THERAPY Magazin
Differentiated learning strategies in gait rehabilitation
New research reveals how stroke patients differ in explicit and implicit motor learning—offering insights for tailored gait rehabilitation strategies that target both cognitive and physical recovery.
Jakob Tiebel
Business Owner, N+ Digital Health Agency
New insights into the influence of explicit and implicit motor learning strategies on the improvement of walking ability in post-stroke patients
The study highlights the significance of explicit and implicit motor learning in the rehabilitation of stroke patients. Whereas explicit learning takes place through conscious control and feedback, implicit learning is based on unconscious adaptations to sensory stimuli. Both mechanisms are crucial for the restoration of walking ability, but they are differently impaired after a stroke – with a greater impairment of explicit processes. The results suggest that a combined use of both approaches could optimise rehabilitation. To further improve clinical applicability, however, a deeper investigation of their interactions as well as adaptation to various degrees of impairment is needed.
Background
Stroke is one of the most common causes of permanent disability in adulthood worldwide. Typical consequences include asymmetrical
gait patterns, reduced walking speed and an increased risk of falling, which significantly impair the quality of life of those affected. Gait rehabilitation aims to improve these deficits through targeted motor learning strategies. Both explicit (conscious) and implicit (unconscious) learning mechanisms play a key role in this.
Explicit motor learning is consciously controlled through instructions and feedback and requires the active planning and adaptation of movements, which are primarily processed in the prefrontal cortex. In contrast, implicit learning occurs automatically, with the nervous system detecting and correcting sensory prediction errors. This form of learning, which is con-trolled by the cerebellum and subcortical structures, enables intuitive adaptation to sensory stimuli without conscious control by the patient.
Implicit learning in particular makes a crucial contribution to the sustainable integration of movement patterns, as it requires fewer cognitive resources and supports everyday movements. In contrast, explicit learning enables precise corrections, particularly in the early rehabilitation phase.
Stroke is one of the most common causes of permanent disability in adulthood worldwide. Typical consequences include asymmetrical
gait patterns, reduced walking speed and an increased risk of falling, which significantly impair the quality of life of those affected. Gait rehabilitation aims to improve these deficits through targeted motor learning strategies. Both explicit (conscious) and implicit (unconscious) learning mechanisms play a key role in this.
Explicit motor learning is consciously controlled through instructions and feedback and requires the active planning and adaptation of movements, which are primarily processed in the prefrontal cortex. In contrast, implicit learning occurs automatically, with the nervous system detecting and correcting sensory prediction errors. This form of learning, which is con-trolled by the cerebellum and subcortical structures, enables intuitive adaptation to sensory stimuli without conscious control by the patient.
Implicit learning in particular makes a crucial contribution to the sustainable integration of movement patterns, as it requires fewer cognitive resources and supports everyday movements. In contrast, explicit learning enables precise corrections, particularly in the early rehabilitation phase.
Through targeted manipulation of asymmetric gait patterns, it was possible to differentiate and analyse motor learning.
Stroke patients exhibit deficits in both conscious learning and automatic adaptation.
The study by Wood and colleagues at the US Department of Physical Therapy at the University of Delaware examines the impairments of explicit and implicit learning processes in individuals with chronic stroke and highlights their relevance for clinical practice under practical laboratory conditions.
Methodology
In the study, the researchers used an experimental paradigm to investigate explicit and implicit motor learning in individuals with chronic stroke and a control group of healthy participants. The sample included individuals with chronic stroke and an age- and gender-matched control group to make differences in motor learning comparable between the groups.
The experiments were carried out on what is known as a split-belt treadmill. This treadmill has two independently movable belts that can run at different speeds under each foot. This arrangement generates asymmetrical gait patterns, which are specifically used to analyse motor adaptation processes. The split-belt treadmill provides a controlled environment in which motor learning can be provoked and measured through manipulation of walking conditions.
After an initial baseline phase, in which both treadmills ran at the same speed and no visual feedback was given, the subjects completed a 3-minute practice phase. This phase served to familiarise the participants with the visual feedback and to ensure that they were able to consciously respond to the feedback by specifically adapting their step lengths. The step length targets were initially set to the individual baseline step lengths of the participants, before being shifted in the short term by ±10 cm to promote conscious control and strategic adaptation of movement.
This was followed by an 8-minute adaptation phase, in which the treadmills ran at different speeds. The faster treadmill moved at the maximum walking speed of the participants, while the slower treadmill was operated at half speed. This difference in speed produced asymmetrical step lengths, which the participants were expected to strategically correct by consciously using the visual feedback. The first 40 steps of this phase were focused on explicit learning, as the participants were instructed to actively correct step errors using the visual information.
