The authors have declared that no competing interests exist.
Conceived and designed the experiments: CM LJB JSR. Performed the experiments: ML MG. Analyzed the data: ML JSR JB MG LJB CM. Contributed reagents/materials/analysis tools: JB MG LJB. Wrote the paper: ML JSR JB MG LJB CM.
Most patients receiving intensive rehabilitation to improve their upper limb function experience pain. Despite this, the impact of pain on the ability to learn a specific motor task is still unknown. The aim of this study was to determine whether the presence of experimental tonic pain interferes with the acquisition and retention stages of motor learning associated with training in a reaching task. Twenty-nine healthy subjects were randomized to either a Control or Pain Group (receiving topical capsaicin cream on the upper arm during training on Day 1). On two consecutive days, subjects made ballistic movements towards two targets (NEAR/FAR) using a robotized exoskeleton. On Day 1, the task was performed without (baseline) and with a force field (adaptation). The adaptation task was repeated on Day 2. Task performance was assessed using index distance from the target at the end of the reaching movement. Motor planning was assessed using initial angle of deviation of index trajectory from a straight line to the target. Results show that tonic pain did not affect baseline reaching. Both groups improved task performance across time (p<0.001), but the Pain group showed a larger final error (under-compensation) than the Control group for the FAR target (p = 0.030) during both acquisition and retention. Moreover, a Group x Time interaction (p = 0.028) was observed on initial angle of deviation, suggesting that subjects with Pain made larger adjustments in the feedforward component of the movement over time. Interestingly, behaviour of the Pain group was very stable from the end of Day 1 (with pain) to the beginning of Day 2 (pain-free), indicating that the differences observed could not solely be explained by the impact of pain on immediate performance. This suggests that if people learn to move differently in the presence of pain, they might maintain this altered strategy over time.
Pain is one of the most common and disabling symptoms following injury to the peripheral (e.g. amputation) or central nervous system (e.g. spinal cord injury or stroke). In motor rehabilitation programs, a large proportion of patients receiving intensive training to improve their upper limb function experience neuropathic pain. Epidemiologic data indicate that compared to patients with similar injuries but without associated pain, patients experiencing pain exhibit poorer motor outcomes, suggesting that they might have limited ability to relearn previous motor patterns, or to learn new ones in order to compensate for residual deficits. For example, four months after a stroke, 32% of patients suffer from moderate to severe pain,
In parallel, studies using experimental pain in humans and animals have shown that pain exerts modulatory influences over the activity in motor pathways. It has been demonstrated that pain can lead to a reduction of maximal voluntary contraction, a decrease in endurance and changes in coordination during dynamic motor tasks (see
An important question regarding the effect of pain on motor learning concerns its impact on the retention of performance. To establish that a novel task has been learned, other criteria need to be met beyond a simple improvement in performance occurring within a training session. Indeed, motor learning involves several steps: improvement in performance during training (acquisition), transfer to longer-term memory (consolidation), and the ability to recall the stored motor memory (retention).
Application of low-frequency (inhibitory) repetitive TMS to M1 prior to force field exposure has been shown to disrupt the retention of this type of learning when tested 24 h later, without affecting performance during the acquisition phase. It suggests that M1 might be important for initiating the development of long-term motor memories.
The general objective of this study was to determine whether the presence of experimental tonic pain interferes with motor learning associated with training in a reaching task. To achieve this goal, we compared two experimental groups (Control and Pain) during a force field adaptation task performed on two consecutive days. Subjects were tested during both baseline and perturbed reaching movements on Day 1 (with or without pain), and only in the perturbed movement condition without pain on Day 2. Specific objectives were to determine the effect of tonic pain on motor performance during:
The ethical review board of the Institut de réadaptation en déficience physique de Québec (IRDPQ, Québec, Canada) approved the study. All participants provided their written informed consent prior to inclusion. The individual appearing on
Thirty right-handed healthy individuals with no reported history of neurological or musculoskeletal problems that could affect performance during the task were recruited and randomized to either the Control group or the Pain group. One subject from the Control group was excluded from the analyses because he reported shoulder pain at the end of the experiment on Day 1.
Experiments were carried out on two consecutive days. The reaching task was assessed using a KINARM robotized exoskeleton (BKIN Technologies, Canada), that allows force field application at a given joint.
Subjects were informed of their group assignment after Baseline 1. The experimental tonic pain was induced with a single topical application of 1% capsaicin cream. A thin layer (∼1 mm) of cream forming a 1 cm-wide ring was applied around the upper arm (just above the elbow) of the trained limb. Once the cream was applied, subjects had to verbally rate their pain intensity every 3 minutes using a numeric rating scale (NRS) ranging from 0 (no pain) to 10 (the worst pain that can be imagined). The experiment resumed when pain reached a plateau (average of 25.1±3.9 minutes). A wait period of 30 minutes was also imposed to the Control group between Baseline 1 and Baseline 2.
