Understanding how pain interacts with motor learning is important for sports training or rehabilitation of motor function in the presence of pain. Pain affects the performance of movements, and under certain conditions can interfere with motor adaptations associated with learning (1,2). However, few studies have assessed whether experiencing pain during learning impacts subsequent performance of the same task after resolution of nociceptive input.
Adaptations to pain can interfere with movement planning and execution (2), but the consequences for performance appear task-specific (3). When performing low-intensity tasks in the presence of pain, such as isometric contractions around 5% of maximal force, reorganization of muscle activity can cause small deviations in force direction, even when the overall level of activity, recorded by surface electrodes, remains unchanged (4,5). In contrast, during high-intensity tasks, such as fast arm-reaching movements, reduced activation of the painful muscle can significantly impact task performance, for example, reducing force magnitude and movement acceleration (6,7).
The effects of pain on the acquisition and retention of motor skills and motor adaptations appear to be task-dependent. For example, when the negative consequences of poor task execution were negligible (e.g., tongue movements), pain was associated with less improvement in performance and suppressed gains in cortical excitability that were normally observed during pain-free training (8,9). However, when movement accuracy was essential (e.g., avoiding falls during walking), results showed no differences in performance when learning a new task with or without pain (10,11).
As improvements in performance within a training session do not necessarily translate into long-term retention of motor skills (12), recent work has begun to focus on the effects of pain beyond the practice period. In the context of sports training and rehabilitation, long-term retention of motor skills is more important than performance during a single training session. Force field perturbations offer a suitable framework for assessing the initial acquisition and subsequent retention of motor performance (13). Perturbations initially cause large kinematic errors, which are gradually reduced over consecutive trials. Upon reexposure to the same perturbation, smaller errors and faster learning rates indicate retention of the capacity to compensate for the perturbation; a phenomenon known as “savings.” In a study assessing locomotor adaptation to a force field over two sessions, performance gains observed during pain were not retained when the task was repeated after resolution of pain (10). Although the results suggest that skills learned during pain may not properly transfer to pain-free conditions, pain did not modify the muscle activation strategy used to resist the perturbation—possibly because the constraints of walking preclude the possibility for substantial adaptations. During arm-reaching movements, pain reduced gains in performance during force field adaptation (9). After resolution of pain, initial improvements in performance were retained and movement accuracy gradually converged to pain-free levels. Large errors in the angle of movement during pain suggest strong feedforward adjustments, but muscle activation was not recorded; hence, it is impossible to infer about the motor strategies used—and even subtle changes in strategy may have consequences for tissue health (14). Furthermore, these studies induced topical skin pain, and it is unclear whether similar effects would be observed during musculoskeletal pain, as nociceptive inputs that originate from cutaneous and muscle tissue trigger distinct sensory and motor responses (15). In a previous study, we found that saline-induced muscle pain did not affect locomotor performance during low-intensity ankle perturbations, but caused delayed muscle activation within the step cycle compared with the pain-free control group (16). Although these activation delays were reduced upon pain-free training of the same task, suggesting a shift toward the strategy used by controls, it is impossible to rule out the influence of automatic spinal circuits involved in gait.
The current study aimed to assess whether acute muscle pain affects movement accuracy and muscle activation strategies during arm-reaching movements: (i) performed without a force field, (ii) during adaptation to a force field perturbation, and (iii) during subsequent pain-free reexposure to the same perturbation within the same experiment session. We expected that pain would reduce activation of the painful muscle, but with limited impact on movements under the null field due to the low task intensity. In contrast, we expected reduced movement speed and accuracy during force field adaptation, when vigorous muscle activation is required to compensate for the perturbations. After resolution of pain, we expected task error to progress toward that of pain-free participants. An important aspect was to determine whether motor strategies developed during pain would reoccur upon subsequent training, despite absence of pain.
Twenty-two participants volunteered for this study (12 women; age, 28 ± 6 yr). Participants reported no major respiratory, neurological, or cardiovascular conditions, and no history of pain in the upper limbs. Written informed consent was obtained from each participant prior to commencement of the study, which was approved by the Human Research Ethics Committee at The University of Queensland and conformed to the Declaration of Helsinki.
Participants performed horizontal arm-reaching movements in the forward direction while grasping the handle of a robotic manipulandum (vBOT) (Fig. 1A). The vBOT is a two-link carbon fiber manipulandum, which is driven by motors operating on timing pulleys to control endpoint torque (for details, see Howard et al. (17)). Participants were instructed to move as fast as possible to a target 15 cm in front of the neutral start position. Real-time feedback on hand position was provided via a computer monitor mounted above the table and projected via a mirror, effectively blocking the participants’ vision of the actual movement. Start and target positions were displayed as 1-cm-radius white and yellow circles, respectively (see Fig. 1), and hand position was represented as a 0.5-cm-radius red circle.
