Acute Pain Impairs Retention of Locomotor Learning, Regardless of the Context of Retention Testing

Abstract

Our group and others have shown that the presence of an acute painful stimulus may interfere with retention of motor learning. Conversely, other evidence suggests this effect may not be truly due to pain, but due to a change in context when testing retention; i.e., testing retention in a non-painful context when learning occurred in a painful context. Yet to our knowledge, no study has directly compared retention of learning acquired under painful conditions with versus without a context change. To answer this question, we tested 30 young healthy adults on a locomotor learning and retention paradigm. All participants walked on a treadmill with a monitor displaying distorted real-time visual feedback of step lengths to induce learning of an asymmetric stepping pattern. Retention was assessed 24 hours later. Participants were randomized into one of three groups: one received no intervention; one received a painful stimulus during learning on day 1 only; and one received the same painful stimulus during both learning on day 1 and retention testing on day 2. Pain was induced by applying a combination of topical capsaicin cream and superficial heat to the skin of one leg. We found that while all groups successfully learned the asymmetric pattern, retention was reduced in both groups that experienced pain during learning, regardless of the pain context during retention testing. These findings indicate that pain experienced during acquisition of a motor skill has a unique and deleterious effect on retention of that motor skill, which could negatively impact rehabilitation efforts.

Graphical Abstract

NEW & NOTEWORTHY

Here, we show that acute pain experienced during locomotor learning reduces its 24-hour retention regardless of the context in which retention is tested. These findings indicate pain has a deleterious effect on retention of newly acquired motor skills, possibly impacting the efficacy of motor learning-based rehabilitation interventions for people with painful conditions.

INTRODUCTION

Pain is a growing healthcare crisis, with 20.5% of Americans experiencing pain on most, if not every day.¹ Those experiencing pain report an increased difficulty performing activities of daily living, engaging in social activities, and participating in work, resulting in both a high personal cost and economic burden.² Rehabilitation, namely physical therapy, is a major component of a multidisciplinary approach to pain management for these individuals, and 18.6% of those with chronic pain will utilize physical therapy. Given that much of physical therapy is based around principles of motor learning, it is important to better understand the effect pain has on motor learning and its retention.³

Most current literature indicates that experimentally-induced pain likely has little effect on motor skill acquisition.⁴ However, results have been more mixed regarding whether the presence of pain during acquisition impairs retention of recently learned motor behaviors.^5–10 A limitation of many of these studies is the use of experimentally induced movement-evoked pain, rather than tonic cutaneous pain.^7,10–11 Under conditions of movement-evoked pain, individuals frequently alter their movement strategies (consciously or subconsciously) to avoid the pain. The presence of such alterations results in ambiguity as to whether pain truly impairs the magnitude of motor learning and its retention or simply affects the performance during learning, which subsequently appears as reduced retention. To avoid this problem, our group and others have used capsaicin-induced cutaneous pain as a form of non-movement-evoked pain to examine the impact of pain itself (and not movement avoidance or altered movement strategies) on one’s ability to learn and retain new locomotor patterns.^5,12 In these studies, acute experimental pain experienced during locomotor learning appears to impair retention at a 24-hr follow-up test.

Conversely, other evidence⁶ suggests this effect may not be truly due to pain, but due to a change in context when testing retention. Specifically, both Bouffard et al. (2014) and our group (Galgiani et al. 2024) tested retention in a nonpainful context when participants initially learned in a painful context. In their 2016 follow-up study, Bouffard et al. found that a group receiving a painful stimulus during learning and retention testing showed no reduction in retention compared to a group receiving no painful stimulus during learning and retention. While intriguing, this study did not definitively test whether the prior reported reductions in retention might be due to a change in context from learning to retention tests because neither group experienced a change in context.