After these 40 steps, the visual feedback was removed, and the participants were instructed to continue walking in a gait pattern that was comfortable for them. This change deliberately eliminated controlled adaptations and focused instead on implicit learning. During this phase, adaptations occurred unconsciously through sensory prediction errors, with the nervous system automatically responding to the asymmetric load, to restore a symmetrical gait pattern.
To evaluate the learning mechanisms, a novel adaptation index was developed based on a voluntary correction model approach. This index enabled a precise separation of the contributions of explicit and implicit learning processes:
Explicit learning was quantified by the conscious adaptations of step length during the phase with visual feedback.
Implicit learning was measured based on the after-effects, i.e. the automatic adaptations in step length following the removal of feedback and the return to baseline.
In the study, the researchers used an experimental paradigm to investigate explicit and implicit motor learning in individuals with chronic stroke and a control group of healthy participants. The sample included individuals with chronic stroke and an age- and gender-matched control group to make differences in motor learning comparable between the groups.
The experiments were carried out on what is known as a split-belt treadmill. This treadmill has two independently movable belts that can run at different speeds under each foot. This arrangement generates asymmetrical gait patterns, which are specifically used to analyse motor adaptation processes. The split-belt treadmill provides a controlled environment in which motor learning can be provoked and measured through manipulation of walking conditions.
After an initial baseline phase, in which both treadmills ran at the same speed and no visual feedback was given, the subjects completed a 3-minute practice phase. This phase served to familiarise the participants with the visual feedback and to ensure that they were able to consciously respond to the feedback by specifically adapting their step lengths. The step length targets were initially set to the individual baseline step lengths of the participants, before being shifted in the short term by ±10 cm to promote conscious control and strategic adaptation of movement.
This was followed by an 8-minute adaptation phase, in which the treadmills ran at different speeds. The faster treadmill moved at the maximum walking speed of the participants, while the slower treadmill was operated at half speed. This difference in speed produced asymmetrical step lengths, which the participants were expected to strategically correct by consciously using the visual feedback. The first 40 steps of this phase were focused on explicit learning, as the participants were instructed to actively correct step errors using the visual information.
After these 40 steps, the visual feedback was removed, and the participants were instructed to continue walking in a gait pattern that was comfortable for them. This change deliberately eliminated controlled adaptations and focused instead on implicit learning. During this phase, adaptations occurred unconsciously through sensory prediction errors, with the nervous system automatically responding to the asymmetric load, to restore a symmetrical gait pattern.
To evaluate the learning mechanisms, a novel adaptation index was developed based on a voluntary correction model approach. This index enabled a precise separation of the contributions of explicit and implicit learning processes:
Explicit learning was quantified by the conscious adaptations of step length during the phase with visual feedback.
Implicit learning was measured based on the after-effects, i.e. the automatic adaptations in step length following the removal of feedback and the return to baseline.
Individually tailored learning strategies are the key to effective recovery of motor skills after a stroke.
This methodology allowed a detailed differentiation of the two learning mechanisms and their effectiveness in the participants, providing new insights into the motor adaptation processes following a stroke.
Following completion of the adaptation phase, there was an 8-minute de-adaptation phase, in which both treadmills were operated at the same speed as in the baseline phase. In this phase, the participants were instructed to “walk comfortably”. The focus here was on measuring the after-effects of the asymmetrical load that was generated during the adaptation phase.
The de-adaptation phase allowed for a more precise evaluation of implicit adaptation by observing how the nervous system responded to the changed treadmill conditions. The extent to which subjects maintained asymmetric stepping patterns, despite the external asymmetry being removed, was used as a measure of the strength of the implicit after-effect. This reflects the total magnitude of implicit adaptation that had taken place during the adaptation phase.
The data from the de-adaptation phase were also integrated into the adaptation index to obtain a complete picture of the implicit learning processes. The phase thus served as a crucial component for the separation and quantification of the automatic adaptation mechanisms that operate independently of conscious control processes.
Following completion of the adaptation phase, there was an 8-minute de-adaptation phase, in which both treadmills were operated at the same speed as in the baseline phase. In this phase, the participants were instructed to “walk comfortably”. The focus here was on measuring the after-effects of the asymmetrical load that was generated during the adaptation phase.
The de-adaptation phase allowed for a more precise evaluation of implicit adaptation by observing how the nervous system responded to the changed treadmill conditions. The extent to which subjects maintained asymmetric stepping patterns, despite the external asymmetry being removed, was used as a measure of the strength of the implicit after-effect. This reflects the total magnitude of implicit adaptation that had taken place during the adaptation phase.