Subjects made right arm ballistic reaching movements to visual targets using the KINARM robotic exoskeleton. They performed the reaching movements, with the arm in the horizontal plane, to one of two targets (random sequence) located 10 cm away from the central starting position (one at 120° (FAR) and one at 300° (NEAR) – i.e. requiring multijoint coordination of the elbow [flexion-extension] and shoulder [horizontal adduction-abduction]). The central starting position was determined in order to achieve a standardized posture of 50° of horizontal abduction [with respect to the sagittal plane] and 90° of elbow flexion). Note that the KINARM exoskeleton allows movement at the elbow and the shoulder, but not at the wrist. Visual feedback of index position, start position, and targets was presented in the same plane as the arm using an overhead projector and a half-silver mirror.
From the starting position, subjects were instructed to “shoot” through the target as quickly and precisely as possible, beginning their movement as soon as the target appeared in the virtual environment. As soon as the index crossed an invisible 10-cm radius circle centered on the starting position, the robot produced a dampening field to rapidly stop the movement.
One hundred trials (50/target) were performed in each period (i.e. for each Baseline 1 and 2, Acquisition and Retention). A velocity-dependent resistive force (−3 Nm·s/rad) was applied at the elbow by the KINARM during the Acquisition and Retention periods. Joint angular positions of the elbow and shoulder were obtained from KINARM motor encoders and sampled at 1 kHz. The position of the index was computed in real-time by the Dexterit-E software of the KINARM system. Data processing was made with Matlab (MathWorks, R2011b).
Two main variables were used to assess motor performance: the initial angle of deviation (iANG) and the final error (fERR). The iANG reflects subject's motor planning as it is based on the initial part of the movement only (prior to the first peak of acceleration). iANG was computed as the angle between: 1) a line joining the position of the index at movement onset and the target; and 2) another line joining the position of the index at movement onset and at its first peak of acceleration (see
This figure depicts the kinematic variables extracted from index finger trajectories, using examples of typical trials early in the Acquisition period (i.e. trajectories are strongly deviated in the direction of the force field). The fERR is measured as the distance between the index and the target when the index crossed the invisible 10-cm radius circle centered on the starting position. The iANG is computed as the angle between: 1) a line joining the position of the index at movement onset and the target; and 2) another line joining the position of the index at movement onset and at its first peak of acceleration.
As the force field applied was velocity-dependent, average movement speed (total index distance/total movement time (excluding reaction time)) was also computed and analyzed to ensure a comparable force-field perturbation during the Adaptation phase on both days.
Each variable was plotted as a function of trial number to obtain time course curves. The following sections of the time course curves were then defined and used for statistical comparisons:
Baseline 1 and 2 – last 10 trials of each Baseline period.
Early Adaptation - trials 2 to 11 of the Adaptation period (e.g. during force-field application; Day 1 or 2) (the first trial was never included in the analysis window as the force field was turned on unexpectedly).
Late Adaptation - last 10 trials of the Adaptation period (e.g. during force-field application; analyzed for Day 1 only).
Descriptive statistics are reported as mean ± standard deviation. Statistical analyses for all variables consisted of three-way repeated measures ANOVAs to evaluate the between-group effect (Pain vs. Control), and the within group effects of Target (NEAR/FAR), and Time. For Objective 1, Time factor had two levels, Baseline 1 and Baseline 2. For Objectives 2 and 3 (tested in the same analysis), Time factor had three levels, Early Adaptation Day 1, Late Adaptation Day 1 and Early Adaptation Day 2. Post-hoc analyses were performed using a Sidak correction for multiple comparisons. All statistical analyses were performed using SPSS 13 software (SPSS Inc., Chicago, IL, USA).
The Control (n = 14, 8 females, aged 26.6±4.8) and Pain Groups (n = 15, 7 females, aged 25.8±4.1) were similar in terms of age or sex. During the application of pain on Day 1, the average pain intensity was of 7.8±0.9 at the beginning of Baseline 2 and of 7.5±1.0 at the end of the Adaptation period. Pain was therefore at a high level and remained stable throughout the experiment.
ANOVAs performed to assess the impact of pain on baseline performance revealed no effect of Group or Time x Group interaction on either iANG or fERR, which shows that pain induction between both baselines for the Pain group did not alter motor performance. The only significant difference was found between targets for both iANG (p = 0.003) and fERR (p<0.001), reflecting small differences in the movement kinematics.
The fact that pain had no impact on baseline motor performance is essential to the interpretation of subsequent analyses, especially considering that the Pain group was tested with pain on Day 1 (during acquisition), and without pain on Day 2.
