difficulty reaching overhead without pain or adapting
Sensorimotor adaptation occurs on a very rapid timescale, the movement to account for objects of different weights, the
generally resulting in a change in motor behavior over therapist could consider modifying the intervention to include
minutes. Movements are adjusted repetition-by-repetition verbal cues for proper scapular dynamics during the reaching
to quickly reduce sensory prediction errors, allowing for movement. This would increase the relative contribution of
an impressive flexibility of movement in the context of instructive learning.
many different task demands.58 Importantly, these rapid Another example that demonstrates how physical therapists
adjustments are not simply transient changes in movement can carefully design interventions to target these learning
performance. Rather, the newly learned movement must mechanisms is a patient with Parkinson disease who has
actively be unlearned. That is, when the error-inducing a goal of improving their gait speed. To achieve this goal,
perturbation is removed, participants will initially exhibit the therapist has the patient practice walking on a treadmill
movement errors in the opposite direction until the adapted while providing verbal and visual cues to increase right and
motor command is unlearned through the same error- left step lengths (Fig. 2B). With this intervention, massed
reducing process (this behavioral phenomenon is called stepping practice will drive use-dependent motor learning,
an “aftereffect”). There is also evidence that the adapted and the performance-based feedback from external cues
motor behavior is automatically stored for future use. about step lengths will engage instructive motor learning.
When participants are reexposed to a perturbation, they To increase the contribution of sensorimotor adaption, the
(1) make smaller movement errors initially and (2) relearn therapist could have the patient practice walking with longer
the new movement faster.72 These behavioral phenomena step lengths over surfaces with variable levels of compliance
are called “recall” and “savings,” respectively, and they that would include sensorimotor prediction errors (ie, the
are considered evidence of lasting motor memories formed same propulsive force would lead to different step lengths
through sensorimotor adaptation. Repeated exposure to a relative to the initial practice conditions). Finally, in this
perturbation may ultimately lead to the establishment of a case, the therapist would likely try to avoid an intervention
new, independent motor command,73 but the time course of structure that would rely on outcomes-based feedback,
this process across motor tasks is unclear.74,75 because attempting to leverage reinforcement motor learning
6 Updates in Motor Learning

Downloaded from [Link] by guest on 01 July 2023

Figure 2. Motor learning mechanisms driven by example physical therapist interventions to treat shoulder impingement (A) and gait impairment (B). The
relative contribution of each mechanism (represented by circle size and fill) to the targeted motor behavior change depends on the behavioral drivers that
are integrated into the intervention by the physical therapist. Note that multiple mechanisms may often occur in parallel as a motor behavior is being
learned.

in a patient with a pathology of the basal ganglia may be the physical therapist’s approach to movement rehabilitation
unsuccessful. across patient populations and areas of practice.
The multiplicity of motor learning mechanisms that con- Research related to the motor learning mechanisms
tribute to sustained improvements in motor behavior is likely described here is progressing rapidly, and the field has clearly
to influence practice in 2 main ways. First, prior to treatment advanced beyond what is currently being taught in most entry-
the physical therapist can strategically design an intervention level doctor of physical therapy curricula. Given the broad
that accounts for individual patient characteristics and tar- applicability of motor learning to all areas of physical therapy,
gets the motor learning mechanisms most likely to change this progress warrants an update to the motor learning–
motor behavior. For example, if a patient exhibits cognitive related knowledge base of physical therapists. Providing
impairments, the physical therapist might structure the inter- newly trained physical therapists with a comprehensive
vention to preferentially target motor learning mechanisms understanding of motor learning research will promote the
that require less cognitive processing (eg, sensorimotor adap- translation of up-to-date motor learning principles that can
tation). Another important patient characteristic to consider be used to guide clinical practice.
is the integrity of different areas of the nervous system, given
that each mechanism of motor learning is primarily subserved
by a distinct neural substrate. This is specifically important Future Research Directions
in the context of a patient with neurological damage or As we begin the process of translating these scientific advances
disease. For instance, a person with cerebellar degeneration into clinical practice, it is important to be aware of the
may benefit from a physical therapist intervention that targets questions that remain unanswered. First, we do not yet know
learning through reinforcement, instructive, or use-dependent how to optimize long-lasting motor learning through sensori-
motor learning rather than sensorimotor adaptation. motor adaptation or reinforcement learning mechanisms. Our
Secondly, having access to multiple mechanisms of motor foundational understanding of motor learning optimization
learning may improve the therapist’s capacity to adjust (well-known principles of neuroplasticity and practice struc-
interventions based on patient responsiveness throughout ture) was established primarily through studies of instructive
the course of treatment. Recent evidence demonstrates that and use-dependent learning. The degree to which these same
there are interindividual differences in the way people learn optimization principles translate to sensorimotor adaptation
new movements. One study found that approximately 20% or reinforcement learning is currently unknown. The results
of neurotypical participants were not able to learn a new of a recent study76 suggest that the same “rules of thumb”
upper extremity movement through reinforcement.51 This may not apply to sensorimotor adaptation, highlighting the
suggests that when a patient with any diagnosis exhibits a need for additional research to fully understand how we might
lack of responsiveness to an intervention, it may be useful to optimize learning through each of these mechanisms.77
intentionally integrate another motor learning mechanism Future research is also needed to elucidate how different
into the treatment approach. The ability to sharpen our learning mechanisms interact during the course of learning a
clinical practice in these ways will have a significant impact on new motor behavior. As previously mentioned, even though
Leech et al 7

