Failure of motor learning for large initial errors

Visual feedbacks influence short-term learning of torque versus motion profile with robotic guidance among young adults

Robotic assistance can improve the learning of complex motor skills. However, the assistance designed and used up to now mainly guides motor commands for trajectory learning, not dynamics learning. The present study explored how a complex motor skill involving the right arm can be learned without suppressing task dynamics, by means of an innovative device with robotic guidance that allows a torque versus motion profile to be learned with admittance control. In addition, we assessed how concurrent visual feedback on this profile can enhance learning without creating dependency, by means of a fading procedure (i.e., feedback reduction across trials). On Day 1, a Control group performed an acquisition session (6 blocks) featuring concurrent visual feedback, while a Fading group performed the session with a gradual reduction in feedback (from 100% to 0% over the 6 blocks). On Day 2, both groups performed a block first without feedback (i.e., Transfer test), then with feedback (i.e., Retention test). Results revealed that on Day 1, movement rehearsal induced a significant improvement in spatiotemporal parameters for the Control group, compared with the Fading group. On Day 2, the opposite was found when this visual feedback was removed, as the Fading group performed significantly better than the Control group on the Transfer test. Vision allows a relationship to be established between the required torque and the motion profile. Its suppression then forces the processing of more intrinsic information, leading to the development of a stable internal representation of the task.

Theoretical limits on the speed of learning inverse models explain the rate of adaptation in arm reaching tasks

An essential aspect of human motor learning is the formation of inverse models, which map desired actions to motor commands. Inverse models can be learned by adjusting parameters in neural circuits to minimize errors in the performance of motor tasks through gradient descent. However, the theory of gradient descent establishes limits on the learning speed. Specifically, the eigenvalues of the Hessian of the error surface around a minimum determine the maximum speed of learning in a task. Here, we use this theoretical framework to analyze the speed of learning in different inverse model learning architectures in a set of isometric arm-reaching tasks. We show theoretically that, in these tasks, the error surface and, thus the speed of learning, are determined by the shapes of the force manipulability ellipsoid of the arm and the distribution of targets in the task. In particular, rounder manipulability ellipsoids generate a rounder error surface, allowing for faster learning of the inverse model. Rounder target distributions have a similar effect. We tested these predictions experimentally in a quasi-isometric reaching task with a visuomotor transformation. The experimental results were consistent with our theoretical predictions. Furthermore, our analysis accounts for the speed of learning in previous experiments with incompatible and compatible virtual surgery tasks, and with visuomotor rotation tasks with different numbers of targets. By identifying aspects of a task that influence the speed of learning, our results provide theoretical principles for the design of motor tasks that allow for faster learning.

Let the force guide you: a performance-based adaptive algorithm for postural training using haptic feedback

Motor learning is an essential component of human behavior. Many different factors can influence the process of motor learning, such as the amount of practice and type of feedback. Changes in task difficulty during training can also considerably impact motor learning. Typical motor learning studies include a sequential variation of task difficulty, i.e., easy to challenging, irrespective of user performance. However, many studies have reported the importance of performance-based task difficulty variation for effective motor learning and skill transfer. A performance-based adaptive algorithm for task difficulty variation based on the challenge-point framework is proposed in this study. The algorithm is described for postural adaptation during simultaneous upper-limb training. Ten healthy participants (28 ± 2.44 years) were recruited to validate the algorithm. Participants adapted to a postural target of 20° in the anterior direction from the initial upright posture while performing a unimanual reaching task using a robotic device. Results suggest a significant decrease in postural error after training. The algorithm successfully adapted the task difficulty based on the performance of the user. The proposed algorithm could be modified for different motor skills and can be further evaluated for different applications in order to maximize the potential benefits of rehabilitation sessions.

Task space exploration improves adaptation after incompatible virtual surgeries

Humans have a remarkable capacity to learn new motor skills, a process that requires novel muscle activity patterns. Muscle synergies may simplify the generation of muscle patterns through the selection of a small number of synergy combinations. Learning new motor skills may then be achieved by acquiring novel muscle synergies. In a previous study, we used myoelectric control to construct virtual surgeries that altered the mapping from muscle activity to cursor movements. After compatible virtual surgeries, which could be compensated by recombining subject-specific muscle synergies, participants adapted quickly. In contrast, after incompatible virtual surgeries, which could not be compensated by recombining existing synergies, participants explored new muscle patterns, but failed to adapt. Here, we tested whether task space exploration can promote learning of novel muscle synergies, required to overcome an incompatible surgery. Participants performed the same reaching task as in our previous study, but with more time to complete each trial, thus allowing for exploration. We found an improvement in trial success after incompatible virtual surgeries. Remarkably, improvements in movement direction accuracy after incompatible surgeries occurred faster for corrective movements than for the initial movement, suggesting that learning of new synergies is more effective when used for feedback control. Moreover, reaction time was significantly higher after incompatible than after compatible virtual surgeries, suggesting an increased use of an explicit adaptive strategy to overcome incompatible surgeries. Taken together, these results indicate that exploration is important for skill learning and suggest that human participants, with sufficient time, can learn new muscle synergies.