For each movement, the target appeared after the hand position cursor had remained inside the start circle for 500 ms, and disappeared when the target was acquired, determined when the cursor remained within the target circle continuously for 100 ms. A loud tone signaled target appearance and acquisition. Text feedback, together with a distinct auditory beep, was provided to adjust performance if the movement started before the target was shown (“too fast”), if movement did not start within 500 ms from target appearance (“too slow”), or if the peak movement speed was lower than 95 cm·s−1 (“too slow”). Throughout the study, an air-sled supported the weight of the participant’s forearm to minimize potential effects of fatigue. The position and velocity of the hand and the forces applied to the handle were recorded at 1000 Hz.
Throughout the experiment, the vBOT generated one of three force field dynamic environments (Fig. 1B): 1) a null field, where the robot imposed no forces to the hand; 2) a viscous curl field (see below); and 3) a strong force “channel” that restricted lateral deviations and constrained movements within a straight line to the target so that the lateral forces applied to the handle could be measured (17). During the viscous force field, the vBOT created a velocity-dependent force field, given by;
where the force field constant k was set to 13 N·m·s−1. During channel trials, a high-stiffness elastic force field confined the movement to a straight path with a spring constant of 4000 N·m.
Participants were familiarized with the setup and experimental task by performing 100 to 200 movements under a null field. After familiarization, they performed six blocks of 100 movements each (Fig. 1C): baseline 1, baseline 2, force field 1, washout 1, force field 2, washout 2. During each block, one of five movements was randomly set to “channel” trial (i.e., 20 per block). The remaining 80 movements (“nonchannel” trials) were performed under null field during baseline and washout conditions and under a velocity-dependent field during force field conditions. Short rest breaks were given after baseline and washout conditions, but not after force field conditions. Instead, the force field was suddenly switched off in the transition from force field to washout conditions as participants continued to perform reaching movements.
Participants were randomly assigned to either a pain or a control group (11 on each group). The pain group received two injections of hypertonic saline (5% NaCl) to temporarily induce pain in the anterior deltoid muscle: one immediately before baseline 2, and another before the force field 1 condition (Fig. 1C). The control group received two nonpainful injections of isotonic saline (0.9% NaCl) into the same muscle. Participants were asked to verbally rate the intensity of their pain approximately every 25 movements on a numerical rating scale anchored with “no pain” at zero and “worst pain imaginable” at 10. The first injection was used to assess the effects of nociceptive input on null field movements (baseline), and the second injection was used to assess how nociceptive input affects motor learning, that is, adaptation and deadaptation to the force field. The force field 2 condition was initiated at least 1 min after pain had completely resolved.
EMG was recorded using pairs of surface Ag/AgCl electrodes (Kendall; Covidien, USA) over the major muscles that perform shoulder and elbow movements during reaching: the anterior deltoid (ADEL), posterior deltoid (PDEL), biceps brachii (BB), triceps brachii (TRB), and pectoralis major muscles (PECM). EMG data were amplified 1000 times, band-pass filtered between 10 and 1000 Hz (ZeroWire EMG, Cometa, Italy), sampled at 2000 Hz (NI-DAQ; National Instruments, USA), and synchronized to the vBOT data. EMG data from four participants (two in each group) were excluded because the battery of the wireless preamplifiers ran out in the middle of the experiment.
Force data, measured during channel trials, were low-pass filtered at 6 Hz using a fourth-order Butterworth filter. Movement onset was detected as the time when hand speed first exceeded 2 cm·s−1, and movement offset as the time when hand speed first dropped below 2 cm·s−1 after peak speed. To quantify movement accuracy during each trial, the peak hand speed and the peak perpendicular error (i.e., signed peak orthogonal distance from a straight path to the target) were calculated. Channel trials were used to calculate a force adaptation index as the slope of linear regression between the ideal force profile (i.e., the force that would exactly overcome the force-field to maintain a straight path: force-field constant × speed) and the actual force applied to the handle (13), which would equal one if these force profiles were identical to one another. For this analysis, baseline channel force profiles were subtracted from each force time series. To measure the initial rate of learning, the force adaptation index of the first 30 trials of force field and washout conditions were fit to an exponential function using the MatLab fit function. To make the estimation more robust, we used bootstrap to resample the data from each group with replacement 2000 times, and for each bootstrap sample we compared the exponential coefficients of the fitted functions between groups. If the presence of pain during training significantly affected motor adaptations, the Control group would show faster learning rates (i.e., greater exponential coefficient) than the Pain group in at least 95% of the bootstrap samples (considering a significance level of P = 0.05).
