Although practice is considered to be the most

Transcription

1 Determining the Optimal Challenge Point for Motor Skill Learning in Adults With Moderately Severe Parkinson s Disease Somporn Onla-or, PhD, Carolee J. Winstein, PhD Objective. To test the predictions of the Challenge Point Framework (CPF) for motor learning in individuals with Parkinson s disease (PD) by manipulating nominal task difficulty and conditions of practice. Methods. Twenty adults with PD and 20 nondisabled controls practiced 3 variations of a laboratory-based goal-directed arm movement over 2 days. A between-group (PD, nondisabled) 2-factor design compared 2 levels of nominal task difficulty (low, high) and 2 levels of practice condition (low, high demand). Learning was assessed with a no-feedback recall test 1 day after practice. Performance was quantified using a root mean square error difference between the goal and participant-generated movement. Results. All participants improved with practice. Under the low-demand practice condition, adults with PD demonstrated comparable learning to that of controls when nominal task difficulty was low but not high. In contrast, under the high-demand practice condition, adults with PD demonstrated preserved motor learning for both levels of task difficulty, but only if recall was tested under the same context as that used during practice. Conclusions. In general, the predictions of CPF were supported. Together, the level of nominal task difficulty and the inherent demand of the practice condition played a critical role in determining the optimal challenge point for motor learning in individuals with PD. More important, and in contrast to the predictions of CPF, a highdemand practice condition appeared to have a facilitative effect on motor learning. However, this benefit revealed the context specificity of motor learning in adults with PD. From the Department of Physical Therapy, Faculty of Associated Medical Sciences, Chiang Mai University, Chiang Mai, Thailand (SO); the Division of Biokinesiology and Physical Therapy, School of Dentistry (CJW); and Department of Neurology, Keck School of Medicine, University of Southern California, Los Angeles (CJW). Address correspondence to Somporn Onla-or, PhD, Department of Physical Therapy, Faculty of Associated Medical Sciences, Chiang Mai University, 110 Intrawaroroj Road, Sripoom, Maung, Chiang Mai Thailand. onlaor@chiangmai.ac.th. Onla-or S, Winstein CJ. Determining the optimal challenge point for motor skill learning in adults with moderately severe Parkinson s disease. Neurorehabil Neural Repair 2008;22: DOI: / Key Words: Motor learning Practice order Feedback frequency Task demand Parkinson s disease Challenge Point Framework. Although practice is considered to be the most important factor responsible for the permanent improvement in the ability to perform a motor skill (ie, motor learning), other factors such as the conditions of practice, the difficulty of the task, and the learner s skill level can significantly influence the proficiency of motor skill learning. The Challenge Point Framework (CPF) has recently been proposed as a means to determine the optimal challenge point for motor learning. 1 Using CPF, interactions between nominal task difficulty of the to-be-learned task and the learner s skill level together with the specific conditions of practice create a level of functional task difficulty that determines how much information will be available for motor learning. This idea emerges largely from studies done with control populations. The fundamental assumption of CPF is that learning is a problem-solving process and that the information available during and after each attempt to solve the problem is remembered and forms the basis for learning. 1 Too much or too little information will retard learning. According to CPF, the characteristic of the learner affects how they respond to certain practice conditions. Thus, a practice condition proven to benefit motor learning for nondisabled adults may not benefit or may even degrade motor learning in adults with neurologic deficits such as Parkinson s disease (PD). 2 Evidence from functional neuroanatomic studies and clinical studies reveals that the integrity of the basal ganglia is essential for both cognitive and motor operations of procedural learning. 3-6 Although numerous studies have shown that motor learning is impaired in adults with PD, 4,7-12 the learning deficits are neither global nor consistently found The inconsistencies among previous findings are due to variations across studies in task demand and the conditions Copyright 2008 The American Society of Neurorehabilitation 385

2 Onla-or, Winstein of practice (ie, functional task difficulty in CPF) as well as disease severity (ie, learner s skill level in CPF). In addition, there were variations in the methods used to assess motor learning. For example, several studies did not account for the well-known learning-performance distinction in the experimental design. 20 Manipulations of practice order and feedback frequency have been shown to influence motor learning capability. A random practice order (eg, A-C-B-, C-B-A-, B-A-C-, each letter represents a different task) generally benefits motor learning more than a blocked practice order (eg, A-A-A-, B-B-B-, C-C-C-). The benefit of random practice over blocked practice is due to the higher level of contextual interference (CI) induced by random practice order. 21,22 One interpretation of the CI effect is that random practice requires more cognitive processing during practice compared to that for blocked practice CPF predicts that for less skilled learners, a low CI practice condition would benefit motor learning more than a high CI practice condition. In fact, the findings from a recent study in adults with PD support this prediction. 