Therefore, the goal of the current study was to directly compare the effect of pain on locomotor learning and retention with versus without a change in context during retention testing. We compared the acquisition and retention of a novel step length asymmetry pattern using three groups of healthy, young adults: a no stimulus group, a group who received a painful stimulus during initial learning but not retention testing, and a group who received the same painful stimulus during initial learning and retention testing. If pain did not really impair retention, i.e., the reduced retention previously reported during pain-free retention testing was merely an effect of context change, then the group that received the painful stimulus during both learning and retention should show intact retention. On the other hand, if pain did impair retention, then the group that received the painful stimulus during both learning and retention should show an impairment in retention similar to that of the group that experienced pain only during learning. We hypothesized that pain experienced during learning would impair retention of the novel step length asymmetry pattern, regardless of whether pain was, or was not, experienced during retention testing.

METHODS

Participants

Thirty young, healthy adults (15 female, 15 male) were recruited from the University of Delaware community. Enrollment criteria were: age 18–35 years, absence of acute or chronic pain, not taking analgesics, no musculoskeletal or neurological conditions, naïve to the experimental paradigm, and willingness to receive the experimentally induced pain. Additionally, participants were excluded if they had any condition that could result in an adverse reaction to experimental pain, including decreased sensation or circulation in the area targeted for the pain stimulus, any skin lesion in the area, or skin allergies. Participants were stratified by sex and then randomized into one of three groups, no stimulus (NO PAIN; n=10, female=4; 22.9 ± 3.9 years, mean age ± 1 SD), pain during learning (PAIN D1; n=10, female=5; 24.3 ± 3.7 years), or pain during learning and retention (PAIN D1+D2; n=10, female=6; 22.5 ± 1.7 years). All participants provided informed consent and this study was approved by the University of Delaware Institutional Review Board.

Experimental Design

Participants walked on a dual-belt treadmill (with belt speeds tied) at a self-selected comfortable, but brisk pace between 1.0 and 1.2 m/s. Treadmill speed was selected by each participant starting at 1.0 m/s and incrementally increasing the speed, if desired by the participant, until they indicated they had reached a comfortable speed. The fastest allowable treadmill speed was 1.2 m/s, which we implemented to ensure similar numbers of steps for all participants. Importantly, treadmill speeds across groups were not statistically different (F(2,27) = 0.22, p = 0.81): NO PAIN (1.11 ± 0.03 m/s, mean ± SD); PAIN D1 (1.12 ± 0.08 m/s); PAIN D1+D2 (1.10 ± 0.08 m/s). Participants wore a ceiling-mounted harness, which did not provide any body-weight support, and lightly held onto a handrail for safety (Fig. 1A). On day 1, participants learned a new asymmetric walking pattern driven by distorted visual feedback of their step lengths.^12,13 On day 2 (24 hours later), participants were tested for retention of this asymmetric walking pattern.

Figure 1.

(A) Treadmill set-up: participants walked on a treadmill with a ceiling-mounted harness and visual feedback was displayed in real-time on screen in front of them. (B) Visual feedback: depictions of what the screen displayed at various key epochs during the walking paradigm. The blue bar represents the left leg; the green bar represents the right leg; the pink horizontal line represents the target line. Each panel shows the display at the moment of heel strike of one of the legs. In the middle and right panels, the open bars with dashed lines represent the actual step lengths taken (not visible to participants), while the solid blue and green bars show the step lengths with the gain in place, visible to participants (right leg appearing to take a 9% longer step; left leg appearing to take a 9% shorter step). The left panel shows veridical feedback as it typically appeared during baseline orientation. The middle panel shows the distorted visual feedback as it typically appeared during early learning, with the distortion indicating inaccurate step lengths. The right panel shows the distorted visual feedback as it typically appeared during late learning, with the distortion indicating accurate step lengths. (C) Experimental timeline: Time course of the induced step length asymmetry via distorted visual feedback (thick blue line) and duration of each phase. Gray shading indicates phases with no visual feedback. Below is the time course of experimental pain application with ramp-up of pain intensity during baseline and tonic pain intensity during the remainder of day 1, and/or day 2, depending on group assignment. Abbreviations: Bsl, Baseline; min, minutes; s, seconds.