needed to test the efficacy of these different learning mecha-

nisms to learn a wide range of motor tasks.
Perhaps the most limiting gap in our knowledge is that we
do not yet know how these motor learning mechanisms can
be harnessed to address limitations in activity or participation.
The majority of motor learning studies, even those conducted
with clinical populations, have been designed to evaluate the
effect of different motor learning mechanisms on changing
specific movement kinematics. However, the link between
kinematic changes and functional improvement is not clear.
Therefore, further research is needed to determine how each
of these distinct motor learning mechanisms can be effectively
Figure 3. A proposed contribution of use-dependent, instructive, integrated into the design of functional mobility interventions.
reinforcement, and sensorimotor adaptation motor learning mechanisms Finally, it is important to note that studies focused on

Downloaded from [Link] by guest on 01 July 2023

to the stages of motor learning described by Fitts and Posner.34 the translation of these advances in motor learning research
into clinically applicable treatment approaches are largely
preliminary. To date, few studies have evaluated the effects
of long-term interventions that are primarily underpinned by
we may be able to influence their relative contributions by instructive, reinforcement, or sensorimotor adaptation-based
carefully controlling practice parameters, multiple learning motor learning. Whereas a number of studies have tested
mechanisms are often engaged in parallel during a single the effects of multiday training protocols in neurotypical and
learning experience. It is of particular importance to under- patient populations, very few studies have evaluated longer-
stand how the relative contributions of different mechanisms term interventions that are designed to intentionally harness
change over the course of learning. Considering the well- these motor learning mechanisms. Future work must expand
known stages of learning described by Fitts and Posner34 upon the large body of work that demonstrates the efficacy of
(Fig. 3), it is possible that instructive and reinforcement motor use-dependent learning-based physical therapist interventions
learning are more heavily engaged in the initial, cognitive stage and (1) determine how to translate our knowledge of these
of learning when performance is variable and the learner has other motor learning mechanisms into clinically applicable
to explore different movement solutions, whereas the associa- interventions, and (2) assess the effects of these interventions
tive stage engages primarily reinforcement and use-dependent on motor behaviors of interest.
mechanisms as the learner continues to refine their behavior
with extensive practice. In the autonomous stage, further
adjustments to motor behavior and a shift to automaticity
may primarily be driven by sensorimotor adaptation and
Conclusions
continued movement practice (similar to that suggested by the The concepts described here denote an expansion of the scien-
authors of previous reviews21,40 ). This working theory aligns tific theory related to human motor learning. We now appre-
with Gentile’s description of explicit and implicit processes ciate that motor learning occurs through multiple distinct
that contribute to motor skill acquisition.78 Future research mechanisms and neural processes. A more comprehensive
is needed to investigate these proposed weightings across the understanding of motor learning has many practical impli-
time course of learning because this information would allow cations for physical therapy and movement rehabilitation. It
physical therapists to strategically evolve their treatments to is critical that we integrate this information into our work-
support different phases of learning over time. Improving ing knowledge of motor learning. Doing so will enrich our
our understanding of how each of these mechanisms con- understanding of the science that motivates our interventions
tributes to motor learning over time also holds the potential and is likely to stimulate clinical and research innovation.
to disentangle what is needed to facilitate the formation of This process—evolution of thought and behavior based on
a habitual motor behavior—the end goal of most movement new information—is essential for continued evidence-based
rehabilitation. practice in physical therapy.
Thus far, research about these motor learning mechanisms
has primarily involved the study of upper extremity motor
tasks. However, upper extremity, lower extremity, and whole-
body movements are thought to be largely controlled by
Author Contributions
different components of the central nervous system (ie, corti- Concept/idea/research design: K.A. Leech, J. Gordon,
cal, spinal, propriospinal controllers, respectively). Therefore, D.S. Reisman, K.M. Cherry-Allen
Writing: K.A. Leech, R.T. Roemmich, J. Gordon,
there may be differences in how motor learning occurs and is
K.M. Cherry-Allen
promoted with each type of motor task. For example, massed
Project management: K.A. Leech
reaching practice likely involves more cognitive processing Fund procurement: K.A. Leech
than massed stepping practice. However, even with tasks like
walking that are considered more automatic, the structure
of an intervention can influence the amount of cognitive
processing required (eg, gait training with obstacle negotiation Funding
or community mobility). Nonetheless, active engagement in This work was supported by a Magistro Family Foundation Research
the task is a baseline requirement to promote use-dependent Grant from the Foundation for Physical Therapy Research and a grant
learning during both reaching and walking. Future research is from the National Institutes of Health (ref. no. K12 HD055929–14).
8 Updates in Motor Learning