25 Feedback is often provided in the form of knowledge of results (KR). Previous work using relatively simple tasks suggests that feedback using a low frequency of KR is as effective or more beneficial for motor learning than that using a high frequency of KR One interpretation of the KR frequency effect is that less frequent KR enhances learning by forcing the individual to engage in cognitive processing essential for skill acquisition. In contrast, a high frequency of KR provides the solution for the learner and, thus, diminishes the cognitive demand for solving the task problem. 26,27,30,31 Similar to the CPF predictions for the CI effect, learners with less skill would benefit more from a high frequency of KR compared to that for a low frequency of KR. In agreement with CPF prediction, previous studies have shown that for a relatively difficult task and/or less skilled learners, high KR frequency benefits motor learning more than low KR frequency. 29,32-34 Within CPF, nominal task difficulty reflects a constant amount of task difficulty regardless of the performer or the environmental circumstances. Thus, it includes such factors as perceptual and motor requirements for the task. For example, when spatial accuracy is controlled, a movement with a short movement time (MT) goal would have a higher level of nominal task difficulty than a movement with a longer MT goal. In this study, a 1500 ms MT goal was used for the low level of nominal task difficulty, whereas a 900 ms MT goal was used for the high level of nominal task difficulty. In contrast to nominal task difficulty, the level of functional task difficulty is influenced by several factors including the conditions of practice. For example, when a task with low nominal difficulty (ie, 1500 ms MT goal) is being performed under a low-demand practice condition (ie, blocked order and 100% KR), the task is considered to poses low functional task difficulty. In contrast, when a task with high nominal difficulty (ie, 900 ms MT goal) is being performed under a highdemand practice condition (ie, random order and 60% KR), the task is considered to possess high functional task difficulty. Finally, when a task with low nominal task difficulty is being performed under a high-demand practice condition or when a task with high nominal task difficulty is being performed under a low-demand practice condition, the task is considered to possess intermediate functional task difficulty. The purpose of this study was to test the predictions of the CPF for motor learning in individuals with PD by manipulating nominal task difficulty and the conditions of practice. To our knowledge, this study is the first to explore motor learning through a systematic manipulation of 3 important factors: the learner s skill level (ie, PD, controls), the conditions of practice that include practice order (ie, blocked, random) and feedback frequency (ie, 100% KR, 60% KR), and nominal task difficulty (ie, low and high). We tested the hypothesis that participants with PD will demonstrate comparable learning to that of age-matched, nondisabled controls when functional task difficulty is low. In contrast, participants with PD will demonstrate poorer motor learning relative to nondisabled controls when functional task difficulty is high. METHODS Participants Twenty adults with PD and 20 nondisabled, agematched volunteers were enrolled. Participants with PD were recruited if they had moderate illness classified by the modified Hoehn and Yahr stage II or III 35 and were optimally medicated for PD. They participated in the investigation during the on medication cycle. To ensure homogeneity within PD subgroups and to control for relevant confounding factors, all participants were screened for dementia (scores 28 on the Mini-Mental State Examination [MMSE], 36 depression (score < 10 on the Center for Epidemiologic Studies Depression [CES-D] short form scale), 37 and hand-dominance (right-handed with 80% cutoff on the Edinburgh Handedness Inventory 38 ). All participants gave written informed consent and participated in the study under a protocol approved by the Institutional Review Board of the University of Southern California Health Sciences Campus. Instrumentation and Task A lightweight lever affixed to a frictionless vertical axle was attached to a table and positioned parallel to 386 Neurorehabilitation and Neural Repair 22(4);2008

3 Motor Skill Learning in PD Experimental Design and Procedure Figure 1. (A) Three goal movement trajectories with the same spatial scaling but different temporal scaling at 900, 1200, and 1500 ms. (B) An example of the 2 augmented visual feedback, a graphic representation of the subject s movement trajectory for that trial superimposed over the goal trajectory, and the root mean square error (RMSE) score on the top right corner. the floor. A handle at the end of the lever was adjusted to accommodate the participant s forearm. A linear potentiometer was attached to the transducer at the base of the vertical axle. Signals from the transducer were converted to a digital signal by an A-D board of a Dell 466v computer and sampled at 200 Hz. The Template software program 39 was used to manipulate the movement trajectory and the interval duration and to store data from each trial for further analyses. The task was to grasp the handle of a lever and move it horizontally at the correct speed and distance to replicate a goal movement trajectory, which was displayed on the computer monitor before each trial. The goal was to learn a continuous movement with 2 elbow extensionflexion reversal actions. This goal movement was to be performed under 3 different movement times (ie, 900, 1200, and 1500 ms; Figure 1A). During acquisition, after each trial, the participant was presented with either the trial number together with 2 types of feedback or the trial number alone. The 2 types of feedback were (1) an overall error score, the