Brain Plasticity Mechanisms Underlying Motor Control Reorganization: Pilot Longitudinal Study on Post-Stroke Subjects

Functional Electrical Stimulation (FES) has demonstrated to improve walking ability and to induce the carryover effect, long-lasting persisting improvement. Functional magnetic resonance imaging has been used to investigate effective connectivity differences and longitudinal changes in a group of chronic stroke patients that attended a FES-based rehabilitation program for foot-drop correction, distinguishing between carryover effect responders and non-responders, and in comparison with a healthy control group. Bayesian hierarchical procedures were employed, involving nonlinear models at within-subject level-dynamic causal models-and linear models at between-subjects level. Selected regions of interest were primary sensorimotor cortices (M1, S1), supplementary motor area (SMA), and angular gyrus. Our results suggest the following: (i) The ability to correctly plan the movement and integrate proprioception information might be the features to update the motor control loop, towards the carryover effect, as indicated by the reduced sensitivity to proprioception input to S1 of FES non-responders; (ii) FES-related neural plasticity supports the active inference account for motor control, as indicated by the modulation of SMA and M1 connections to S1 area; (iii) SMA has a dual role of higher order motor processing unit responsible for complex movements, and a superintendence role in suppressing standard motor plans as external conditions changes.

Modified sensory feedback enhances the sense of agency during continuous body movements in virtual reality

The sense of agency refers to the feeling of control over one's own actions, and through them, the external events. This study examined the effect of modified visual feedback on the sense of agency over one's body movements using virtual reality in healthy individuals whose motor control was disturbed. Participants moved a virtual object using their right hand to trace a trajectory (Experiment 1) or a leading target (Experiment 2). Their motor control was disturbed by a delay in visual feedback (Experiment 1) or a 1-kg weight attached to their wrist (Experiment 2). In the offset conditions, the virtual object was presented at the median point between the desired position and the participants' actual hand position. In both experiments, participants reported improved sense of agency in the offset condition compared to the aligned condition where the visual feedback reflected their actual body movements, despite their motion being less precise in the offset condition. The results show that sense of agency can be enhanced by modifying feedback to motor tasks according to the goal of the task, even when visual feedback is discrepant from the actual body movements. The present study sheds light on the possibility of artificially enhancing body agency to improve voluntary motor control.

Automated functional electrical stimulation training system for upper-limb function recovery in poststroke patients

BACKGROUND

This paper describes the design and test of an automated functional electrical stimulation (FES) system for poststroke rehabilitation training. The aim of automated FES is to synchronize electrically induced movements to assist residual movements of patients.

METHODS

In the design of the FES system, an accelerometry module detected movement initiation and movement performed by post-stroke patients. The desired movement was displayed in visual game module. Synergy-based FES patterns were formulated using a normal pattern of muscle synergies from a healthy subject. Experiment 1 evaluated how different levels of trigger threshold or timing affected the variability of compound movements for forward reaching (FR) and lateral reaching (LR). Experiment 2 explored the effect of FES duration on compound movements.

RESULTS

Synchronizing FES-assisted movements with residual voluntary movements produced more consistent compound movements. Matching the duration of synergy-based FES to that of patients could assist slower movements of patients with reduced RMS errors.

CONCLUSIONS

Evidence indicated that synchronization and matching duration with residual voluntary movements of patients could improve the consistency of FES assisted movements. Automated FES training can reduce the burden of therapists to monitor the training process, which may encourage patients to complete the training.

The dynamics of motor learning through the formation of internal models

A medical student learning to perform a laparoscopic procedure or a recently paralyzed user of a powered wheelchair must learn to operate machinery via interfaces that translate their actions into commands for an external device. Since the user's actions are selected from a number of alternatives that would result in the same effect in the control space of the external device, learning to use such interfaces involves dealing with redundancy. Subjects need to learn an externally chosen many-to-one map that transforms their actions into device commands. Mathematically, we describe this type of learning as a deterministic dynamical process, whose state is the evolving forward and inverse internal models of the interface. The forward model predicts the outcomes of actions, while the inverse model generates actions designed to attain desired outcomes. Both the mathematical analysis of the proposed model of learning dynamics and the learning performance observed in a group of subjects demonstrate a first-order exponential convergence of the learning process toward a particular state that depends only on the initial state of the inverse and forward models and on the sequence of targets supplied to the users. Noise is not only present but necessary for the convergence of learning through the minimization of the difference between actual and predicted outcomes.