Role of the Funding Source 19. Kitago T, Krakauer JW. Motor learning principles for neuroreha-
bilitation. Handb Clin Neurol. 2013;110:93–103.
The funders played no role in the writing of this article.
20. Roemmich RT, Bastian AJ. Closing the loop: from motor neuro-
science to neurorehabilitation. Annu Rev Neurosci. 2018;41:415–
429.
Disclosures 21. Spampinato D, Celnik P. Multiple motor learning processes in
The authors completed the ICMJE Form for Disclosure of Potential humans: defining their neurophysiological bases. Neuroscientist.
Conflicts of Interest and reported no conflicts of interest. 2021;27:246–267.
22. Reisman DS, McLean H, Keller J, Danks KA, Bastian AJ. Repeated
split-belt treadmill training improves poststroke step length asym-
metry. Neurorehabil Neural Repair. 2013;27:460–468.
References 23. Wolf SL, Winstein CJ, Miller JP, et al. Effect of constraint-
1. Sullivan KJ. President’s perspective. J Neurol Phys Ther. induced movement therapy on upper extremity function 3 to 9
2009;33:58–59. months after stroke: the EXCITE randomized clinical trial. JAMA.
2. Scheets PL, Hornby TG, Perry SB, et al. Moving forward. J Neurol 2006;296:2095–2104.
Phys Ther. 2021;45:46–49. 24. Lohse KR, Lang CE, Boyd LA. Is more better? Using metadata