root mean square error (RMSE), representing the difference between the goal movement task and the participant s response, and (2) a graphic representation of the response superimposed with the goal movement task (Figure 1B). A 2 Group (control [Con], PD) 2 Practice condition (low-demand practice condition, high-demand practice condition) between-factor design was implemented. This overall design resulted in 4 experimental groups: Con-low demand, Con-high demand, PD-low demand, and PD-high demand. The 3 movement time goals (900, 1200, and 1500 ms) were constant under both conditions of practice (ie, within factor design). The experimental protocol included an acquisition phase (days 1 and 2) and delayed retention phase (day 3). During acquisition, participants in the low-demand condition practiced the tasks under blocked order and received 100% KR frequency, whereas those in the high-demand condition practiced the tasks under random order and received 60% KR frequency. The 60% KR was scheduled as follows: 100% for trials 1 to 15, 60% for trials 16 to 30, and 20% for trials 31 to 45 in each trial set. All participants practiced 6 sets of 45 trials during acquisition (3 sets each day), resulting in a total of 270 trials. There were 2 retention test conditions (blocked and random retention). The order of the retention tests was counterbalanced across participants. During the retention test, participants performed the same movements (ie, recall test) they practiced the day before but without any KR feedback. Outcome Measures and Statistical Analyses Global performance accuracy (RMSE) was quantified across acquisition and delayed retention phases. RMSE is the average difference between goal movement trajectory and the participant s response calculated over the participant s total movement time. 40 Individual RMSE data were grouped into 9-trial blocks for the acquisition phase (blocks 1-10) and 5-trial blocks for the retention (blocks 11-12) phase. Although 3 movement time goals were used to grade nominal task difficulty, we were particularly interested in the 2 extreme nominal task difficulty levels, 1500 ms (low) MT goal and 900 ms (high) MT goal. The 1200 ms MT goal was included in the design to reduce the number of repetitive representations of the same trajectory in the random practice condition. Therefore, only data from the 900 ms and 1500 ms MT goals were included in the analyses. A 2 Group (Con, PD) 2 Practice condition (low demand, high demand) 2 Nominal task difficulty (low, high) 10 Block analysis of variance (ANOVA) with repeated measures on the last factor was conducted on the acquisition data to provide an overall description of the performance for the 2 groups. Performance during the retention test was used to assess motor learning. Retention data from each practice condition were analyzed separately to Neurorehabilitation and Neural Repair 22(4);

4 Onla-or, Winstein Table 1. Group Mean Comparisons for the Demographic Characteristics. Group by Condition of Practice Assessment Con-low PD-low P Value Con-high PD-high P Value Sample size Age (years) 66.0 ± ± ± ± Education (years) 16.4 ± ± ± ± Cognitive: MMSE (max 30) 29.5 ± ± ± ± Emotional status: CES-D (max 30) 3.4 ± ± ± ± Hoehn and Yahr (1-5) NA 2.6 ± 0.6 NA 2.5 ± 0.7 PD duration (years) NA 7.3 ± 4.2 NA 10.2 ± 5.8 Abbreviations: PD, Parkinson s disease; MMSE, Mini-Mental State Examination; CES-D, Center for Epidemiologic Studies Depression. Mean ± standard deviation and P value of each variable for participants in the low-demand practice condition (Con-low demand and PD-low demand) are shown in the left column, and those for participants in the high-demand practice condition (Con-high demand and PD-high demand) are shown in the right column. accommodate inclusion of the 2 retention tests (random, blocked) as a within-factor. Therefore, retention data for the PD-low demand subgroup were compared to the Con-low demand subgroup and those for the PD-high demand subgroup were compared to the Con-high demand subgroup. Two separate 2 Group (Con, PD) 2 Nominal task difficulty (low, high) 2 Retention test (blocked, random) ANOVA were conducted on mean RMSE to address our questions regarding the motor learning capability of individuals with PD compared to nondisabled controls. Post hoc analyses of any significant main effects and interactions were conducted using pairwise comparisons with a Bonferroni correction. A partial eta squared statistic was used to index the strength of association (ie, effect size, ES). 41 Independent 2-sample t tests were conducted to assess differences in demographic characteristics between PD and nondisabled groups. For all statistical tests, significance level was set at P <.05. RESULTS Participant Demographic Characteristics Group mean comparisons for the demographic characteristics are summarized in Table 1. Group comparisons were made separately within the same practice condition. For the low-demand practice condition, there were no differences between groups for age, education, or MMSE score. For the high-demand practice condition, there was no difference between groups for age. However, participants with PD had a higher education level but lower MMSE score than the nondisabled controls. None of the participants had clinical signs of dementia or depression. Mean score on the CES-D was, however, higher for the PD group than the control group for both practice conditions. The disease severity was similar for the 2 PD subgroups (mean Hoehn and Yahr stage: PD-low demand = 2.6, PD-high demand = 2.5). For all participants with PD, their PD-related symptoms presented bilaterally with some postural instability. However, they were physically independent. Average PD duration was 7.3 years (range, 3-15 years) for the PD-low demand subgroup and 10.2 years (range, 4-20 years) for the PD-high demand subgroup. Acquisition Performance RMSE