EMG signals were digitally band-pass filtered between 30 and 500 Hz, full-wave rectified, and low-pass filtered at 10 Hz (all second-order Butterworth filters) to generate EMG envelopes. For each trial, EMG envelopes were assessed in a time window from 100 ms prior to movement onset until movement offset. The average EMG amplitude of each muscle was calculated during each reaching movement. To account for individual differences in magnitude of EMG signals, the average EMG amplitudes were normalized by the median value obtained during baseline 1. We also performed more detailed analysis on the individual EMG waveforms: EMG envelopes from each movement were amplitude-normalized to the median peak EMG recorded during baseline 1 and time-normalized to 60 samples, centered at the time of peak speed. To assess adaptations in motor strategy relative to baseline, we averaged the EMG envelope of each participant across the last 50 trials of baseline 1 and subtracted it from each individual EMG envelope across all conditions. The resultant EMG envelopes were then statistically tested between and within groups (see Statistics). To assess the time profile of muscle cocontraction, we used the original EMG envelopes (i.e., without subtracting the baseline patterns) and extracted the magnitude of (normalized) EMG that was equal and opposite in agonist–antagonist muscle pairs (18). For each sample, we recorded the value of the smaller EMG envelope, that is, discarding the amount of EMG that was not matched by EMG in the opposing muscle, and the resultant time series was compared between groups using the same procedure used for the EMG envelopes (see “Statistics”). This cocontraction time series has been originally termed “wasted contraction,” because it does not contribute to the net torque (18). However, it has been shown that the increased cocontraction of antagonist muscles, rather than wasted, performs a critical role in the learning of novel dynamics by stabilizing the limb, resisting to force field perturbations, and allowing for the refinement of movement accuracy (19,20). Therefore, this time series will be simply referred to as “cocontraction.” Force parameters (i.e., force profiles and force adaptation index) were assessed from channel trials, whereas position, velocity, and EMG parameters were assessed during “normal” (nonchannel) trials.
Average pain scores of both groups were compared using unpaired two-sample t tests. Measures of movement accuracy and average EMG amplitude were assessed using a three-way repeated-measures ANOVA with group (pain, control) as between-subjects factor, and condition (baseline 1, baseline 2, force field 1, washout 1, force field 2, washout 2) and epoch (early: first 20 trials, late: last 20 trials) as within-subject factors. When the residuals were not normally distributed, data were transformed using either logarithm or square power transformations (the latter used when negative values were present), which resulted in normally distributed residuals. Bonferroni post hoc tests with corrections for multiple comparisons were used to assess significant main effects and interactions. Data are presented as mean ± SEM in text and figures.
A linear mixed effects (LME) model was used to statistically compare the shape of EMG envelopes between groups based on the technique of wavelet functional ANOVA (for details see McKay et al. (21)). Briefly, EMG envelopes are transformed in the wavelet domain, where temporally localized features can be well represented by a small number of orthogonal (independent) wavelet coefficients, which are then statistically tested. To achieve this, we transformed the EMG envelopes from the early (first 20) and late (last 20) trials of each condition into the wavelet domain using third-order coiflets. The wavelet coefficients from each epoch (early, late) were compared between groups using an LME model, and significant between-group contrasts were identified and transformed back into the time domain for visualization. The same technique was also used to test the cocontraction time series. We then used a similar procedure to assess the degree of within-group retention of motor strategy between the first and second exposures to the force field by comparing the EMG envelopes between conditions force fields 1 and 2, and between washouts 1 and 2.
Pain and movement accuracy
The proportion of male and female participants was balanced between groups (females pain, 7; control, 5; t test P = 0.42). Injections of hypertonic saline resulted in higher pain scores than isotonic saline (t test pain vs control, average pain scores first injection: 4.2 ± 0.3 vs 0.2 ± 0.1, t10 = 10.4, P < 0.001; second injection: 3.0 ± 0.4 vs 0.3 ± 0.2, t10 = 5.2, P < 0.001). The difference in pain scores persisted until after the last movement of baseline 2 (pain vs control: 3.2 ± 0.4 vs 0.1 ± 0.1), as well as force field 1 and washout 1 conditions (pain vs control: 1.4 ± 0.3 vs 0.1 ± 0.1).