Downloaded from [Link] by guest on 01 July 2023

3. Fisher BE, Morton SM, Lang CE. From motor learning to physical to explore dose-response relationships in stroke rehabilitation.
therapy and back again: the state of the art and science of motor Stroke. 2014;45:2053–2058.
learning rehabilitation research. J Neurol Phys Ther. 2014;38:149– 25. Hornby TG, Straube DS, Kinnaird CR, et al. Importance of speci-
150. ficity, amount, and intensity of locomotor training to improve
4. Schmidt RA. Motor learning principles for physical therapy. In: ambulatory function in patients poststroke. Top Stroke Rehabil.
Lister M, ed., Contemporary Management of Motor Control Prob- 2011;18:293–307.
lems. Proceedings of the II STEP Conference. Alexandria, VA, USA: 26. Adkins DL, Boychuk J, Remple MS, Kleim JA. Motor training
Foundation for Physical Therapy; 1991: 49–63. induces experience-specific patterns of plasticity across motor
5. Winstein C, Lewthwaite R, Blanton SR, Wolf LB, Wishart L. cortex and spinal cord. J Appl Physiol (1985). 2006;101:
Infusing motor learning research into neurorehabilitation practice: 1776–1782.
a historical perspective with case exemplar from the accelerated 27. Winchester P, McColl R, Querry R, et al. Changes in supraspinal
skill acquisition program. J Neurol Phys Ther. 2014;38:190–200. activation patterns following robotic locomotor therapy in motor-
6. Boudreau SA, Farina D, Falla D. The role of motor learning and incomplete spinal cord injury. Neurorehabil Neural Repair.
neuroplasticity in designing rehabilitation approaches for muscu- 2005;19:313–324.
loskeletal pain disorders. Man Ther. 2010;15:410–414. 28. Knikou M. Plasticity of corticospinal neural control after loco-
7. van Vliet PM, Heneghan NR. Motor control and the management motor training in human spinal cord injury. Neural Plast.
of musculoskeletal dysfunction. Man Ther. 2006;11:208–213. 2012;2012:254948.
8. Low M. A time to reflect on motor control in musculoskeletal 29. Thompson AK, Wolpaw JR. H-reflex conditioning during locomo-
physical therapy. J Orthop Sports Phys Ther. 2018;48:833–836. tion in people with spinal cord injury. J Physiol. 2021;599:2453–
9. van Dillen LR, Lanier VM, Steger-May K, et al. Effect of motor skill 2469.
training in functional activities vs strength and flexibility exercise 30. Kleim JA, Jones TA. Principles of experience-dependent neural
on function in people with chronic low back pain: a randomized plasticity: implications for rehabilitation after brain damage. J
clinical trial. JAMA Neurol. 2021;78:385–395. Speech Lang Hear Res JSLHR. 2008;51:S225–S239.
10. Dumoulin C, Hay-Smith J, Habée-Séguin GM, Mercier J. Pelvic 31. Beets IAM, Macé M, Meesen RLJ, Cuypers K, Levin O, Swinnen
floor muscle training versus no treatment, or inactive control SP. Active versus passive training of a complex bimanual task:
treatments, for urinary incontinence in women: a short version is prescriptive proprioceptive information sufficient for inducing
Cochrane systematic review with meta-analysis. Neurourol Uro- motor learning? PLoS One. 2012;7:e37687.
dyn. 2015;34:300–308. 32. Lotze M, Braun C, Birbaumer N, Anders S, Cohen LG. Motor
11. Wu YM, Mclnnes N, Leong Y. Pelvic floor muscle training versus learning elicited by voluntary drive. Brain J Neurol. 2003;126:866–
watchful waiting and pelvic floor disorders in postpartum women: 872.
a systematic review and meta-analysis. Female Pelvic Med Reconstr 33. Bernstein N. The Coordination and Regulation of Movements.
Surg. 2018;24:142–149. New York, NY: Pergamon; 1967.
12. Lin C-HJ, Winstein CJ, Fisher BE, Wu AD. Neural correlates of 34. Fitts PM, Posner MI. Human Performance. Belmont, CA, USA:
the contextual interference effect in motor learning: a transcranial Brooks/Cole; 1971.
magnetic stimulation investigation. J Mot Behav. 2010;42:223– 35. Waddell KJ, Strube MJ, Bailey RR, et al. Does task-specific training
232. improve upper limb performance in daily life poststroke? Neurore-
13. Memmert D. Long-term effects of type of practice on the learn- habil Neural Repair. 2017;31:290–300.
ing and transfer of a complex motor skill. Percept Mot Skills. 36. French MA, Morton SM, Reisman DS. Use of explicit processes
2006;103:912–916. during a visually guided locomotor learning task predicts 24-hour
14. Winstein CJ. Knowledge of results and motor learning— retention after stroke. J Neurophysiol. 2021;125:211–222.
implications for physical therapy. Phys Ther. 1991;71:140–149. 37. Schween R, McDougle SD, Hegele M, Taylor JA. Assessing
15. Salmoni AW, Schmidt RA, Walter CB. Knowledge of results and explicit strategies in force field adaptation. J Neurophysiol.
motor learning: a review and critical reappraisal. Psychol Bull. 2020;123:1552–1565.
1984;95:355–386. 38. Slachevsky A, Pillon B, Fourneret P, et al. The prefrontal cortex and
16. Bland MD, Birkenmeier RL, Barco P, Lenard E, Lang CE, Lenze EJ. conscious monitoring of action. Neuropsychologia. 2003;41:655–
Enhanced medical rehabilitation: effectiveness of a clinical training 665.
model. NeuroRehabilitation. 2016;39:481–498. 39. French MA, Morton SM, Charalambous CC, Reisman DS. A
17. Wulf G, Lewthwaite R. Optimizing performance through intrinsic locomotor learning paradigm using distorted visual feedback elicits
motivation and attention for learning: the OPTIMAL theory of strategic learning. J Neurophysiol. 2018;120:1923–1931.
motor learning. Psychon Bull Rev. 2016;23:1382–1414. 40. Taylor JA, Ivry RB. The role of strategies in motor learning. Ann
18. Krakauer JW, Mazzoni P. Human sensorimotor learning: adap- N Y Acad Sci. 2012;1251:1–12.
tation, skill, and beyond. Curr Opin Neurobiol. 2011;21: 41. Schmidt RA, Lee TD. Motor Control and Learning: A Behavioral
636–644. Emphasis. 5th ed. Champaign, IL, USA: Human Kinetics; 2011.
Leech et al 9