Visual feedback display

All participants viewed the same real-time visual feedback of their step lengths, displayed on a computer monitor in front of the treadmill (The Motion Monitor Toolbox, Innovative Sports Training Inc., Chicago, IL). Feedback consisted of a dynamic graph with two vertical bars, side-by-side, representing participants’ instantaneous left and right step lengths during the swing phase of gait. At the initiation of swing, the height of the bar would grow vertically from the bottom of the screen until heel strike, at which point the height of the bar would hold constant until the next step with that leg began. This was provided for both legs so participants would receive feedback of both legs’ step lengths continuously as they walked. A baseline step length average was calculated and represented as a horizontal “target line”, individualized to each participant, on the screen. When the target line was visible, participants were instructed to “match the target line as best you can” for each side and for each correct target hit, a word appeared on the screen (i.e. “Great!”) to alert participants of success. During learning and retention phases, the visual feedback of the vertical step length bars was distorted to drive an asymmetric stepping pattern. A gain was added to each bar, displaying a 9% shorter step length on the left side and a 9% longer step length on the right side. Therefore, to successfully match the target line, they needed to counter this visual perturbation by adopting a 9% longer left step length and a 9% shorter right step length (Fig. 1B). We chose to use a 9% gain because in our prior work we have identified that asymmetries below 7–8% risk being too small to distinguish from normal variability and asymmetries much above ~11% become noticeably more difficult to execute, which we wanted to avoid.^12,14 The target line had a thickness of 1.2% of each participant’s baseline step length, so that every heel strike that produced a step length within ±0.6% of the target step length was considered a target hit.

Experimental schedule

Across day 1 and 2, participants in all groups walked under the same three blocks: baseline, learning, and retention (Fig. 1C). The baseline block was broken into three different phases: baseline 1, orientation, and baseline 2. During baseline 1 (1 min), participants were asked to walk normally and no visual feedback was provided. Next, participants were oriented (2 min) to the visual display by viewing veridical feedback and were encouraged to take shorter and longer steps to ensure understanding of the scaling of the feedback. In baseline 2 (1 min), participants were asked again to walk normally with no visual feedback.

The learning block was broken into three different phases: learning, catch, and practice. Prior to learning, participants were informed that the visual feedback “may not be entirely accurate during this phase” and their task was to “hit the target line as best you can with each leg.” During learning (10 min), the visual distortion was gradually applied starting at 1% and increasing by 1% each minute until the full 9% asymmetry was reached. This level of asymmetry was held constant for the final 2 minutes of learning. Between learning and practice, a catch phase (30 sec) was performed where the visual feedback was turned off and participants were asked to “look ahead and walk normally.” The purpose of the catch phase was to assess the magnitude of implicit learning from the prior learning phase. That is, when instructed to return to normal walking, any residual asymmetry that was greater than baseline asymmetry could be assumed to be a consequence of an implicit learning process.^13,14 In the practice phase (5 min), the full 9% distorted visual feedback was restored, and participants were again asked to “hit the target line as best you can with each leg.” The end of the practice phase marked the completion of day 1. Participants were asked to avoid using a treadmill between day 1 and day 2.

About 24 hours later (24 ± 4 hrs), participants returned to the lab for retention testing. The retention block consisted of a single 5-minute phase identical to the practice phase of day 1 in which the 9% distorted visual feedback was displayed and participants were instructed to “hit the target line as best you can with each leg.”

Pain stimulus

Participants in the PAIN D1 and PAIN D1+D2 groups received the same harmless but painful stimulus induced by combining capsaicin cream and superficial heat. Capsaicin and heat induced pain was selected as it is a robust model of tonic, cutaneous pain that is not movement-evoked^15,16 and therefore unlikely to alter motor performance. Prior to baseline walking, a 0.1% topical preparation of capsaicin was applied to the anterior right lower shank, just inferior to the pes anserine, covering approximately a 100 cm² area. Disposable, air-activated heating pads were applied over the capsaicin-treated skin and secured in place with Cover-Roll stretch tape (3M, St. Paul, MN). Baseline walking was begun immediately following application of the painful stimulus as the baseline block was completed within 5 minutes and the “ramp-up” period of the painful stimulus occurs over 10–20 minutes, in accordance with rising skin temperature (Fig. 1C). We have also shown that, as expected, application of this exact painful stimulus does not interfere with motor performance of locomotion.¹² Following the practice phase of day 1, the heat wrap was removed and the treated skin was washed thoroughly with soap and cool water until pain was abolished. Those in the PAIN D1+D2 group (but not those in the PAIN D1 group) received this same painful stimulus again on day 2 and retention testing was performed once pain had increased to the same intensity (as assessed by the Defense and Veterans Pain Rating Scale, described below) as each participant’s average pain intensity during the previous day’s learning phase. Participants in the NO PAIN group did not receive the painful capsaicin and heat stimulus at any point during day 1 or day 2 (Fig. 1C).