The position, velocity, and force profiles were generally similar between groups throughout the study (Fig. 2). During baseline trials, hand trajectories followed a straight path and only small amplitude forces were applied to the handle. Application of the force field initially caused large deviations in hand trajectory, which were gradually reduced as participants learned to compensate for the force perturbation. During washout, the force field was suddenly removed and deviations in hand trajectory to the opposite direction were observed in the initial trials, before position and force profiles returned to baseline levels.
The movement accuracy parameters across each experimental condition are shown in Figure 3. No significant main effect of Group was found in the parameters assessed (all F1,20 < 1.41, P > 0.24). Despite minor, nonsignificant group differences in peak hand speed during early baseline 1 (pain vs control: 123.7 ± 9.9 vs 109.4 ± 13.6 cm·s−1), velocity profiles from both groups converged to similar levels at the end of baseline 1 (113.1 ± 11.0 vs 107.8 ± 15.7 cm·s−1) and remained similar throughout the experiment, with no significant interaction between group and condition (F5,100 = 2.06, P = 0.08). The peak perpendicular error and force adaptation index were also not different between groups (group–condition interaction, both F5,100 < 1.3, P > 0.2). Significant condition–epoch interactions in the ANOVA indicated that movement accuracy was influenced by exposure to the force field (all F5,100 > 2.7, P < 0.02). Post hoc tests revealed no significant differences in the accuracy of movements performed before (baseline 1) and after (baseline 2) the first injection of saline (painful or nonpainful) in the null field (all t66 < 0.8, P = 1). In contrast, upon the first exposure to the force field (after the second saline injection), participants showed greater deviation in hand trajectory relative to a straight path (t66 = 15.3, P < 0.001) than during baseline 1, regardless of pain. In addition, the magnitude of the force adaptation index increased rapidly (t66 = 7.25, P < 0.001), which indicates that participants quickly learned to compensate for the force perturbation. Both the peak perpendicular error and the force adaptation index remained higher than baseline 1 during early Washout (all t66 > 12.9, P < 0.001), but returned to baseline levels during late Washout trials (all t66 < 0.25, P = 1). Upon the second exposure to the force field, both groups showed smaller deviations in hand trajectory than during the first exposure (early force field 2 vs early force field 1, t66 = 5.77, P < 0.001), indicating retention of the capacity to compensate for the perturbation, developed during the first force field exposure, again regardless of the exposure to pain. Estimates of initial learning rates (i.e., exponential rate of increase of the force adaptation index) revealed that the control group adapted significantly more quickly to the force field than the Pain group in 95% of the bootstrap samples during force field 1, whereas during washout 1 this proportion was reduced to 84.5% of the samples. In the second exposure to the force field, between-group differences were smaller (i.e., percentage closer to 50%), with the control group adapting more quickly in 23.9% and 42.6% of the samples during force field 2 and washout 2, respectively.
Each group adapted their motor strategy differently to the saline injection (for details see Figs. 2 and 4). After the first injection, under null field conditions (baseline 2), the Pain group showed lower ADEL EMG than controls (group–condition interaction F5,80 = 3.91, P = 0.003, post hoc t66 = 4.15, P = 0.0005). Group differences were more pronounced during adaptation to the force field. During force field adaptation, the control group showed a 70% increase in ADEL EMG compared with baseline, whereas the Pain group showed a reduction of 4% relative to baseline. Following the second saline injection (i.e., force field 1 and washout 1), the pain group showed lower ADEL, BB, and TRB EMG than controls (group–condition interaction F5,80 > 2.5, P < 0.04, all post hoc t66 > 4.3, P < 0.0004). The ADEL and TRB EMG remained lower in the Pain group during subsequent reexposure and de-adaptation to the same force field, despite the absence of pain (i.e., force field 2 and washout 2, t66 > 4.8, P < 0.0001), but not BB EMG (t66 < 1.2, P = 1). The pain group also showed overall lower PDEL EMG than controls, indicated by a main effect of group in the ANOVA (F1,16 = 5.56, P = 0.03). Post hoc comparisons between conditions confirmed that the strategy of muscle activation learned during the first exposure to the force field (following the second saline injection) was sustained during reexposure (in the absence of pain) for both groups, as no differences were found in average EMG between conditions force field 1 and 2 or between washout 1 and 2 (PDEL: main effect of Condition F5,80 = 29.5, P < 0.001, all post hoc t15 < 0.7, P = 1; other muscles: group–condition interaction, all post hoc t66 < 2.8, P > 0.4).