42. Taylor JA, Ivry RB. Cerebellar and prefrontal cortex contributions 61. Wallman J, Fuchs AF. Saccadic gain modification: visual error
to adaptation, strategies, and reinforcement learning. Prog Brain drives motor adaptation. J Neurophysiol. 1998;80:2405–2416.
Res. 2014;210:217–253. 62. Shadmehr R, Mussa-Ivaldi FA. Adaptive representation of dynam-
43. Seidler RD, Bo J, Anguera JA. Neurocognitive contributions to ics during learning of a motor task. J Soc Neurosci. 1994;14:3208–
motor skill learning: the role of working memory. J Mot Behav. 3224.
2012;44:445–453. 63. Reisman DS, Block HJ, Bastian AJ. Interlimb coordination during
44. Seidler RD, Kwak Y, Fling BW, Bernard JA. Neurocognitive locomotion: what can be adapted and stored? J Neurophysiol.
mechanisms of error-based motor learning. Adv Exp Med Biol. 2005;94:2403–2415.
2013;782:39–60. 64. Blakemore SJ, Frith CD, Wolpert DM. The cerebellum is involved
45. French MA, Cohen ML, Pohlig RT, Reisman DS. Fluid cognitive in predicting the sensory consequences of action. Neuroreport.
abilities are important for learning and retention of a new, explic- 2001;12:1879–1884.
itly learned walking pattern in individuals after stroke. Neuroreha- 65. Morton SM. Cerebellar contributions to locomotor adaptations
bil Neural Repair. 2021;35:419–430. during splitbelt treadmill walking. J Neurosci. 2006;26:9107–
46. Sutton R, Barto G. An Introduction to Reinforcement Learning. 9116.
Cambridge, MA, USA: MIT Press; 1998. 66. Morton SM, Bastian AJ. Prism adaptation during walking gen-
47. Lee D, Seo H, Jung MW. Neural basis of reinforcement learning eralizes to reaching and requires the cerebellum. J Neurophysiol.