EMG-based vibro-tactile biofeedback training: effective learning accelerator for children and adolescents with dystonia? A pilot crossover trial

Background

This study is aimed at better understanding the role of a wearable and silent ElectroMyoGraphy-based biofeedback on motor learning in children and adolescents with primary and secondary dystonia.

Methods

A crossover study with a wash-out period of at least 1 week was designed; the device provides the patient with a vibration proportional to the activation of an impaired target muscle. The protocol consisted of two 5-day blocks during which subjects were trained and tested on a figure-8 writing task: their performances (at different levels of difficulty) were evaluated in terms of both kinematics and muscular activations on day 1 and day 5, while the other 3 days were purely used as training sessions. The training was performed with and without using the biofeedback device: the week of use was randomized. Data were collected on 14 subjects with primary and secondary (acquired) dystonia (age: 6–19 years).

Results

Results comparing kinematic-based and EMG-based outcome measures pre- and post-training showed learning due to practice for both subjects with primary and secondary dystonia. On top of said learning, an improvement in terms of inter-joint coordination and muscular pattern functionality was recorded only for secondary dystonia subjects, when trained with the aid of the EMG-based biofeedback device.

Conclusions

Our results support the hypothesis that children and adolescents with primary dystonia in which there is intact sensory processing do not benefit from feedback augmentation, whereas children with secondary dystonia, in which sensory deficits are often present, exhibit a higher learning capacity when augmented movement-related sensory information is provided. This study represents a fundamental investigation to address the scarcity of noninvasive therapeutic interventions for young subjects with dystonia.

Does severity of motor impairment affect reactive adaptation and fall-risk in chronic stroke survivors?

Background

A single-session of slip-perturbation training has shown to induce long-term fall risk reduction in older adults. Considering the spectrum of motor impairments and deficits in reactive balance after a cortical stroke, we aimed to determine if chronic stroke survivors could acquire and retain reactive adaptations to large slip-like perturbations and if these adaptations were dependent on severity of motor impairment.

Methods

Twenty-six chronic stroke participants were categorized into high and low-functioning groups based on their Chedoke-McMaster-Assessment scores. All participants received a pre-training, slip-like stance perturbation at level-III (highest intensity/acceleration) followed by 11 perturbations at a lower intensity (level-II). If in early phase, participants experienced > 3/5 falls, they were trained at a still lower intensity (level-I). Post-training, immediate scaling and short-term retention at 3 weeks post-training was examined. Perturbation outcome and post-slip center-of-mass (COM) stability was analyzed.

Results

On the pre-training trial, 60% of high and 100% of low-functioning participants fell. High-functioning group tolerated and adapted at training-intensity level-II but low-functioning group were trained at level-I (all had > 3 falls on level-II). At respective training intensities, both groups significantly lowered fall incidence from 1st through 11th trials, with improved post-slip stability and anterior shift in COM position, resulting from increased compensatory step length. Both groups demonstrated immediate scaling and short-term retention of the acquired stability control.

Conclusion

Chronic stroke survivors are able to acquire and retain adaptive reactive balance skills to reduce fall risk. Although similar adaptation was demonstrated by both groups, the low-functioning group might require greater dosage with gradual increment in training intensity.

Basic and Translational Neuroscience of Childhood-Onset Dystonia: A Control-Theory Perspective

Dystonia is a collection of symptoms with involuntary muscle activation causing hypertonia, hyperkinetic movements, and overflow. In children, dystonia can have numerous etiologies with varying neuroanatomic distribution. The semiology of dystonia can be explained by gain-of-function failure of a feedback controller that is responsible for stabilizing posture and movement. Because postural control is maintained by a widely distributed network, many different anatomic regions may be responsible for symptoms of dystonia, although all features of dystonia can be explained by uncontrolled activation or hypersensitivity of motor cortical regions that can cause increased reflex gain, inserted postures, or sensitivity to irrelevant sensory variables. Effective treatment of dystonia in children requires an understanding of the relationship between etiology, anatomy, and the specific mechanism of failure of postural stabilization.