Data Collection

Motion capture

Kinetic data were collected by force plates (Bertec, Columbus, OH) imbedded under each belt of the dual-belt instrumented treadmill, while kinematic data was collected by an eight-camera Vicon MX40 motion capture system with Nexus software (Vicon Motion Systems, Inc., Centennial, CO). Seven reflective markers, one on each heel, lateral malleolus, and fifth metatarsal head and one on the left first metatarsal, were used to capture kinematic data. Kinetic data, sampled at 1,000 Hz, and kinematic data, sampled at 100 Hz, were time-synchronized in Vicon Nexus and we used The Motion Monitor (Innovative Sports Training Inc., Chicago, IL) to provide the real-time step length visual feedback to participants.

Pain measures

Pain levels were assessed regularly during baseline, learning, and retention walking blocks, regardless of group assignment. The Defense and Veterans Pain Rating Scale (DVPRS) was used to assess pain level as this scale combines a numeric 0–10 pain rating scale, cartoon facial expressions, and written descriptions of each level.¹⁷ Prior to application of the painful stimulus, participants were oriented to the DVPRS and expressed understanding. Pain level was assessed every 3–5 minutes during the “ramp-up” period and every 5 minutes once a tonic level of pain was reached, defined as no change in pain level over the previous three pain measures. Pain level was assessed while participants continued the locomotor task. For participants in the PAIN D1+D2 group, the pain stimulus was applied again at the beginning of day 2 and for this group only, the retention block was not begun until each participant reached their average pain intensity from the prior day during the learning block. Specifically, for each participant in the PAIN D1+D2 group, we waited until their pain reached their average pain on day 1 (averaged from measures at the beginning of the learning phase, the end of the learning phase, and the end of the practice phase). Once this individualized pain intensity was met, retention testing was begun.

Data Analyses

Kinematic and kinetic data were analyzed with custom-written MATLAB code (MathWorks, Natick, MA). Kinematics were low-pass filtered at 10 Hz using a fourth-order Butterworth filter. The timing of heel strikes for each stride were identified as the time when vertical ground reaction forces first exceeded 20 Nm on the corresponding belt after the anterior-posterior velocity of the heel marker on that side first changed direction. Step lengths were then determined by the anterior-posterior distance between left and right heel markers at the time of heel strike. Our primary outcome measure was step-length asymmetry (SLA), which has been used previously to quantify learning in this locomotor paradigm:^{13,14,18–20}

$$\mathit{\text{SLA}}(\%)=\frac{({\mathit{\text{Step}}\mathit{\text{Length}}}_{\mathit{\text{Left}}}-{\mathit{\text{Step}}\mathit{\text{Length}}}_{\mathit{\text{Right}}})}{({\mathit{\text{Step}}\mathit{\text{Length}}}_{\mathit{\text{Left}}}+{\mathit{\text{Step}}\mathit{\text{Length}}}_{\mathit{\text{Right}}})}\times 100$$

Thus, a SLA of 0% would represent perfect symmetrical step lengths whereas a SLA of +9% would represent perfect learning of the desired asymmetry in this paradigm.