Downloaded from [Link] by guest on 01 July 2023

and decision making. Annu Rev Neurosci. 2012;35:287–308. 2004;92:2497–2509.
48. Schultz W. Reward functions of the basal ganglia. J Neural Transm 67. Lang CE, Bastian AJ. Cerebellar subjects show impaired adap-
(Vienna). 2016;123:679–693. tation of anticipatory EMG during catching. J Neurophysiol.
49. Schultz W. Dopamine reward prediction error coding. Dialogues 1999;82:2108–2119.
Clin Neurosci. 2016;18:23–32. 68. Statton MA, Vazquez A, Morton SM, Vasudevan EVL, Bastian AJ.
50. Uehara S, Mawase F, Celnik P. Learning similar actions by rein- Making sense of cerebellar contributions to perceptual and motor
forcement or sensory-prediction errors rely on distinct physiologi- adaptation. Cerebellum. 2018;17:111–121.
cal mechanisms. Cereb Cortex. 2018;28:3478–3490. 69. Mazzoni P, Krakauer JW. An implicit plan overrides an
51. Uehara S, Mawase F, Therrien AS, Cherry-Allen KM, Celnik P. explicit strategy during visuomotor adaptation. J Soc Neurosci.
Interactions between motor exploration and reinforcement learn- 2006;26:3642–3645.
ing. J Neurophysiol. 2019;122:797–808. 70. Roemmich RT, Long AW, Bastian AJ. Seeing the errors you feel
52. Holland P, Codol O, Galea JM. Contribution of explicit pro- enhances locomotor performance but not learning. Curr Biol.
cesses to reinforcement-based motor learning. J Neurophysiol. 2016;26:2707–2716.
2018;119:2241–2255. 71. Long AW, Roemmich RT, Bastian AJ. Blocking trial-by-trial error
53. Therrien AS, Wolpert DM, Bastian AJ. Effective reinforcement correction does not interfere with motor learning in human walk-
learning following cerebellar damage requires a balance between ing. J Neurophysiol. 2016;115:2341–2348.
exploration and motor noise. Brain. 2016;139:101–114. 72. Klassen J, Tong C, Flanagan JR. Learning and recall of incremental
54. Abe M, Schambra H, Wassermann EM, Luckenbaugh D, kinematic and dynamic sensorimotor transformations. Exp Brain
Schweighofer N, Cohen LG. Reward improves long-term retention Res. 2005;164:250–259.
of a motor memory through induction of offline memory gains. 73. Reisman DS, Bastian AJ, Morton SM. Neurophysiologic and reha-
Curr Biol. 2011;21:557–562. bilitation insights from the split-belt and other locomotor adapta-
55. Shmuelof L, Huang VS, Haith AM, Delnicki RJ, Mazzoni P, tion paradigms. Phys Ther. 2010;90:187–195.
Krakauer JW. Overcoming motor “forgetting” through reinforce- 74. Leech KA, Day KA, Roemmich RT, Bastian AJ. Movement and
ment of learned actions. J Soc Neurosci. 2012;32:14617–14621. perception recalibrate differently across multiple days of locomo-
56. Quattrocchi G, Greenwood R, Rothwell JC, Galea JM, Bestmann tor learning. J Neurophysiol. 2018;120:2130–2137.
S. Reward and punishment enhance motor adaptation in stroke. J 75. Martin TA, Keating JG, Goodkin HP, Bastian AJ, Thach WT.
Neurol Neurosurg Psychiatry. 2017;88:730–736. Throwing while looking through prisms. II. Specificity and
57. Tseng Y-W, Diedrichsen J, Krakauer JW, Shadmehr R, Bastian AJ. storage of multiple gaze-throw calibrations. Brain. 1996;119:
Sensory prediction errors drive cerebellum-dependent adaptation 1199–1211.
of reaching. J Neurophysiol. 2007;98:54–62. 76. Helm EE, Pohlig RT, Kumar DS, Reisman DS. Practice structure
58. Bastian AJ. Understanding sensorimotor adaptation and learning and locomotor learning after stroke. J Neurol Phys Ther JNPT.
for rehabilitation. Curr Opin Neurol. 2008;21:628–633. 2019;43:85–93.
59. Gentile AM. Skill acquisition: action, movement, and the neuro- 77. Leech KA, Holleran CL. Commentary on: “practice structure and
motor processes. In: Movement Science: Foundations for Physical locomotor learning after stroke”. J Neurol Phys Ther. 2019;43:94–
Therapy in Rehabilitation. Rockville, MD, USA: Aspen; 1987: 93– 95.
154. 78. Gentile AM. Movement science: implicit and explicit processes dur-
60. Schmidt RA. A schema theory of discrete motor skill learning. ing acquisition of functional skills. Scand J Occup Ther. 1998;5:7–
Psychol Rev. 1975;82:225–260. 16.