Mapping upper-limb motor performance after stroke - a novel method with utility for individualized motor training

Background

Chronic upper limb motor impairment is a common outcome of stroke. Therapeutic training can reduce motor impairment. Recently, a growing interest in evaluating motor training provided by robotic assistive devices has emerged. Robot-assisted therapy is attractive because it provides a means of increasing practice intensity without increasing the workload of physical therapists. However, movements practised through robotic assistive devices are commonly pre-defined and fixed across individuals. More optimal training may result from individualizing the selection of the trained movements based on the individual’s impairment profile. This requires quantitative assessment of the degree of the motor impairment prior to training, in relevant movement tasks. However, standard clinical measures for profiling motor impairment after stroke are often subjective and lack precision. We have developed a novel robot-mediated method for systematic and fine-grained mapping (or profiling) of individual performance across a wide range of planar arm reaching movements. Here we describe and demonstrate this mapping method and its utilization for individualized training. We also present a novel principle for the individualized selection of training movements based on the performance maps.

Methods and Results

To demonstrate the utility of our method we present examples of 2D performance maps produced from the kinetic and kinematics data of two individuals with stroke-related upper limb hemiparesis. The maps outline distinct regions of high motor impairment. The procedure of map-based selection of training movements and the change in motor performance following training is demonstrated for one participant.

Conclusions

The performance mapping method is feasible to produce (online or offline). The 2D maps are easy to interpret and to be utilized for selecting individual performance-based training. Different performance maps can be easily compared within and between individuals, which potentially has diagnostic utility.

Electronic supplementary material

The online version of this article (10.1186/s12984-017-0335-x) contains supplementary material, which is available to authorized users.

Multiscale modeling in the clinic: diseases of the brain and nervous system

Computational neuroscience is a field that traces its origins to the efforts of Hodgkin and Huxley, who pioneered quantitative analysis of electrical activity in the nervous system. While also continuing as an independent field, computational neuroscience has combined with computational systems biology, and neural multiscale modeling arose as one offshoot. This consolidation has added electrical, graphical, dynamical system, learning theory, artificial intelligence and neural network viewpoints with the microscale of cellular biology (neuronal and glial), mesoscales of vascular, immunological and neuronal networks, on up to macroscales of cognition and behavior. The complexity of linkages that produces pathophysiology in neurological, neurosurgical and psychiatric disease will require multiscale modeling to provide understanding that exceeds what is possible with statistical analysis or highly simplified models: how to bring together pharmacotherapeutics with neurostimulation, how to personalize therapies, how to combine novel therapies with neurorehabilitation, how to interlace periodic diagnostic updates with frequent reevaluation of therapy, how to understand a physical disease that manifests as a disease of the mind. Multiscale modeling will also help to extend the usefulness of animal models of human diseases in neuroscience, where the disconnects between clinical and animal phenomenology are particularly pronounced. Here we cover areas of particular interest for clinical application of these new modeling neurotechnologies, including epilepsy, traumatic brain injury, ischemic disease, neurorehabilitation, drug addiction, schizophrenia and neurostimulation.

The integration of probabilistic information during sensorimotor estimation is unimpaired in children with Cerebral Palsy

Background

It is important to understand the motor deficits of children with Cerebral Palsy (CP). Our understanding of this motor disorder can be enriched by computational models of motor control. One crucial stage in generating movement involves combining uncertain information from different sources, and deficits in this process could contribute to reduced motor function in children with CP. Healthy adults can integrate previously-learned information (prior) with incoming sensory information (likelihood) in a close-to-optimal way when estimating object location, consistent with the use of Bayesian statistics. However, there are few studies investigating how children with CP perform sensorimotor integration. We compare sensorimotor estimation in children with CP and age-matched controls using a model-based analysis to understand the process.

Methods and findings

We examined Bayesian sensorimotor integration in children with CP, aged between 5 and 12 years old, with Gross Motor Function Classification System (GMFCS) levels 1–3 and compared their estimation behavior with age-matched typically-developing (TD) children. We used a simple sensorimotor estimation task which requires participants to combine probabilistic information from different sources: a likelihood distribution (current sensory information) with a prior distribution (learned target information). In order to examine sensorimotor integration, we quantified how participants weighed statistical information from the two sources (prior and likelihood) and compared this to the statistical optimal weighting. We found that the weighing of statistical information in children with CP was as statistically efficient as that of TD children.

Conclusions

We conclude that Bayesian sensorimotor integration is not impaired in children with CP and therefore, does not contribute to their motor deficits. Future research has the potential to enrich our understanding of motor disorders by investigating the stages of motor processing set out by computational models. Therapeutic interventions should exploit the ability of children with CP to use statistical information.