Our analysis focused on SLA values at three key epochs: late baseline, late practice, and early retention, defined as the average SLA for all strides of the baseline 2 phase, the last 25 strides of the practice phase, and the first 4 strides of the retention phase. These epochs were identified in our prior work,¹² and represent baseline asymmetry, learning magnitude, and retention magnitude, respectively. We further quantified retention through a forgetting index, calculated by subtracting early retention from late practice for each participant. Finally, we also assessed the degree to which learning was achieved implicitly (as opposed to explicitly) by calculating the percent of implicit learning, or the ratio of the average SLA during the catch trial to the average SLA during late learning (last 25 strides of the learning phase) and expressing as a percentage.

Statistical analyses

All statistical analyses were performed in SPSS software (Microsoft, Chicago, IL). Average pain level during the learning block was compared between the PAIN D1 and PAIN D1+D2 groups via two-sample Student’s t test. For the PAIN D1+D2 group, we also conducted a paired Student’s t test comparing average pain during the learning block (day 1) to average pain during the retention block (day 2) to determine if pain intensity differed among participants from day 1 to day 2. For our main analysis, which compared learning of the novel asymmetric walking pattern and its retention, a 3 × 3 marginal linear mixed model was performed across the three groups for the three epochs of interest, late baseline, late practice, and early retention. An autoregressive with heterogenous variances, ARH(1), covariance matrix was selected to model repeated measures as it resulted in the best fit via Akaike Information Criteria and Bayesian Information Criteria. Normality of residuals was met for each cell of the study design and no outliers were present. Forgetting indices and the percent implicit learning were compared via separate one-way between-subjects ANOVAs, for which assumptions of independence, normality of residuals, and homogeneity of variance were met. Any significant interactions found in the marginal linear mixed model were followed up by comparisons of simple main effects. Significant main effects in the one-way ANOVAs were followed up with pairwise comparisons. Bonferroni corrections were used, where applicable and an alpha value of 0.05 was set for all comparisons.

RESULTS

Pain Levels were Consistent for PAIN D1 and PAIN D1+D2 Groups

All participants in the NO PAIN group reported a pain level of 0 throughout the entire locomotor paradigm. Figure 2 shows pain intensity ratings among the two painful groups. Average pain intensity during the learning block did not differ between the PAIN D1 and PAIN D1+D2 groups (t(18)=−1.2, p=0.25). Additionally, pain intensity for the PAIN D1+D2 group did not differ between day 1 and day 2 (t(9)=−1.3, p=0.22). On average, reported pain levels were 4.6 ± 0.58 for participants in the PAIN D1 group and 5.1 ± 0.43 for participants in the PAIN D1+D2 group at the end of the learning block on day 1. No participants reached our cutoff pain intensity of 8/10 and average pain intensity during the learning block ranged from 2.3 – 6.7.

Figure 2.

Average pain intensity ratings (using the 0–10 numeric pain rating scale) on day 1 for PAIN D1 and PAIN D1+D2 groups and day 2 for the PAIN D1+D2 group. Timepoints noted above correspond with the following minutes of the walking paradigm: Early Baseline, 0 minutes; Late Baseline, 4 minutes; Early Learning, 5 minutes; Late Practice, 20 minutes; Early Retention, 0 minutes (day 2); Late Retention, 5 minutes (day 2). Error bars, ± 1 SEM.

24-hour Retention, but not Learning, Was Reduced in both Groups with Pain

Figure 3 shows stride-by-stride group averaged SLA data from baseline, learning and retention blocks on days 1 and 2. All three groups showed low SLA during baseline, as expected, and gradual acquisition of the desired 9% SLA, also as expected. However, during early retention on day 2, the NO PAIN group showed good retention; i.e., SLA values were close to where they were at the end of the previous day, whereas for the PAIN D1 and PAIN D1+D2 groups, this was not the case (see Fig. 3 inset).

Figure 3.

Group average step length asymmetries (SLA) shown on a stride-by-stride basis for baseline, learning, catch, practice, and retention phases for all three groups. Data are binned in groups of 3 strides and truncated here to the length of the individual with the least number of strides for each phase. Gray shading indicates phases with no visual feedback. The dashed line indicates the 9% SLA target for the end of learning and all of practice and retention. The dotted line indicates 0 asymmetry. Shading, ± 1 SEM. The inset shows a zoomed in view of the first 3 bins of the retention phase. Abbreviations: Bsl, Baseline.