The data from the de-adaptation phase were also integrated into the adaptation index to obtain a complete picture of the implicit learning processes. The phase thus served as a crucial component for the separation and quantification of the automatic adaptation mechanisms that operate independently of conscious control processes.
Results
The results show that individuals with stroke exhibited a limited capacity for explicit learning compared to the control group. This impairment was evident in the smaller difference in the adaptation index between phases with and without visual feedback during the adaptation phase (mean [95% HDI]: 0.09 [-0.05 0.25], probability of a difference = 88.2%). In addition, the adaptation index during the feedback phase was significantly lower for the stroke group than for the control group (mean difference = 0.23 [0.11 0.34], probability of a difference = 100.0%). These results demonstrate that individuals with stroke use visual feedback less effectively and have a reduced ability for conscious adaptation to errors.
The results show that individuals with stroke exhibited a limited capacity for explicit learning compared to the control group. This impairment was evident in the smaller difference in the adaptation index between phases with and without visual feedback during the adaptation phase (mean [95% HDI]: 0.09 [-0.05 0.25], probability of a difference = 88.2%). In addition, the adaptation index during the feedback phase was significantly lower for the stroke group than for the control group (mean difference = 0.23 [0.11 0.34], probability of a difference = 100.0%). These results demonstrate that individuals with stroke use visual feedback less effectively and have a reduced ability for conscious adaptation to errors.
Combined approaches that specifically address cognitive and motor deficits could significantly improve rehabilitation outcomes in the future.
Implicit adaptation was also impaired in the stroke group. This was measured by smaller implicit after-effects, which indicate lower adaptations by the nervous system (group difference = 0.10 [−0.02 0.21], probability of a difference = 94.3%). At the end of the adaptation phase, the stroke group also showed significantly lower values than the control group (group difference = 0.17 [0.07 0.28], probability of a difference = 99.9%). These results suggest that both explicit learning and implicit adaptation are impaired following a stroke.
For a more detailed examination of learning processes, a model was used that separately analysed conscious (explicit) and unconscious (implicit) learning. The results showed that adaptability was overall lower in the stroke group compared to the control group, indicating greater differences and variabilities within the stroke group.
Conscious learning was measured through the adaptation index, which quantified how quickly and effectively participants could respond to visual feedback. Here the stroke group performed significantly worse than the control group. This confirms that conscious learning is limited in post-stroke individuals.
The analysis of the unconscious adaptations, in contrast, showed only small differences between the two groups in most of the areas investigated. However, a clear difference was noted in the ability to retain previously learned information over the long term. This ability was noticeably less pronounced in the stroke group, which offers a possible explanation for the overall slower automatic adaptation processes following a stroke.
For a more detailed examination of learning processes, a model was used that separately analysed conscious (explicit) and unconscious (implicit) learning. The results showed that adaptability was overall lower in the stroke group compared to the control group, indicating greater differences and variabilities within the stroke group.
Conscious learning was measured through the adaptation index, which quantified how quickly and effectively participants could respond to visual feedback. Here the stroke group performed significantly worse than the control group. This confirms that conscious learning is limited in post-stroke individuals.
The analysis of the unconscious adaptations, in contrast, showed only small differences between the two groups in most of the areas investigated. However, a clear difference was noted in the ability to retain previously learned information over the long term. This ability was noticeably less pronounced in the stroke group, which offers a possible explanation for the overall slower automatic adaptation processes following a stroke.
Discussion
The results show that individuals with stroke have difficulties using conscious learning effectively. This manifested itself in fewer behavioural changes after the visual feedback was removed and in slower adaptation processes, which were confirmed by model analyses. These differences are not solely attributable to motor limitations but indicate specific difficulties in conscious error correction.
Unconscious adaptability was also impaired in participants with stroke. Their adaptation of movements was slower and less complete compared to the control group with healthy subjects. The model analyses suggest that these delays might be attributable to a limited capacity for long-term storage of new movement patterns. This could be explained by damage in the brain regions responsible for movement and learning.
The results illustrate that both conscious and unconscious learning processes require different approaches in rehabilitation. Conscious learning could be promoted through clear instructions and targeted feedback, while unconscious adaptations should be supported through repeated practice and long-term stabilisation.
The results show that individuals with stroke have difficulties using conscious learning effectively. This manifested itself in fewer behavioural changes after the visual feedback was removed and in slower adaptation processes, which were confirmed by model analyses. These differences are not solely attributable to motor limitations but indicate specific difficulties in conscious error correction.
Unconscious adaptability was also impaired in participants with stroke. Their adaptation of movements was slower and less complete compared to the control group with healthy subjects. The model analyses suggest that these delays might be attributable to a limited capacity for long-term storage of new movement patterns. This could be explained by damage in the brain regions responsible for movement and learning.