Performance analysis of risk-aware control in upper limb of patients with post-stroke hemiparesis

Evidence suggests that humans tune their motor behavior in response to environmental uncertainty. Such risk-aware sensorimotor performance is hypothetically affected in stroke due to impaired sensorimotor control, but its existence and the detailed pattern of impairment remain unclear - particularly due to lack of approaches for measurement of dynamic performance in stroke. Here we present the design and implementation of a platform for testing the upper-limb control in presence of motor uncertainty. The platform monitors how a patient with stroke, exclusively with the affected arm, navigates a virtual car by adjusting the hand position. We focus on how the control behavior changes when the car is perturbed with progressively larger Gaussian noise. Using this platform, we successfully collected data from 4 healthy subjects and 4 patients with post-stroke hemiparesis. Initial evidence suggests that patients with stroke failed to tune their performance with increasing level of Gaussian perturbation. The results imply that the risk-aware behavior potentially provides a new modality of biofeedback useful in therapeutic interventions.

Matching Task Difficulty to Patient Ability During Task Practice Improves Upper Extremity Motor Skill After Stroke: A Proof-of-Concept Study

OBJECTIVE

To test the feasibility of the Fugl-Meyer Assessment of the Upper Extremity "keyform," derived from Rasch analysis, as a method for systematically planning and progressing rehabilitation.

DESIGN

Feasibility study, single group design.

SETTING

University rehabilitation research laboratory.

PARTICIPANTS

Participants (N=10; mean age, 59.70±9.96y; 24.1±30.54mo poststroke) with ischemic or hemorrhagic stroke >3 months prior, voluntarily shoulder flexion ≥30°, and simultaneous elbow extension ≥20°.

INTERVENTIONS

The keyform method defined initial rehabilitation targets (goals) and progressed the rehabilitation program after every third session. Targets were repetitively practiced within the context of client-selected functional tasks not in isolation.

MAIN OUTCOME MEASURES

Feasibility was defined by subject's pain or fatigue, upper extremity motor function (Wolf Motor Function Test), and movement patterns (kinematics). Assessments were administered pre- and posttreatment and compared using paired t tests. Task-difficulty and patient-ability measures were calculated using Rasch analysis and compared using paired t tests (P<.05).

RESULTS

Ten participants completed 9 sessions, 200 movement repetitions per session in <2 hours without pain or fatigue. Participants gained upper extremity motor function (Wolf Motor Function Test: pretreatment, 22.23±24.26s; posttreatment, 15.46±22.12s; P=.01), improved shoulder-elbow coordination (index of curvature: pretreatment, 1.30±0.15; posttreatment, 1.21±0.11; P=.01), and exhibited reduced trunk compensatory movement (trunk displacement: pretreatment, 133.97±74.15mm; posttreatment, 108.08±64.73mm; P=.02). Task-difficulty and patient-ability measures were not statistically different throughout the program (person-ability measures of 1.01±0.05, 1.64±0.45, and 2.22±0.65 logits and item-difficulty measures of 0.93±0.37, 1.70±0.20, and 2.06±0.24 logits at the 3 testing time points, respectively; P>.05).

CONCLUSIONS

The Fugl-Meyer Assessment of the Upper Extremity keyform is a feasible method to ensure that the difficulty of tasks practiced were well matched to initial and evolving levels of upper extremity motor ability.

Sinusoidal error perturbation reveals multiple coordinate systems for sensorymotor adaptation

A coordinate system is composed of an encoding, defining the dimensions of the space, and an origin. We examine the coordinate encoding used to update motor plans during sensory-motor adaptation to center-out reaches. Adaptation is induced using a novel paradigm in which feedback of reach endpoints is perturbed following a sinewave pattern over trials; the perturbed dimensions of the feedback were the axes of a Cartesian coordinate system in one session and a polar coordinate system in another session. For center-out reaches to randomly chosen target locations, reach errors observed at one target will require different corrections at other targets within Cartesian- and polar-coded systems. The sinewave adaptation technique allowed us to simultaneously adapt both dimensions of each coordinate system (x-y, or reach gain and angle), and identify the contributions of each perturbed dimension by adapting each at a distinct temporal frequency. The efficiency of this technique further allowed us to employ perturbations that were a fraction the size normally used, which avoids confounding automatic adaptive processes with deliberate adjustments made in response to obvious experimental manipulations. Subjects independently corrected errors in each coordinate in both sessions, suggesting that the nervous system encodes both a Cartesian- and polar-coordinate-based internal representation for motor adaptation. The gains and phase lags of the adaptive responses are not readily explained by current theories of sensory-motor adaptation.