On Day 1 (acquisition), the force field perturbation induced substantial movement errors for both groups in the early period of Adaptation. Rapidly, task performance improved and motor strategy adjusted as outlined by the time course of these variables (
For iANG, there was no Group effect (p = 0.249), while the expected main effect of Time was observed (p<0.001). Post-hoc analysis for the main effect of Time showed that subjects exhibited a reversal of the direction of their initial deviation during the Acquisition phase (Day 1 Late adaptation vs. Day 1 Early adaptation; p<0.001): early in the force field the hand was deflected in the direction of the force field, but later they started to initiate their movement in the opposite direction in anticipation of the perturbation (
For fERR, analyses showed no Group effect (p = 0.100) and the expected significant main effect of Time (p<0.001). Further a Target x Group interaction (p = 0.035) was observed (
The analysis of average movement speed showed only a significant effect of Time (p<0.001). Post-hoc analysis indicated that movement speed was significantly smaller (p<0.01) on Day 1 Early adaptation (37.0 cm/s±1.8), but did not differ (p = 0.255) between Day 1 Late adaptation (40.3 cm/s±1.9) and Day 2 Early Adaptation (41.5 cm/s±2.0). The lack of difference between groups ensures that no systematic difference in the velocity-dependent force-field exposure can account for the differences observed between groups on the other variables.
Results of this study show that the presence of a significant level of acute tonic pain (∼7.5/10) did not influence baseline motor performance in a simple reaching task. Moreover subjects still significantly improved their performance in a new reaching adaptation task when training with pain, and exhibited retention when tested pain free 24 hours later. However two main differences were observed between the Pain and Control Groups. First, the Pain group showed larger changes in their feedforward strategy to minimize movement error in the force field over time, irrespective of the target. Second, the Pain group exhibited a systematic under-compensation for the force field in their final errors for one of the two targets (FAR).
The large majority of previous studies that assessed the impact of pain on upper limb motor control have been designed to understand the impact of musculoskeletal pain. In these studies, pain was applied to a very focused region, often modulated by the movement itself, and induced no or only slight kinematics changes.
Interestingly, the evolution of the motor strategy over time to attain the task's goal differed between groups. When they initially faced the force field on Day 1, the Pain group tended to be more deviated in the direction of the force field than the Control group, despite experiencing similar force field intensity. Progressively both groups modified their motor planning to counteract the force field, and initiated their movement trajectory at an angle (away from a straight line)
A few studies have investigated the role of proprioceptive feedback in adaptation to force fields during reaching, and have shown that visual feedback can largely compensate for deficient proprioceptive feedback.
In this experiment, fERR, used to assess subjects' task performance, showed an impact of pain both during the acquisition (with pain) and on next-day retention (tested pain-free). The fact that the differences observed remained on Day 2 indicates that increased errors cannot solely be explained by the impact of pain on immediate performance. It rather suggests that pain interfered with the acquisition process itself. Importantly however, this deficit was observed only for the FAR target. How can differences between targets be explained? Adapting to the force field for the FAR target does not appear to be a more challenging task, as the errors observed for the Control group on Early Day 1 were not statistically different than for the NEAR target. However both targets required different patterns of multijoint coordination. Reaching towards the NEAR target involved horizontal abduction (average excursion of 21.2±1.3) combined with elbow flexion (20.4±1.1), while the FAR target required horizontal adduction (20.9°±1.2) combined with elbow extension (25.4°±1.7). The fact that the FAR target required more elbow excursion, and that both pain and the force field were acting at the elbow, might contribute to explain the observed results. A recent study has shown that the patterns of motor control adaptation to a noxious stimulation (hypertonic saline injection) differ according to the task performed.
Such task dependency of motor control adaptation to pain might also contribute to explain discrepancies between the results obtained in previous studies. Boudreau et al. showed that local pain induced by capsaicin interfered with motor acquisition during a tongue protrusion tracking task.
Interestingly, behaviour of the Pain group was very similar from the end of Day 1 (with pain) to the beginning of Day 2 (pain-free). This indicates that the differences observed between groups cannot be explained solely by the impact of pain on immediate performance. This is a clinically relevant observation as it suggests that motor strategies developed in the presence of acute pain might be maintained over time. It has been argued that departure from “normal” movement patterns in response to pain may not be ideal and might lead to detrimental effects in the long term, although the alternative strategy employed is effective in the short term to achieve the task's goal.
In conclusion, the results of this study show that tonic pain: 1) had no impact on baseline reaching performance; 2) resulted in more final error (under-compensation) for one target in both the acquisition and retention phases of learning during reaching tasks perturbed by a force-field; 3) resulted in a slightly altered motor strategy consisting in a larger adjustment in the feedforward component of the movement. Importantly, the strategy and performance of Day 1 carried over to the next day, despite the fact that subjects were retested in the absence of pain. This is an important observation as it suggests that even in the case of short-duration pain, moving differently while in pain can have an impact on how people will move subsequently even when pain is gone.
Comparison of these results with previous studies stresses the need for more studies on the effect of pain on motor learning to investigate the effect of different pain models, but also of different motor tasks involving visual feedback or not.
The authors wish to thank Mathieu Baril for technical development.