block means for each group by condition of practice and motor task during acquisition are displayed in Figure 2. All participants improved their performance with practice (Block Effect; P <.001). Overall, performance was not different between the 2 groups (Group Effect; P =.07). Participants who practiced in blocked order with 100% KR frequency (ie, Con-low demand and PD-low demand, top row) were generally more accurate than those who practiced in random order with 60% KR frequency (ie, Con-high demand and PD-high demand, bottom row) (Practice Condition Effect; P <.001). Similarly, all participants were more accurate in reproducing the low nominal task difficulty (ie, 1500 ms trajectory, right column) than that of the relatively higher nominal task difficulty (ie, 900 ms trajectory, left column) (Nominal task difficulty Effect; P <.001). Across the 2 conditions of practice, individuals with PD had higher error than nondisabled controls for the high but not for the low nominal task difficulty (Group Nominal task difficulty interaction; P <.05). The estimated effect size for this interaction was large (ES =.41). 42 This group difference for the high nominal task difficulty was due primarily to the high error of the PD-low demand subgroup during the first half of practice (Block 2-5, Figure 2 top left). The Group Practice condition interaction was not significant (P =.64), suggesting that during acquisition, 388 Neurorehabilitation and Neural Repair 22(4);2008

5 Motor Skill Learning in PD Figure 2. Root mean square error (RMSE) block means during acquisition days 1 (block 1-5) and 2 (block 6-10). Top row illustrates data from the low-demand practice condition (blocked practice order with 100% knowledge of results [KR]), and bottom row illustrates those from the high-demand practice condition (random practice order with 60% KR). Left column illustrates data from the high nominal task difficulty (900 ms trajectory), and right column illustrates those from the low nominal task difficulty (1500 ms trajectory). RMSE is shown in degrees. Error bars are standard error of mean (SEM). participants with PD benefited from practice order and feedback frequency as nondisabled controls. Retention Performance Low-demand practice condition subgroups (con-low demand, PD-low demand). For participants who practiced under blocked order and received 100% KR frequency, the PD group was as accurate as the nondisabled, control group in reproducing the low nominal difficulty task (1500 ms trajectory) either when tested in blocked retention or random retention (Figure 3, top right). However, when nominal task difficulty was high (900 ms trajectory), the PD group performed with significantly greater error than the control group for both blocked and random retention tests (Figure 3, top left). This Group Nominal task difficulty interaction was significant (P <.02, ES =.22). High-demand practice condition subgroups (con-high demand, PD-high demand). For participants who practiced under random order and received 60% faded KR frequency, retention data showed that the context of the retention test (whether it is the same as or different from the practice condition) primarily affected the learning capability of the PD group. Individuals with PD exhibited greater error than controls when tested in blocked retention for both low and high nominal task difficulty conditions (1500 ms and 900 ms trajectories). Conversely, individuals with PD were as accurate as nondisabled controls when tested in random retention, again for both low and high nominal task difficulty conditions (Figure 3, bottom row). These findings resulted in a significant Group Retention Test interaction; P <.01, ES =.33. Findings that the PD-high demand subgroup performed relatively well compared to controls in the random retention but not in the blocked retention test suggest that practice under the random order together with receiving Neurorehabilitation and Neural Repair 22(4);

6 Onla-or, Winstein Figure 3. Mean root mean square error (RMSE) during retention (Ret) tests (blocked and random). Top row illustrates data from the low-demand practice condition (blocked practice order with 100% knowledge of results [KR]), and bottom row illustrates those from the high-demand practice condition (random practice order with 60% KR). Left column illustrates data from the high nominal task difficulty (900 ms trajectory), and right column illustrates those from the low nominal task difficulty (1500 ms trajectory). RMSE is shown in degrees. Error bars are standard error of mean (SEM). 60% KR frequency evoked context-specific learning in individuals with PD. This context-specific learning effect in the PD-high demand subgroup was evident independent of the order of the retention test condition (ie, participants performed the random retention before or after the blocked retention). This effect is illustrated for 2 exemplar participants who had different retention test orders (PD-01, PD-05) in Figure 4. The context-dependent learning effect in adults with PD is illustrated in Figure 5. The retention data (averaged across the 2 nominal task difficulties) for each practice condition were compared within each subgroup. For the control group (Figure 5A), both Con-low demand and Con-high demand subgroups showed a smaller error in the blocked compared with the random retention test (Retention Effect; P <.003, ES =.41). Statistical analysis revealed no significant Practice Condition main effect or interaction, suggesting that the 2 practice conditions (low, high) had similar effects on retention performance of nondisabled controls. Unlike the control groups, the pattern of findings was different between the 2 PD subgroups. Specifically, the PD-low demand subgroup showed a smaller error in the blocked relative to random retention. In contrast, the PD-high demand subgroup showed a smaller error in the random relative to blocked retention (Figure 5B). This resulted in a significant Practice condition Retention condition interaction (P <.004, ES =.39). Post hoc analyses identified a significant difference between blocked and random retention for the PD-high demand subgroup (P <.02, ES =.50) but not the PD-low demand subgroup (P =.07, ES =.40). 390 Neurorehabilitation and Neural Repair 22(4);2008