Short-Duration and Intensive Training Improves Long-Term Reaching Performance in Individuals With Chronic Stroke

Previous studies have shown that multiple sessions of reach training lead to long-term improvements in movement time and smoothness in individuals post-stroke. Yet such long-term training regimens are often difficult to implement in actual clinical settings. In this study, we evaluated the long-term and generalization effects of short-duration and intensive reach training in 16 individuals with chronic stroke and mild to moderate impairments. Participants performed 2 sessions of unassisted intensive reach training, with 600 movements per session, and with display of performance-based feedback after each movement. The participants' trunks were restrained with a belt to avoid compensatory movements. Training resulted in significant and durable (1 month) improvements in movement time (20.4% on average) and movement smoothness (22.7% on average). The largest improvements occurred in individuals with the largest initial motor impairments. In addition, training induced generalization to nontrained targets, which persisted in 1-day and in 1-month retention tests. Finally, there was a significant improvement in the Box and Block test from baseline to 1-month retention test (23% on average). Thus, short-duration and intensive reach training can lead to generalized and durable benefits in individuals with chronic stroke and mild to moderate impairments.

Current and emerging strategies for treatment of childhood dystonia

Childhood dystonia is a movement disorder characterized by involuntary sustained or intermittent muscle contractions causing twisting and repetitive movements, abnormal postures, or both (Sanger et al, 2003). Dystonia is a devastating neurological condition that prevents the acquisition of normal motor skills during critical periods of development in children. Moreover, it is particularly debilitating in children when dystonia affects the upper extremities such that learning and consolidation of common daily motor actions are impeded. Thus, the treatment and rehabilitation of dystonia is a challenge that continuously requires exploration of novel interventions. This review will initially describe the underlying neurophysiological mechanisms of the motor impairments found in childhood dystonia followed by the clinical measurement tools that are available to document the presence and severity of symptoms. Finally, we will discuss the state-of-the-art of therapeutic options for childhood dystonia, with particular emphasis on emergent and innovative strategies.

Translating the science into practice: shaping rehabilitation practice to enhance recovery after brain damage

The revolution in neuroscience provided strong evidence for learning-dependent neuroplasticity and presaged the role of motor learning as essential for restorative therapies after stroke and other disabling neurological conditions. The scientific basis of motor learning has continued to evolve from a dominance of cognitive or information processing perspectives to a blend with neural science and contemporary social-cognitive-psychological science, which includes the neural and psychological underpinnings of motivation. This transformation and integration across traditionally separate domains is timely now that clinician scientists are developing novel, evidence-based therapies to maximize motor recovery in the place of suboptimal solutions. We will review recent evidence pertaining to therapeutic approaches that spring from an integrated framework of learning-dependent neuroplasticity along with the growing awareness of protocols that directly address the patient's fundamental psychological needs. Of importance, there is mounting evidence that when the individual's needs are considered in the context of instructions or expectations, the learning/rehabilitation process is accelerated.

Gradual training reduces the challenge to lateral balance control during practice and subsequent performance of a novel locomotor task

Locomotor balance control mechanisms and impairments have been well described in the literature. In contrast, the role of evidence-based motor learning strategies in the recovery or restoration of locomotor balance control has received much less attention. Little is known about the efficacy of motor learning strategies to improve locomotor tasks and their unique requirements, such as lateral balance control. This study examined whether gradual versus sudden training influenced lateral balance control among unimpaired adults (n=16) during training and 24-h transfer performance of a novel locomotor task. This was accomplished by examining the variability of whole-body frontal plane kinematics throughout training and 24-h transfer performance of asymmetric split-belt treadmill walking. Compared to sudden training, gradual training significantly reduced the challenge to lateral balance control (exhibited by a reduction in frontal plane kinematic variability) during training and during subsequent transfer task performance. These results indicate that gradual training could play an important role in restoring locomotor balance control during physical rehabilitation.

Task-oriented rehabilitation robotics

Task-oriented training is emerging as the dominant and most effective approach to motor rehabilitation of upper extremity function after stroke. Here, the authors propose that the task-oriented training framework provides an evidence-based blueprint for the design of task-oriented robots for the rehabilitation of upper extremity function in the form of three design principles: skill acquisition of functional tasks, active participation training, and individualized adaptive training. The previous robotic systems that incorporate elements of task-oriented trainings are then reviewed. Finally, the authors critically analyze their own attempt to design and test the feasibility of a TOR robot, ADAPT (Adaptive and Automatic Presentation of Tasks), which incorporates the three design principles. Because of its task-oriented training-based design, ADAPT departs from most other current rehabilitation robotic systems: it presents realistic functional tasks in which the task goal is constantly adapted, so that the individual actively performs doable but challenging tasks without physical assistance. To maximize efficacy for a large clinical population, the authors propose that future task-oriented robots need to incorporate yet-to-be developed adaptive task presentation algorithms that emphasize acquisition of fine motor coordination skills while minimizing compensatory movements.