The results from the wavelet-based LME model, shown in Figure 4, confirm that the Pain group used a motor strategy involving generally lower EMG magnitude than the Control group. Immediately after the first saline injection, i.e., during null field in baseline 2, the positive contrasts indicate that the pain group showed consistently less EMG of the ADEL and PDEL muscles. During the first adaptation to the force field, the control group responded with greater increase in EMG of the shoulder and arm muscles than the pain group. This difference in motor adaptation resulted in large between-group contrasts in EMG envelopes during force field 1 and washout 1, particularly in ADEL EMG, which remained mostly under baseline levels in the pain group. Strikingly, similar group contrasts in ADEL, PDEL, and TRB EMG envelopes were observed during the second exposure to the force field (force field 2 and washout 2), despite complete absence of pain.
Within-group comparisons confirmed that, although different to each other, both groups repeated similar motor strategies between the first and second exposures to the force field (Fig. 5), i.e., between force fields 1 and 2 and washouts 1 and 2—particularly for the shoulder muscles ADEL and PDEL. When reexposed to the force field, the control group demonstrated lower BB EMG (mostly during washout), as well as small adjustments in ADEL, PDEL, and TRB EMG compared with the EMG profiles observed during initial exposure. In contrast, during force field 2, the pain group showed greater BB (and to a lesser extent, also greater TRB and PECM) EMG than during force field 1, but differences appear less pronounced between washout conditions.
The wavelet-based analysis of the cocontraction time series (Fig. 6) revealed some group differences during baseline 1, although differences were small during the last 20 trials. After the first saline injection (baseline 2), positive contrasts indicate that the pain group used consistently less cocontraction of shoulder (ADEL-PDEL) and elbow (BB-TRB) muscles than Controls. These between-group differences were intensified during the first adaptation to the force field (force field 1) and during recovery (washout 1). Similar group differences were observed during reexposure to the force field (force field 2 and washout 2), corroborating with retention of motor strategies after resolution of pain. The time series in Figure 6 also reveal that the group differences in cocontraction of shoulder muscles (ADEL-PDEL) were more pronounced in the first half of the movement, with negligible differences at the endpoint, whereas differences in arm muscles (BB-TRB) were largest in the middle-to-end portion.
Although nociceptive stimulation did not prevent adaptation to novel dynamics in an arm-reaching task, it reduced the initial rate of learning and altered the motor strategy used to compensate for the perturbation. Participants who learned the motor task while experiencing pain developed a distinct strategy of muscle activation, which involved less activity and less cocontraction of shoulder and elbow muscles compared to the pain-free control group, particularly during force field perturbations. A novel observation was that the motor strategy developed during the first exposure to the force field was repeated upon reexposure to the same perturbation, despite the absence of pain. These results suggest that pain changes the motor strategy developed during training of a novel task, and that the same alternative strategy may be used when subsequently training the same task after resolution of pain. Such an effect could be problematic if it applies to learning motor skills in sport and rehabilitation.
Changes in motor strategy during pain
Pain induced by hypertonic saline into the ADEL muscle caused a reduction in ADEL EMG and reduced cocontraction during arm-reaching movements under the null field (baseline 2), during force field adaptation (force field 1) and deadaptation (washout 1) compared with participants who received a nonpainful isotonic saline injection. Similar reductions in activity of the painful muscle have been previously reported during rapid arm movements (in the absence of a force field) following saline injection in ADEL, BB, or TRB (22,23). Consistent with our results, Muceli et al. (23) also reported no changes in movement accuracy after saline injection in ADEL, despite reduced ADEL EMG. Our data indicate that task accuracy was achieved by concomitant reduction in antagonist muscle activity. Extending these findings, our results reveal more substantial reductions in EMG during force field adaptation: the motor strategy used to overcome the perturbation in the presence of pain involved less activity of ADEL, PDEL, BB, and TRB muscles around the arm and shoulder than during pain-free training. Reduced muscle activity during pain is generally believed to be part of a protective adaptation aimed at reducing the stress in the painful area and protecting from further pain (2). Because the peripheral properties of the neuromuscular system are not affected by injections of hypertonic saline (24), reduced EMG has been attributed to reduced central drive to the muscles (7,25). This reduction in descending drive is believed to originate, at least in part, at the cortical level, as acute experimental pain reduces the amplitude of motor-evoked potentials following transcranial magnetic stimulation (26,27), but can increase responses to electrical stimulation of the corticospinal tract, i.e., motoneuron facilitation (28), depending on the muscle.