7 Motor Skill Learning in PD Figure 5. Mean root mean square error (RMSE) (averaged across 2 nominal task difficulty) for each group by conditions of practice. (A) Data from the control subgroups (Con-low demand, Con-high demand). (B) Data from PD subgroups (PD-low demand, PD-high demand). The control group showed lower error in blocked than in random retention (ret) independent of the conditions of practice (Retention Effect; P <.003). In contrast, the PD group showed lower error in the similar practice-retention context than in the different practice-retention context. This resulted in a significant Practice condition Retention interaction (P <.004). RMSE is shown in degrees. Error bars are SEM. PD = Parkinson s disease. Figure 4. Example of individual data from the PD-high demand subgroup over acquisition and 2 retention tests. Top row shows data from participant PD-01 who performed random retention test first and then blocked retention test. Bottom row shows data from a different participant (PD-05) who performed blocked retention test first then random retention test. Participants with PD who practiced in the highdemand condition showed higher error in the blocked retention than in the random retention independent of the order of which retention test they performed first. B indicates, blocked retention test; R, random retention test; PD, Parkinson s disease; RMSE = root mean square error. DISCUSSION The aim of the present study was to test the predictions of the Challenge Point Framework for motor learning in individuals with PD by manipulating nominal task difficulty and the cognitive demand inherent in the practice condition. The CPF offered useful scaffolding for understanding the importance of the motor task and conditions of practice for the information processing capability of this population. This study appears to be the first to manipulate the conditions of practice (feedback frequency, practice schedule), nominal task difficulty (movement time), and retention test condition (blocked, random) within the same experiment. Thus, there is no definitive information from previous work that would allow us to accurately predict the expected results. When averaged across the 2 nominal task difficulties, our findings show that the 2 practice conditions (low, high) had similar effects on retention performance only for the nondisabled controls. Of particular interest, the results showed contextspecific learning characteristics only for the adults with moderately severe PD. Our findings showed that the level of nominal difficulty of the to-be-learned task and information processing demands inherent in the practice conditions do interact and play a critical role in determining motor learning capability for the PD group. Under a lowdemand practice condition, individuals with moderately severe PD demonstrated comparable learning to nondisabled, age-matched participants when nominal task difficulty was also low but demonstrated learning deficits when nominal task difficulty was high. However, in contrast to the predictions of CPF, and in comparison to the nondisabled group, under the high-demand practice condition, individuals with PD demonstrated comparable motor learning, but only when the context of the recall test was the same as that during practice (ie, random retention). This context-specific effect is not a new concept in the PD literature and has variously been referred to as a deficit in task-switching, stuck-in-set-perseveration, 47 and motor set inflexibility. 48,49 The context-specific learning effect does suggest that the nature of the transfer/retention test is important for future motor learning research, especially for understanding the learning capability in adults with PD. Neurorehabilitation and Neural Repair 22(4);

8 Onla-or, Winstein Discussion of Methods and Experimental Design The Challenge Point Framework led us to speculate that some of the discrepancies in previous studies of motor learning capability in adults with PD were due to differences in the level of nominal task difficulty, the practice conditions employed, and the severity level of the disease that is well known to affect information processing capability. 7,12,50 Therefore, we included only individuals with moderate severity (modified Hoehn and Yahr stage II-III) and systematically manipulated the cognitive demand through scheduling of the motor tasks and the frequency of performance feedback, along with the level of nominal task difficulty. In addition and to account for the well-known temporary performance effects induced during practice, motor learning was assessed by performance during no feedback, delayed recall (ie, a transfer test design). We deliberately chose 2 retention conditions, blocked and random, to examine the generalizability and context-specific nature of motor learning in this group. This was done because there is some evidence that motor set inflexibility is one characteristic of PD that would predict context-specific learning Effects of Low-Demand Practice Condition and 2 Levels of Nominal Task Difficulty on Motor Learning in Individuals With PD In the low-demand practice condition, participants practiced the same task throughout the practice set and received feedback after every trial. It is well established that although a low contextual interference and a high frequency of feedback usually enhance acquisition performance, this combination would likely degrade learning, at least for nondisabled young adults