Adaptive regulation of assistance 'as needed' in robot-assisted motor skill learning and neuro-rehabilitation

We propose a general adaptive procedure to select the appropriate degree of assistance based on a Bayesian mechanism used to estimate psychophysical thresholds. This technique does not need an accurate model of learning and recovery processes. This procedure is validated in the context of a motor skill learning problem (control of a virtual object), in which the controller is used to gradually increase task difficulty as learning proceeds. These automatic adjustments of task difficulty or the degree of assistance can be used to promote not only motor skill learning but also neuromotor recovery.

Feasibility of the adaptive and automatic presentation of tasks (ADAPT) system for rehabilitation of upper extremity function post-stroke

BACKGROUND

Current guidelines for rehabilitation of arm and hand function after stroke recommend that motor training focus on realistic tasks that require reaching and manipulation and engage the patient intensively, actively, and adaptively. Here, we investigated the feasibility of a novel robotic task-practice system, ADAPT, designed in accordance with such guidelines. At each trial, ADAPT selects a functional task according to a training schedule and with difficulty based on previous performance. Once the task is selected, the robot picks up and presents the corresponding tool, simulates the dynamics of the tasks, and the patient interacts with the tool to perform the task.

METHODS

Five participants with chronic stroke with mild to moderate impairments (> 9 months post-stroke; Fugl-Meyer arm score 49.2 ± 5.6) practiced four functional tasks (selected out of six in a pre-test) with ADAPT for about one and half hour and 144 trials in a pseudo-random schedule of 3-trial blocks per task.

RESULTS

No adverse events occurred and ADAPT successfully presented the six functional tasks without human intervention for a total of 900 trials. Qualitative analysis of trajectories showed that ADAPT simulated the desired task dynamics adequately, and participants reported good, although not excellent, task fidelity. During training, the adaptive difficulty algorithm progressively increased task difficulty leading towards an optimal challenge point based on performance; difficulty was then continuously adjusted to keep performance around the challenge point. Furthermore, the time to complete all trained tasks decreased significantly from pretest to one-hour post-test. Finally, post-training questionnaires demonstrated positive patient acceptance of ADAPT.

CONCLUSIONS

ADAPT successfully provided adaptive progressive training for multiple functional tasks based on participant's performance. Our encouraging results establish the feasibility of ADAPT; its efficacy will next be tested in a clinical trial.

The binding of learning to action in motor adaptation

In motor tasks, errors between planned and actual movements generally result in adaptive changes which reduce the occurrence of similar errors in the future. It has commonly been assumed that the motor adaptation arising from an error occurring on a particular movement is specifically associated with the motion that was planned. Here we show that this is not the case. Instead, we demonstrate the binding of the adaptation arising from an error on a particular trial to the motion experienced on that same trial. The formation of this association means that future movements planned to resemble the motion experienced on a given trial benefit maximally from the adaptation arising from it. This reflects the idea that actual rather than planned motions are assigned 'credit' for motor errors because, in a computational sense, the maximal adaptive response would be associated with the condition credited with the error. We studied this process by examining the patterns of generalization associated with motor adaptation to novel dynamic environments during reaching arm movements in humans. We found that these patterns consistently matched those predicted by adaptation associated with the actual rather than the planned motion, with maximal generalization observed where actual motions were clustered. We followed up these findings by showing that a novel training procedure designed to leverage this newfound understanding of the binding of learning to action, can improve adaptation rates by greater than 50%. Our results provide a mechanistic framework for understanding the effects of partial assistance and error augmentation during neurologic rehabilitation, and they suggest ways to optimize their use.

Toward 'optimal' schemes of robot assistance to facilitate motor skill learning

We investigate whether and on what circumstances physical interaction with a robot may facilitate the acquisition of a novel motor skill. We focus on two different motor tasks: (i) intermanual transfer of cursive handwriting and (ii) acquisition of a putting skill. In the case of handwriting, we found that intermanual transfer is facilitated by forms of interaction that account for the temporal aspects of the movements. In the case of putting, we found that guidance is helpful in improving longitudinal error (a matter of speed accuracy), but not directional error (a matter of position accuracy). Based on these results, we draw some tentative conclusions on which tasks can benefit from guidance, and how robots should be programmed to maximize their effect.