The results illustrate that both conscious and unconscious learning processes require different approaches in rehabilitation. Conscious learning could be promoted through clear instructions and targeted feedback, while unconscious adaptations should be supported through repeated practice and long-term stabilisation.
Conclusion
Researchers are coming to the realisation that people after a stroke show limitations in various types of motor learning, which cannot be attributed solely to motor deficits. These findings are important in order to better adapt rehabilitation programmes to the specific needs of those affected in the future and thus promote the restoration of motor skills through the selection of suitable learning strategies. Future research must now
examine how training methods can be structured to specifically strengthen conscious and unconscious learning and secure long-term progress.
Researchers are coming to the realisation that people after a stroke show limitations in various types of motor learning, which cannot be attributed solely to motor deficits. These findings are important in order to better adapt rehabilitation programmes to the specific needs of those affected in the future and thus promote the restoration of motor skills through the selection of suitable learning strategies. Future research must now
examine how training methods can be structured to specifically strengthen conscious and unconscious learning and secure long-term progress.
Implications for practice
The results show that rehabilitation after a stroke should target both conscious learning processes (through clear instructions and specific feedback) as well as unconscious adaptation mechanisms (through repetition-based exercises and sensory challenges). Combined approaches that promote both mechanisms simultaneously or sequentially could shape the restoration of motor skills more effectively. Individually tailored training programmes that take into account the spe-cific deficits of each patient offer the potential to optimise rehabilitation outcomes and ensure long-term improvement of everyday competencies. For this purpose, it is important, within the scope of motor therapies, not only to consider the motor deficits, but also to criti-cally reflect on the cognitive resources when selecting suitable learning strategies.
The results show that rehabilitation after a stroke should target both conscious learning processes (through clear instructions and specific feedback) as well as unconscious adaptation mechanisms (through repetition-based exercises and sensory challenges). Combined approaches that promote both mechanisms simultaneously or sequentially could shape the restoration of motor skills more effectively. Individually tailored training programmes that take into account the spe-cific deficits of each patient offer the potential to optimise rehabilitation outcomes and ensure long-term improvement of everyday competencies. For this purpose, it is important, within the scope of motor therapies, not only to consider the motor deficits, but also to criti-cally reflect on the cognitive resources when selecting suitable learning strategies.
Implikationen für die Praxis
Die Ergebnisse zeigen, dass Rehabilitation nach einem Schlaganfall sowohl auf bewusste Lernprozesse (durch klare Anweisungen und gezieltes Feedback) als auch auf unbewusste Anpassungsmechanismen (durch wiederholungsbasierte Übungen und sensorische Herausforderungen) abzielen sollte. Kombinierte Ansätze, die beide Mechanismen gleichzeitig oder sequenziell fördern, könnten die Wiederher-stellung motorischer Fähigkeiten effektiver gestalten. Individuell angepasste Trainingsprogramme, die die spezifischen Defizite jedes Patienten berücksichtigen, bieten das Potenzial, die Rehabilitationsergebnisse zu optimieren und die Alltagskompetenzen nachhaltig zu verbessern. Hierfür ist es wichtig, im Rahmen der motorischen Therapien nicht nur die motorischen Defizite zu berücksichtigen, sondern auch die kognitiven Ressourcen bei der Wahl geeigneter Lernstrategien kritisch zu reflektieren.
Die Ergebnisse zeigen, dass Rehabilitation nach einem Schlaganfall sowohl auf bewusste Lernprozesse (durch klare Anweisungen und gezieltes Feedback) als auch auf unbewusste Anpassungsmechanismen (durch wiederholungsbasierte Übungen und sensorische Herausforderungen) abzielen sollte. Kombinierte Ansätze, die beide Mechanismen gleichzeitig oder sequenziell fördern, könnten die Wiederher-stellung motorischer Fähigkeiten effektiver gestalten. Individuell angepasste Trainingsprogramme, die die spezifischen Defizite jedes Patienten berücksichtigen, bieten das Potenzial, die Rehabilitationsergebnisse zu optimieren und die Alltagskompetenzen nachhaltig zu verbessern. Hierfür ist es wichtig, im Rahmen der motorischen Therapien nicht nur die motorischen Defizite zu berücksichtigen, sondern auch die kognitiven Ressourcen bei der Wahl geeigneter Lernstrategien kritisch zu reflektieren.
Specific limitations in conscious error correction and the storage of new movement patterns require targeted rehabilitation strategies.