The 3 × 3 marginal linear mixed model confirmed this finding. Figure 4 shows the average SLAs for all 3 groups across the 3 key epochs of late baseline, late practice, and early retention. There was a significant group × time interaction effect (F(4,32.7)=2.97, p=0.03) with planned poc-hoc testing of simple main effects revealing no statistical difference in SLA among groups at late baseline (F(2,26.1)=0.48, p=0.63) or late practice epochs (F(2,27)=0.54, p=0.59), but a significant difference at the retention epoch (F(2,28.6)=5.41, p=0.01). At early retention, SLA for the NO PAIN group (mean SLA, 7.1 ± 0.9%) was statistically greater compared to both the PAIN D1 group (3.4 ± 0.9%, p=0.02) and PAIN D1+D2 group (3.6 ± 0.9%, p=0.03). Moreover, the two painful groups did not differ from each other at the retention epoch (p>0.99).

Figure 4.

Group average step length asymmetries (SLA) at key epochs of baseline, late practice (representing learning magnitude), and early retention (representing retention magnitude). Error bars, ± 1 SEM. P values refer to the post-hoc pairwise comparisons between NO PAIN and PAIN D1 groups, and between NO PAIN and PAIN D1+D2 groups. Asterisks indicate statistical significance.

Groups did not differ in the amount learned at the end of the learning phase (F(2,26.3) = 1.12, p = 0.34), nor did they differ in the percent of implicit learning (NO PAIN: 25.85%, PAIN D1: 23.80%, PAIN D1+D2: 24.24%) (F(2,26.3) = 0.02, p = 0.98). These analyses provide further evidence that the groups did not learn differently from one another.

Painful Groups, Regardless of Context, Had Increased Forgetting after 24 Hours

Additionally, these results were consistent with our forgetting index analysis, shown in Figure 5. The one-way ANOVA indicated a significant difference among groups (F(2, 27)=5.74, p<0.01). Post-hoc pairwise comparisons revealed that forgetting of the learned SLA from day 1 to day 2 was significantly less for the NO PAIN group (mean forgetting, 1.1 ± 0.9% SLA) compared to the PAIN D1 group (5.1 ± 0.9% SLA, p<0.01) and approached significance for the NO PAIN group compared to the PAIN D1+D2 group (4.0 ± 0.9% SLA, p=0.07).

Figure 5.

Group average forgetting indices. Error bars, ± 1 SEM. P values refer to the post hoc pairwise comparisons between NO PAIN and PAIN D1 groups, and between NO PAIN and PAIN D1+D2 groups. Asterisk indicates statistical significance.

DISCUSSION

We examined whether an acute pain stimulus has a unique and negative effect on retention of locomotor learning, as suggested by Bouffard et al. (2014) and Galgiani et al. (2024), or whether the prior studies instead merely reflected context-specific learning and a lack of transfer from a painful context to a non-painful context.⁶ We found that groups experiencing pain during motor learning had significantly impaired retention at 24-hours compared to the no stimulus group, regardless of the context in which retention was tested (painful or non-painful). To the best of our knowledge, this is the first time that pain, motor learning, and the context in which retention is tested has been examined using three separate groups. Our design allowed us to compare learning and retention in two ways: first, between two groups that experienced acute pain during learning but had different contexts during retention testing; and second, between two groups that had no change in context across days, with one group experiencing acute pain and the other not.

Importantly, our painful stimulus of capsaicin and superficial heat produced a stable and moderate level of pain that did not differ between painful groups. Moreover, the pain intensity experienced by the PAIN D1+D2 group did not differ between day 1 and day 2, ensuring the same painful context across days. Similar to other studies that used capsaicin as a painful stimulus,^5,6,12 we did not see any difference in baseline stepping behavior or initial learning between groups who either did or did not experience pain. The use of a painful stimulus that does not produce a movement-evoked pain, and therefore does not alter movement strategy, likely contributed to this phenomenon where pain has no effect on acquisition. Taken together, the findings of prior studies and the current study suggest that capsaicin-induced pain does not alter locomotor movement strategies and does not disrupt acquisition of a novel locomotor task, but does reliably impair retention of locomotor learning.