Visual feedback reduces co-contraction in children with dystonia

Inappropriate muscle activation and co-contraction are important features in childhood dystonia, and clinical interventions are often targeted to reduce the excess muscle activation. Previous research has shown that visual biofeedback of muscle activity can help people to reduce excess muscle activation in a variety of motor disorders. To investigate the effectiveness of similar techniques for dystonia, we had participants perform a tracking task with and without visual feedback of co-contraction. Children with dystonia had greater levels of co-contraction than children without dystonia. Most importantly, individuals were able to reduce their co-contraction significantly when visual biofeedback was provided. These results indicate that children with dystonia are able to control co-contraction, at least to a certain extent, provided attention can be directed to the excess muscle activation. These results also suggest that methods of biofeedback focusing on inappropriate muscle activations might provide a clinical benefit for treatment of children with dystonia.

The effects of error augmentation on learning to walk on a narrow balance beam

Error augmentation during training has been proposed as a means to facilitate motor learning due to the human nervous system's reliance on performance errors to shape motor commands. We studied the effects of error augmentation on short-term learning of walking on a balance beam to determine whether it had beneficial effects on motor performance. Four groups of able-bodied subjects walked on a treadmill-mounted balance beam (2.5-cm wide) before and after 30 min of training. During training, two groups walked on the beam with a destabilization device that augmented error (Medium and High Destabilization groups). A third group walked on a narrower beam (1.27-cm) to augment error (Narrow). The fourth group practiced walking on the 2.5-cm balance beam (Wide). Subjects in the Wide group had significantly greater improvements after training than the error augmentation groups. The High Destabilization group had significantly less performance gains than the Narrow group in spite of similar failures per minute during training. In a follow-up experiment, a fifth group of subjects (Assisted) practiced with a device that greatly reduced catastrophic errors (i.e., stepping off the beam) but maintained similar pelvic movement variability. Performance gains were significantly greater in the Wide group than the Assisted group, indicating that catastrophic errors were important for short-term learning. We conclude that increasing errors during practice via destabilization and a narrower balance beam did not improve short-term learning of beam walking. In addition, the presence of qualitatively catastrophic errors seems to improve short-term learning of walking balance.

Effects of physical guidance on short-term learning of walking on a narrow beam

Physical guidance is often used in rehabilitation when teaching patients to re-learn movements. However, the effects of guidance on motor learning of complex skills, such as walking balance, are not clear. We tested four groups of healthy subjects that practiced walking on a narrow (1.27 cm) or wide (2.5 cm) treadmill-mounted balance beam, with or without physical guidance. Assistance was given by springs attached to a hip belt that applied restoring forces towards beam center. Subjects were evaluated while walking unassisted before and after training by calculating the number of times subjects stepped off of the beam per minute of successful walking on the beam (Failures per Minute). Subjects in Unassisted groups had greater performance improvements in walking balance from pre to post compared to subjects in Assisted groups. During training, Unassisted groups had more Failures per Minute than Assisted groups. Performance improvements were smaller in Narrow Beam groups than in Wide Beam groups. The Unassisted-Wide and Assisted-Narrow groups had similar Failures per Minute during training, but the Unassisted-Wide group had much greater performance gains after training. These results suggest that physical assistance can hinder motor learning of walking balance, assistance appears less detrimental for more difficult tasks, and task-specific dynamics are important to learning independent of error experience.

Can robots help the learning of skilled actions?

Learning to move skillfully requires that the motor system adjusts muscle commands based on ongoing performance errors, a process influenced by the dynamics of the task being practiced. Recent experiments from our laboratories show how robotic devices can temporarily alter task dynamics in ways that contribute to the motor learning experience, suggesting possible applications in rehabilitation and sports training.

Motor Adaptation to Single Force Pulses: Sensitive to Direction but Insensitive to Within-Movement Pulse Placement and Magnitude

Although previous experiments have identified that errors in movement induce adaptation, the precise manner in which errors determine subsequent control is poorly understood. Here we used transient pulses of force, distributed pseudo-randomly throughout a movement set, to study how the timing of feedback within a movement influenced subsequent predictive control. Human subjects generated a robust adaptive response in postpulse movements that opposed the pulse direction. Regardless of the location or magnitude of the pulse, all pulses yielded similar changes in predictive control. All current supervised and unsupervised theories of motor learning presume that adaptation is proportional to error. Current neural models that broadly encode movement velocity and adapt proportionally to motor error can mimic human insensitivity to pulse location, but cannot mimic human insensitivity to pulse magnitude. We conclude that single trial adaptation to force pulses reveals a categorical strategy that humans adopt to counter the direction, rather than the magnitude, of movement error.

Computational motor control in humans and robots

Computational models can provide useful guidance in the design of behavioral and neurophysiological experiments and in the interpretation of complex, high dimensional biological data. Because many problems faced by the primate brain in the control of movement have parallels in robotic motor control, models and algorithms from robotics research provide useful inspiration, baseline performance, and sometimes direct analogs for neuroscience.