Comments
This is a theoretically sound yet highly practical study whose insight goes beyond the mere fact that stroke patients learn more slowly something that experienced practitioners know from experience. The new core finding lies in the differentiated analysis of the specific mechanisms underlying the learning difficulties of stroke patients, and the identification of targeted approaches to address these effectively.
What are the most important findings?
Explicit learning is impaired independently of motor limitations. The study shows that deficits in conscious learning (explicit learning) are not solely conditioned by motor impairments. Even patients with better motor control struggled to consciously correct errors and effectively utilise visual feedback. This suggests that cognitive processes such as working memory or the processing of feedback play a key role.
Slower adaptation through specific impairment of implicit learning processes. The study also shows that not all unconscious (implicit) adjustment is impaired, but specifically the ability to quickly react to errors and store them. This was demonstrated for the first time in a differentiated manner through the modelling of the adaptive learning process in a locomotion task.
Explicit learning does not inhibit implicit adaptation: Another interesting result is that conscious learning (explicit feedback) does not impair unconscious adaptation. This is important as it is often assumed that these two processes could compete with each other. The results show that both processes can run independently of each other, which means that rehabilitation programmes can certainly combine these approaches without risking a deterioration in outcome.
New methodological approaches to the separation of learning processes: The study is the first to analyse both explicit and implicit learning in a locomotion task in stroke patients using a combination of behavioural tests and computer-assisted modelling. The voluntary correction model used offers a new tool for the future to precisely quantify and differentiate learning processes.
Why are these findings important?
The results emphasise the importance of giving greater consideration to cognitive processes such as feedback processing and memory per-formance in motor rehabilitation. In the context of motor therapy, attention is usually only given to motor functions.
The study goes beyond the general understanding that stroke patients learn more
slowly, and provides new mechanistic explanations that have previously received little attention in rehabilitation. The modelling of learning provides an important basis for the future in order to develop more effective personalised training methods
This is a theoretically sound yet highly practical study whose insight goes beyond the mere fact that stroke patients learn more slowly something that experienced practitioners know from experience. The new core finding lies in the differentiated analysis of the specific mechanisms underlying the learning difficulties of stroke patients, and the identification of targeted approaches to address these effectively.
What are the most important findings?
Explicit learning is impaired independently of motor limitations. The study shows that deficits in conscious learning (explicit learning) are not solely conditioned by motor impairments. Even patients with better motor control struggled to consciously correct errors and effectively utilise visual feedback. This suggests that cognitive processes such as working memory or the processing of feedback play a key role.
Slower adaptation through specific impairment of implicit learning processes. The study also shows that not all unconscious (implicit) adjustment is impaired, but specifically the ability to quickly react to errors and store them. This was demonstrated for the first time in a differentiated manner through the modelling of the adaptive learning process in a locomotion task.
Explicit learning does not inhibit implicit adaptation: Another interesting result is that conscious learning (explicit feedback) does not impair unconscious adaptation. This is important as it is often assumed that these two processes could compete with each other. The results show that both processes can run independently of each other, which means that rehabilitation programmes can certainly combine these approaches without risking a deterioration in outcome.
New methodological approaches to the separation of learning processes: The study is the first to analyse both explicit and implicit learning in a locomotion task in stroke patients using a combination of behavioural tests and computer-assisted modelling. The voluntary correction model used offers a new tool for the future to precisely quantify and differentiate learning processes.
Why are these findings important?
The results emphasise the importance of giving greater consideration to cognitive processes such as feedback processing and memory per-formance in motor rehabilitation. In the context of motor therapy, attention is usually only given to motor functions.
The study goes beyond the general understanding that stroke patients learn more
slowly, and provides new mechanistic explanations that have previously received little attention in rehabilitation. The modelling of learning provides an important basis for the future in order to develop more effective personalised training methods
Ambulante Rehabilitation
Fachkreise
Gait
lyra
Produkte
Science
Stationäre Rehabilitation
THERAPY 2025-I
THERAPY Magazine
Jakob Tiebel
Business Owner, N+ Digital Health Agency
Jakob Tiebel studied applied psychology with a focus on health economics. He has clinical expertise from his previous therapeutic work in neurorehabilitation. He conducts research and publishes on the theory-practice transfer in neurorehabilitation and is the owner of Native.Health, an agency for digital health marketing.
References:
- Wood JM, Thompson E, Wright H, Festa L, Morton SM, Reisman DS, Kim HE. Explicit and implicit locomotor learning in individuals with chronic hemiparetic stroke. J Neurophysiol. 2024 Oct 1;132(4):1172-1182. doi: 10.1152/jn.00156.2024. Epub 2024 Sep 4. PMID: 39230337; PMCID: PMC11495209.