However, because of the alterations in cognitive processing, this may not be the case for older adults or individuals with PD. For example, recent work has shown that a condition with high-frequency KR is superior to one with a low frequency of KR for motor learning in PD, 32 and similarly, a blocked order practice condition is superior to a randomorder practice condition for motor learning in PD. 25 Consistent with these findings and our hypothesis, the PD group showed comparable learning to the nondisabled control group when functional task difficulty was low. When functional task difficulty was set at an intermediate level through an increase in nominal task difficulty, individuals with PD showed a distinct deficit in motor learning compared with nondisabled controls and for both recall tests. The added task difficulty was the critical element that surpassed the optimal challenge point for the PD group. To achieve the 900 ms movement time goal, participants had to produce a force rapidly (ie, high acceleration) with the same spatial requirements. It is not surprising that the 900 ms task was difficult for the PD group, given that reduced force generation and time to reach peak velocity have been shown to be a characteristic of motor control in PD Partial eta squared revealed a medium effect size for the Group Nominal task difficulty interaction (ES =.22). The disparate results for the low and high nominal task difficulty, using the same low-demand condition of practice, highlight the importance of nominal task difficulty for motor learning in PD. These results are consistent with previous findings that demonstrated comparable motor learning for adults with PD and nondisabled controls when pursuit rotor speed was low, but worse learning for adults with PD when pursuit rotor speed was high. 50,54 Effects of High-Demand Practice Condition and 2 Levels of Nominal Task Difficulty on Motor Learning in Individuals With PD Our results from the low-demand practice condition described above support the hypothesis that when functional task demand is low, motor learning is preserved in individuals with moderately severe PD. In contrast, when the practice condition evokes a high cognitive demand and functional task demand varies (either high or intermediate), the results depend on the specific retention test used to assess learning. This finding had a medium strength effect size (Group Retention interaction, ES =.33). The significant interaction between retention test condition and group highlights the context-specific learning allowed through the unique experimental design, discussed further in the next section. Contrary to expectations, participants with PD showed comparable learning to nondisabled controls when functional task demand was high. The CPF predicted that high cognitive demand induced through practice together with a high motor demand task would be detrimental for learning in individuals with PD, in large part due to an already compromised cognitive and motor processing capability. 3-6 In contrast, our findings showed that the added cognitive demand induced from high contextual interference and reduced feedback resulted in comparable motor learning. The mechanism underlying this learning effect is unclear and may be different from that of nondisabled controls. The basal ganglia are known to have a role in task-switching, and the cortico-striatal feedback loop is important for internal cuing of motor execution Thus, the random practice order and reduced feedback frequency may directly challenge the task-switching and internal cueing deficits 392 Neurorehabilitation and Neural Repair 22(4);2008

9 Motor Skill Learning in PD of PD. It may be possible that by forcing individuals with moderately severe PD to practice task-switching over a sufficient number of practice trials, they eventually habituate and overcome these known deficits. It should be noted that for each practice set, feedback was given at 100% for the first 15 trials, then 60% for trials 16 to 30, and 20% for trials 31 to 45. This faded feedback frequency may have allowed individuals with PD to use external information during early practice to improve their performance and then to gradually adjust to the reduced feedback by engaging the appropriate cognitive processes necessary to learn the task. Context-Specific Learning in Individuals With PD For the high-demand practice conditions, the context of the recall test was critical for demonstrating comparable motor learning capability to that of the nondisabled age-matched control group. This significant group effect had a large effect size (ES =.50). The evidence of context-specific learning, exclusively for the PD group, is indicative of a dysfunction of the basal ganglia. Set-shifting deficits or stuck-in-set-perseveration, a characteristic that has long been identified in PD, may mediate the less flexible learning capability identified here The set-shifting deficit is thought to be due to impairment of the dorsolateral prefrontal circuit in selection mechanisms, necessary for disengaging from a previous task set and engaging a new one. 43,58 Evidence presented here that individuals with PD showed preserved learning only when practiced under the high cognitive demand condition and only when learning was assessed in the same context underscores the importance of exploring the optimal challenge point for the learner. Even under a high functional task difficulty condition, adults with PD demonstrated comparable motor learning to that of controls (ie, 900 ms movement). This has important implications for designing rehabilitation programs. learner characteristics, and recall/retention test condition should be considered when designing rehabilitation programs for motor skill learning. Our findings of motor learning under conditions of high contextual interference even with reduced levels of feedback suggest that adults with moderately severe PD can overcome the known task-switching deficits if tested in the same context as that of the practice phase. Commonly, individuals with PD are not sufficiently challenged in the rehabilitation settings, in part because of known motor and cognitive deficits. We have demonstrated that, in fact, the challenge itself may be what benefits motor learning. The beneficial effect from a high-demand practice condition is similar to activitydependent neuroplasticity that has recently been identified in individuals with PD who undergo intensive exercise programs. 