Reduced EMG during pain also resulted in less cocontraction of the arm and shoulder muscles than the Control group. In principle, muscle cocontraction is metabolically expensive, and hence considering energetic cost alone, reduced cocontraction represents a metabolically more efficient motor strategy. However, in the presence of noise (e.g., unpredictable perturbations) movement accuracy requires antagonist muscle cocontraction to increase endpoint stiffness of the arm during movements (20,29), and this must be balanced together with metabolic cost. During force field adaptation, joint cocontraction plays a particularly important role in the initial stages of learning (i.e., when movement is not yet accurate) and in environments with high instability (e.g., multiple force field directions) by stabilizing the arm, resisting to the force field perturbation, and facilitating the refinement of an internal feedforward model of inverse dynamics (19,30). Analysis of the cocontraction time series revealed distinct effects of pain in each muscle pair (see Fig. 6): group differences in ADEL-PDEL cocontraction were particularly large during hand acceleration, whereas differences in BB-TRB cocontraction were larger during stabilization, after the movement was initiated. In other words, the differences in cocontraction are larger in periods where activation of these muscles is more functionally relevant. In the context of sport training and rehabilitation, the reduction of joint stability may leave the joint prone to injury and potentially contribute to the progression of musculoskeletal problems, such as articular cartilage damage (31,32).
In addition to reduced central drive, pain also interferes with proprioception, modifying the ability of the central nervous system to interpret proprioceptive information and precisely control force or arm position (33). Hence, overestimation of perceived effort (34) may have contributed for the lower levels of muscle activity observed in the Pain group compared to their pain-free counterparts.
Retention of motor strategies after resolution of pain
In the context of motor learning, motor acquisition refers to the error-driven process whereby a movement strategy is developed to gradually perform a task more accurately or faster within a training session (35). The gains in performance reflect changes in spinal and cortical neural networks (36), which can be retained and carried over between sessions (37). This process of motor retention results in faster adaption or further improvements in performance as individuals are repeatedly exposed to the same training environment (37). The new observations of this study were that the Pain group adopted a different motor strategy to learn the task (which involved less activity of arm and shoulder muscles than the Control group) and that the same strategy was retained when the task was repeated after complete resolution of pain. Our data show that, although pain reduced the initial rate of learning, it did not prevent the acquisition or retention of a new motor skill, as both the Pain and Control groups improved movement accuracy similarly during force field 1 and showed faster adaptation during reexposure to the same perturbation (force field 2). These results contrast with an earlier study of locomotion, which showed impaired retention when a painful perturbed walking task was repeated after resolution of pain (10). The differences suggest that the impact of pain depends on the training task and the required adaptation. Compared to arm-reaching, walking is generally considered to have a stronger spinal control component (38), which might be more resistant to top-down learning. Moreover, the locomotor perturbation consisted of brief ankle displacements, whereas a vigorous velocity-dependent force field was used in the present study, and this required more substantial adaptations in muscle activation strategies. Hence, the potentially less challenging adaptation required during locomotor training might explain why participants did not retain the learned motor strategy after resolution of pain in that study.
Other studies have also reported persistent effects of experimental pain after a quick washout period (27,31). Reductions in the intensity of somatosensory- and motor-evoked potentials observed during pain remained suppressed for up to 25 min after resolution of pain (39). Moreover, it has been suggested that when environmental conditions change, the control of muscle coordination does not follow simple rules of optimal control (i.e., to minimize cost functions such as effort), but rather exhibits a tendency to retain habitual patterns previously developed, as long as movement accuracy is “good enough” (40). This idea is corroborated by studies in which the force field perturbation required specific curvilinear hand paths that would minimize energy expenditure (41). Rather than adopting hand trajectories that required less energy, participants continued to move just as they did in the null field. Similar observations were reported during multidirectional isometric wrist contractions: de Rugy et al. (42) modeled wrist joint forces as a summation of EMG recordings from five forearm muscles, and then modified the model to simulate paralysis of a muscle. Instead of using muscle activation strategies that would minimize cost functions such as effort and/or task variability (43), participants simply increased the activity of all muscles to achieve the new goal, presumably at the cost of a dramatic increase in energy expenditure. Taken together, these results support the tendency of the central nervous system to sustain habitual motor strategies, regardless of the presence (or absence) of pain.
In the present study, the observed pain-induced modifications in muscle activity caused only small reductions in the initial rate of learning. Nevertheless, retention of motor strategies that are potentially suboptimal beyond the duration of nociceptive stimulation might have serious consequences for sports training and rehabilitation regimens. For example, an increase of just 1% in baseline knee adduction moment has been associated with a sixfold increase in the risk of radiographic progression of knee osteoarthritis during a 6-yr follow-up (44). However, there is considerable debate on the implications of reduced muscle activation and cocontraction: a single acute episode of pain in the vastus medialis muscle can reduce EMG of both vastus medialis and vastus lateralis muscles and attenuate knee extensor moment during gait (31)—adaptations that reduce joint stability and may contribute to the development and progression of knee osteoarthritis (45). On the other hand, increased duration of cocontraction of medial knee muscles during walking has been associated with faster progression of knee osteoarthritis (46). Detrimental or not, our results suggest that motor strategies learned during painful episodes are not immediately reverted after resolution of pain. Hence, undergoing motor training during pain might result in perpetuation of the same motor strategy after resolution of pain. Further research is necessary to determine whether the retention of motor strategy persists over longer periods, as assessments of immediate retention are not always consistent with observations performed 24 h later (47).