59 Training programs for individuals with PD that use high contextual interference conditions may not only facilitate learning but also translate to the real-world situation where normally responses must be adapted to the environmental context. Similar suggestions have been offered for this population. 60 Together, our results suggest that motor rehabilitation programs for adults with moderately severe PD should be designed to be challenging with a relatively high cognitive demand and to exploit the context specificity of learning through recall tests that are similar in nature to the practice condition. Although these results are promising for application to the neurorehabilitation setting, there are a few limitations that need to be considered. First, generalizability should be demonstrated using functional activities of daily living in place of our laboratory-based motor tasks. Second, little is known about the dose of practice needed to overcome the learning deficits commonly attributed to individuals with PD. Acquiring motor skills under a high-demand practice condition may require extended practice compared to that under a low-demand condition. Indeed, previous work has shown that extended practice was beneficial for motor learning in adults with PD. 16,61 IMPLICATIONS AND LIMITATIONS Our experimental design allowed us to uncover a significant interplay between nominal task difficulty and the conditions of practice for determining motor learning capability and to reveal the context-specific learning that is characteristic of adults with moderately severe PD. Had we manipulated only a single dimension (ie, conditions of practice or nominal task difficulty or retention test condition) in our design, we might have arrived at a far different conclusion. We suggest that a combination of factors including the conditions of practice, nominal task difficulty, REFERENCES 1. Guadagnoli MA, Lee TD. Challenge point: a framework for conceptualizing the effects of various practice conditions in motor learning. J Mot Behav. 2004;36(2): Kuan L-W, Barker R. New therapeutic approaches to Parkinson s disease including neural transplants. Neurorehabil Neural Repair. 2005;19: Middleton FA, Strick PL. Basal ganglia output and cognition: evidence from anatomical, behavioral, and clinical studies. Brain Cogn. 2000;42(2): Vakil E, Kahan S, Huberman M, Osimani A. Motor and non-motor sequence learning in patients with basal ganglia lesions: the case of serial reaction time (SRT). Neuropsychologia. 2000;38(1):1-10. Neurorehabilitation and Neural Repair 22(4);

10 Onla-or, Winstein 5. Owen AM, Doyon J, Dagher A, Sadikot A, Evans AC. Abnormal basal ganglia outflow in Parkinson s disease identified with PET. Implications for higher cortical functions. Brain. 1998;121(Pt 5): Packard MG, Knowlton BJ. Learning and memory functions of the Basal Ganglia. Annu Rev Neurosci. 2002;25: Doyon J, Gaudreau D, Laforce R Jr, et al. Role of the striatum, cerebellum, and frontal lobes in the learning of a visuomotor sequence. Brain Cogn. 1997;34(2): Verschueren SM, Swinnen SP, Dom R, De Weerdt W. Interlimb coordination in patients with Parkinson s disease: motor learning deficits and the importance of augmented information feedback. Exp Brain Res. 1997;113(3): Krebs HI, Hogan N, Hening W, Adamovich SV, Poizner H. Procedural motor learning in Parkinson s disease. Exp Brain Res. 2001;141(4): Sarazin M, Deweer B, Merkl A, Poser NV, Pillon B, Dubois B. Procedural learning and striatofrontal dysfunction in Parkinson s disease. Mov Disord. 2002;17(2): Shin JC, Ivry RB. Spatial and temporal sequence learning in patients with Parkinson s disease or cerebellar lesions. J Cogn Neurosci. 2003;15(8): Siegert RJ, Taylor KD, Weatherall M, Abernethy DA. Is implicit sequence learning impaired in Parkinson s disease? A metaanalysis. Neuropsychology. 2006;20(4): Soliveri P, Brown RG, Jahanshahi M, Caraceni T, Marsden CD. Learning manual pursuit tracking skills in patients with Parkinson s disease. Brain. 1997;120(Pt 8): Agostino R, Sanes JN, Hallett M. Motor skill learning in Parkinson s disease. J Neurol Sci. 1996;139(2): Sommer M, Grafman J, Clark K, Hallet M. Learning in Parkinson s disease: eyeblink conditioning, declarative learning, and procedural learning. J Neurol Neurosurg Psychiatry. 1999;67(1): Behrman AL, Cauraugh JH, Light KE. Practice as an intervention to improve speeded motor performance and motor learning in Parkinson s disease. J Neurol Sci. 2000;174(2): Smith J, Siegert RJ, McDowall J, Abernethy D. Preserved implicit learning on both the serial reaction time task and artificial grammar in patients with Parkinson s disease. Brain Cogn. 2001;45 (3): Smiley-Oyen AL, Lowry KA, Emerson QR. Learning and retention of movement sequences in Parkinson s disease. Mov Disord. 2006;21(8): Jessop RT, Horowicz C, Dibble LE. Motor learning and Parkinson disease: Refinement of movement velocity and endpoint excursion in a limit of stability balance task. Neurorehabil Neural Repair. 2006;20(4): Cahill L, McGaugh JL, Weinberger NM. The neurobiology of learning and memory: some reminders to remember. Trends Neurosci. 