In conclusion, pain did not prevent learning of a novel arm-reaching task, but reduced the initial rate of learning and modified the motor strategy used to perform the task, even though the final state of motor adaptation was unaffected. Remarkably, motor strategies learned in the presence of pain were sustained upon reexposure to the same perturbation, after complete resolution of the painful stimulus. These results suggest that learning in the presence of pain may underpin the development of suboptimal motor strategies and highlight the importance of pain alleviation during motor learning, with critical implications for the design of sports training programs and chronic pain rehabilitation.
This study was funded by a Program Grant (APP1091302), a Senior Principal Research Fellowship (APP1102905) from the National Health and Medical Research Council (NHMRC) of Australia, and a Future Fellowship grant to TC (FT120100391). The authors have no conflicts of interest to declare.
All the authors are properly listed, and all have contributed substantially to the article.
The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation, and statement that results of the present study do not constitute endorsement by ACSM.
1. Ferguson AR, Huie JR, Crown ED, Grau JW. Central nociceptive sensitization vs. spinal cord training
: opposing forms of plasticity that dictate function after complete spinal cord injury. Front Physiol
2. Hodges PW, Tucker KJ. Moving differently in pain: a new theory to explain the adaptation to pain. Pain
. 2011;152(3 Suppl):S90–8.
3. Bank PJ, Peper CE, Marinus J, Beek PJ, van Hilten JJ. Motor consequences of experimentally induced limb pain: a systematic review. Eur J Pain
4. Tucker KJ, Hodges PW. Changes in motor unit recruitment strategy during pain alters force direction. Eur J Pain
5. Salomoni SE, Ejaz A, Laursen AC, Graven-Nielsen T. Variability of three-dimensional forces increase during experimental knee pain. Eur J Appl Physiol
6. Ervilha UF, Farina D, Arendt-Nielsen L, Graven-Nielsen T. Experimental muscle pain changes motor control
strategies in dynamic contractions. Exp Brain Res
7. Salomoni SE, Tucker KJ, Hug F, McPhee M, Hodges PW. Reduced maximal force during acute anterior knee pain is associated with deficits in voluntary muscle activation. PLoS One
8. Boudreau S, Romaniello A, Wang K, Svensson P, Sessle BJ, Arendt-Nielsen L. The effects of intra-oral pain on motor cortex neuroplasticity associated with short-term novel tongue-protrusion training
in humans. Pain
9. Lamothe M, Roy J-S, Bouffard J, Gagné M, Bouyer LJ, Mercier C. Effect of tonic pain on motor acquisition and retention while learning to reach in a force field. PLoS One
10. Bouffard J, Bouyer LJ, Roy J-S, Mercier C. Tonic pain experienced during locomotor training
impairs retention despite normal performance during acquisition. J Neurosci
11. Bouffard J, Bouyer LJ, Roy JS, Mercier C. Pain induced during both the acquisition and retention phases of locomotor adaptation does not interfere with improvements in motor performance. Neural Plast
12. Reis J, Schambra HM, Cohen LG, et al. Noninvasive cortical stimulation enhances motor skill acquisition over multiple days through an effect on consolidation. Proc Natl Acad Sci U S A
13. Smith MA, Ghazizadeh A, Shadmehr R. Interacting adaptive processes with different timescales underlie short-term motor learning
. PLoS Biol
14. Hodges PW, Smeets RJ. Interaction between pain, movement, and physical activity. Clin J Pain
15. Henderson LA, Bandler R, Gandevia SC, Macefield VG. Distinct forebrain activity patterns during deep versus superficial pain. Pain
16. Bouffard J, Salomoni SE, Mercier C, et al. Effect of experimental muscle pain on the acquisition and retention of locomotor adaptation: different motor strategies for a similar performance. J Neurophysiol
17. Howard IS, Ingram JN, Ingram JN, Wolpert DM. A modular planar robotic manipulandum with end-point torque control. J Neurosci Methods
18. Thoroughman KA, Shadmehr R. Electromyographic correlates of learning an internal model of reaching movements. J Neurosci
19. Franklin DW, Osu R, Burdet E, Kawato M, Milner TE. Adaptation to stable and unstable dynamics achieved by combined impedance control and inverse dynamics model. J Neurophysiol
20. Gribble PL, Mullin LI, Cothros N, Mattar A. Role of cocontraction in arm movement
accuracy. J Neurophysiol
21. McKay J, Welch T, Vidakovic B, Ting L. Statistically significant contrasts between EMG waveforms revealed using wavelet-based functional ANOVA. J Neurophysiol
22. Ervilha UF, Arendt-Nielsen L, Duarte M, Graven-Nielsen T. The effect of muscle pain on elbow flexion and coactivation tasks. Exp Brain Res
23. Muceli S, Falla D, Farina D. Reorganization of muscle synergies during multidirectional reaching in the horizontal plane with experimental muscle pain. J Neurophysiol
24. Farina D, Arendt-Nielsen L, Graven-Nielsen T. Experimental muscle pain decreases voluntary EMG activity but does not affect the muscle potential evoked by transcutaneous electrical stimulation. Clin Neurophysiol
25. Graven-Nielsen T, Lund H, Arendt-Nielsen L, Danneskiold-Samsøe B, Bliddal H. Inhibition of maximal voluntary contraction force by experimental muscle pain: a centrally mediated mechanism. Muscle Nerve
26. Le Pera D, Graven-Nielsen T, Valeriani M, et al. Inhibition of motor system excitability at cortical and spinal level by tonic muscle pain. Clin Neurophysiol
27. Schabrun SM, Hodges PW. Muscle pain differentially modulates short interval intracortical inhibition and intracortical facilitation in primary motor cortex. J Pain
28. Martin PG, Weerakkody NS, Gandevia SC, Taylor JL. Group III and IV muscle afferents differentially affect the motor cortex and motoneurones in humans. J Physiol
29. Selen LPJ, Beek PJ, van Dieën JH. Impedance is modulated to meet accuracy demands during goal-directed arm movements. Exp Brain Res
30. Osu R, Franklin DW, Kato H, et al. Short- and long-term changes in joint co-contraction associated with motor learning
as revealed from surface EMG. J Neurophysiol
31. Henriksen M, Alkjaer T, Lund H, et al. Experimental quadriceps muscle pain impairs knee joint control during walking. J Appl Physiol
32. Burr DB, Radin EL. Microfractures and microcracks in subchondral bone: are they relevant to osteoarthrosis? Rheum Dis Clin North Am
33. Wessberg J, Vallbo AB. Coding of pulsatile motor output by human muscle afferents during slow finger movements. J Physiol
34. Proske U, Gregory JE, Morgan DL, Percival P, Weerakkody NS, Canny BJ. Force matching errors following eccentric exercise. Hum Mov Sci
35. Willingham DB, Willingham DB. A neuropsychological theory of motor skill learning. Psychol Rev
36. Merzenich MM, Merzenich MM, Sameshima K, Sameshima K. Cortical plasticity and memory. Curr Opin Neurobiol
37. Martin TA, Martin TA, Keating JG, et al. Throwing while looking through prisms I. Focal olivocerebellar lesions impair adaptation. Brain
38. Côté MP, Murray LM, Knikou M. Spinal control of locomotion: individual neurons, their circuits and functions. Front Physiol
39. Schabrun SM, Burns E, Hodges PW. New insight into the time-course of motor and sensory system changes in pain. PLoS One
40. Loeb GE. Optimal isn’t good enough. Biol Cybern
41. Kistemaker DA, Kistemaker DA, Wong JD, Wong JD, Gribble PL. The central nervous system does not minimize energy cost in arm movements. J Neurophysiol
42. de Rugy A, Loeb GE, Carroll TJ. Muscle coordination is habitual rather than optimal. J Neurosci
43. Diedrichsen J, Shadmehr R, Ivry RB. The coordination of movement: optimal feedback control and beyond. Trends Cogn Sci
44. Miyazaki T, Wada M, Kawahara H, Sato M, Baba H, Shimada S. Dynamic load at baseline can predict radiographic disease progression in medial compartment knee osteoarthritis. Ann Rheum Dis
45. Hortobágyi T, Westerkamp L, Beam S, et al. Altered hamstring-quadriceps muscle balance in patients with knee osteoarthritis. Clin Biomech (Bristol, Avon)
46. Hodges PW, van den Hoorn W, Wrigley TV, et al. Increased duration of co-contraction of medial knee muscles is associated with greater progression of knee osteoarthritis. Man Ther
47. Kantak SS, Winstein CJ. Learning-performance distinction and memory processes for motor skills: a focused review and perspective. Behav Brain Res