2001;24(10): Shea JB, Morgan R. Contextual interference effects on the acquisition, retention, and transfer of motor skills. J Exp Psychol Hum Learn Mem. 1979;3: Lee TD, Magill RA. The locus of contextual interference in motor skill-acquisition. J Exp Psychol Hum Learn Mem. 1983;9: Hall KG, Magill RA. Variability of practice and contextual interference in motor skill learning. J Mot Behav. 1995;27(4): Albaret JM, Thon B. Differential effects of task complexity on contextual interference in a drawing task. Acta Psychol (Amst). 1998;100(1-2): Lin C, Sullivan KJ, Wu AD, Kantak S, Winstein CJ. The effect of task practice order for motor skill learning in adults with Parkinson s disease: a pilot study. Phys Ther. 2007; 87(9): Weeks DL, Sherwood DE. A comparison of knowledge of results scheduling methods for promoting motor skill acquisition and retention. Res Q Exerc Sport. 1994;65(2): Winstein CJ, Schmidt RA. Reduced frequency of knowledge of results enhances motor skill learning. J Exp Psychol Learn Mem Cogn. 1990;16: Winstein CJ, Merians AS, Sullivan KJ. Motor learning after unilateral brain damage. Neuropsychologia. 1999;37(8): Guadagnoli MA, Dornier LA, Tandy RD. Optimal length for summary knowledge of results: the influence of task-related experience and complexity. Res Q Exerc Sport. 1996;67(2): Schmidt RA, Bjork PA. New conceptualizations of practice: common principles in three paradigms suggest new concept for training. Psychol Sci. 1992;3: Anderson DI, Magill RA, Sekiya H. Motor learning as a function of KR schedule and characteristics of task-intrinsic feedback. J Mot Behav. 2001;33(1): Guadagnoli MA, Leis B, Van Gemmert AW, Stelmach GE. The relationship between knowledge of results and motor learning in Parkinsonian patients. Parkinsonism Relat Disord. 2002;9(2): Lai Q, Shea CH. The role of reduced frequency of knowledge of results during constant practice. Res Q Exerc Sport. 1999;70(1): Wulf G, Horger M, Shea CH. Benefits of blocked over serial feedback on complex motor skill learning. J Mot Behav. 1999;31(1): Goetz CG, Poewe W, Rascol O, et al. Movement Disorder Society Task Force report on the Hoehn and Yahr staging scale: status and recommendations. Mov Disord. 2004;19(9): Folstein MF, Folstein SE, McHugh PR. Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3): Andresen EM, Malmgren JA, Carter WB, Patrick DL. Screening for depression in well older adults: evaluation of a short form of the CES-D (Center for Epidemiologic Studies Depression Scale). Am J Prev Med. 1994;10(2): Oldfield RC, Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia. 1971;9(1): Hary D. Template (version 3.0) Computer Software [computer program]. Santa Monica, CA: Integrated Scientific Resources; Schmidt RA, Lee TD. Conditions of practice. In: Schmidt RA, Lee TD, eds. Motor Control and Learning: A Behavioral Emphasis. Vol. 11. Champaign, IL: Human Kinetics; 2005: Pierce CA, Block RA, Aguinis H. Cautionary note on reporting eta-squared values from multifactorial ANOVA designs. Ed Psychol Meas. 2004;64: Cohen J. Quantitative methods in psychology: a power primer. Psychol Bull. 1992;112(1): Cools R, Barker RA, Sahakian BJ, Robbins TW. Mechanisms of cognitive set flexibility in Parkinson s disease. Brain.2001;124(Pt 12): Hayes AE, Davidson MC, Keele SW, et al. Toward a functional analysis of the basal ganglia. J Cogn Neurosci. 1998;10(2): Fales CL, Vanek ZF, Knowlton BJ. Backward inhibition in Parkinson s disease. Neuropsychologia. 2006;44(7): Onla-or S, Winstein CJ. Function of the direct and indirect pathways of the basal ganglia motor loop: evidence from reciprocal aiming movements in Parkinson s disease. Cogn Brain Res. 2001;10(3): Sandson J, Albert ML. Perseveration in behavioral neurology. Neurology. 1987;37(11): Robertson C, Flowers KA. Motor set in Parkinson s disease. J Neurol Neurosurg Psychiatry. 1990;53(7): Chong RK, Horak FB, Woollacott MH. Parkinson s disease impairs the ability to change set quickly. J Neurol Sci. 2000;175(1): Harrington DL, Haaland KY, Yeo RA, Marder E. Procedural memory in Parkinson s disease: impaired motor but not visuoperceptual learning. J Clin Exp Neuropsychol. 1990;12(2): Stelmach GE, Teasdale N, Phillips J, Worringham CJ. Force production characteristics in Parkinson s disease. Exp Brain Res. 1989;76(1): Neurorehabilitation and Neural Repair 22(4);2008

11 Motor Skill Learning in PD 52. Phillips JG, Martin KE, Bradshaw JL, Iansek R. Could bradykinesia in Parkinson s disease simply be compensation? J Neurol. 1994;241(7): Bishop M, Brunt D, Pathare N, Ko M, Marjama-Lyons J. Changes in distal muscle timing may contribute to slowness during sit to stand in Parkinson s disease. Clin Biomech (Bristol, Avon). 2005; 20(1): Bondi MW, Kaszniak AW. Implicit and explicit memory in Alzheimer s disease and Parkinson s disease. J Clin Exp Neuropsychol. 1991;13(2): Brown RG, Marsden CD. Internal versus external cues and the control of attention in Parkinson s disease. Brain. 1988;111(Pt 2): Georgiou N, Iansek R, Bradshaw JL, Phillips JG, Mattingley JB, Bradshaw JA. An evaluation of the role of internal cues in the pathogenesis of parkinsonian hypokinesia. Brain. 1993;116(Pt 6): Cools AR, van den Bercken JH, Horstink MW, van Spaendonck KP, Berger HJ. Parkinson s disease: a reduced ability to shift to a new grouping if not prompted or guided. Mov Disord. 1990;5(2): Cronin-Golomb A, Corkin S, Growdon JH. Impaired problem solving in Parkinson s disease: impact of a set-shifting deficit. Neuropsychologia. 1994;32(5): Fisher BE, Wu A, Gordon J, Lin C. Effects of High Intensity Exercise Using Body Weight Treadmill Training on Functional Recovery and Neuroplasticity in Individuals With Parkinson s Disease. Platform presentation at Society for Neuroscience: 2006; Atlanta, GA. 60. Morris ME. Movement disorders in people with Parkinson disease: a model for physical therapy. Phys Ther. 2000;80(6): Platz T, Brown RG, Marsden CD. Training improves the speed of aimed movements in Parkinson s disease. Brain. 1998;121 (Pt 3): Neurorehabilitation and Neural Repair 22(4);