The key finding of our current study is that experiencing acute pain during acquisition of a novel locomotor pattern impairs 24-hour retention, regardless of the context in which retention is tested. This is evident from our main analysis, with a significant interaction of group × time and significant differences in post-hoc testing between the NO PAIN and PAIN D1 groups and between the NO PAIN and PAIN D1+D2 groups during the early retention epoch. When measured by the forgetting index, there was a significant difference between the NO PAIN and the PAIN D1 groups, although the comparison between the NO PAIN and the PAIN D1+D2 groups did not reach statistical significance (post hoc, p=0.07). These findings differed from those of Bouffard et al. (2016), who found no retention deficits between two groups: a non-painful control group and a group who experienced pain during both learning and retention testing. There are several differences between this paradigm and ours that may explain these conflicting results, including the specific mechanism driving the learning (mechanical versus visual), the duration of the learning period (5 versus 10 minutes), and the presence or absence of a washout period immediately after learning. We offer here a few reasons why we think the current study represents an advancement from prior work. First, we included three separate groups allowing for direct comparisons between groups who experienced pain during learning, but had altered pain context for retention testing; and for comparison between groups who had no change in context across days, but either did or did not receive the painful stimulus. Second, our motor learning paradigm is well-characterized and has large effect sizes (i.e., large learning and retention magnitudes), allowing room for changes in learning and retention to be detectable. In contrast, the Bouffard et al. paradigm^5,6 has very small learning and retention effects (<1° ankle excursion difference). Finally, here we used the identical paradigm to our prior study¹² of pain and locomotor learning and found the same results (for the two groups that were included in that prior work, NO PAIN and PAIN D1). We think this is an important demonstration of the reproducibility and robustness of the effects of acute pain on retention of locomotor learning.

Of note, the retention deficit observed in our study is transient, occurring over the first 4 steps of the retention epoch. To further examine this, we compared average SLAs across strides 5–9 post hoc and found no differences among groups (F(2,28.1)=0.55, p=0.58). This is not surprising, in that a quick rate of relearning is expected in such an explicit motor learning task,¹³ in which participants are provided with continuous visual cues. Nonetheless, this explicit learning task was purposefully selected for this experiment to closely mimic the conditions founds in rehabilitation settings, where clinicians heavily utilize explicit cueing. We also noticed that the PAIN D1+D2 group appeared to have a somewhat lower SLA throughout the retention block, so we conducted a post hoc analysis of the average SLA during late retention. This was also non-significant (F(2,28.4)=0.63, p=0.54).

The exact mechanism for this retention deficit as a result of pain exposure during learning is unclear and the current study was not designed to answer this question. However, previous literature does support a role for acute pain in disrupting motor memory consolidation, which is a function that has been attributed to motor cortical brain regions. For example, acquisition of a new motor skill is associated with increased excitability within the primary motor cortex (M1).^21,22 Additionally, artificially increasing M1 excitability via non-invasive brain stimulation has been shown to improve consolidation of motor memories, whereas reducing M1 excitability has been shown to impair consolidation.^23–25 Perception of pain has been shown to have direct input to M1 by intracortical recordings²⁶ and chronic pain is associated with altered M1 excitability.²⁷ Further, an acute pain stimulus, and specifically the experimental pain used in this study, has been shown to reduce the excitability of M1.^28,29 Putting these ideas together, we therefore postulate that acute pain may decrease M1 excitability and thereby reduce consolidation of newly learned motor behaviors. Note that although locomotion is commonly viewed as mediated predominantly by brainstem and spinal circuits, adapting walking patterns to novel situations and especially when using visual signals, as is the case in the current paradigm, has been shown to rely on motor cortical brain regions, including M1.^30,31 Further mechanistic work should examine the relationship between this motor learning retention deficit and the degree of reduction in M1 excitability when exposed to acute experimental pain.

Conclusion

Our findings add to a growing body of literature investigating the potentially negative effects of pain on motor learning, and specifically its retention. One limitation of this body of work has been the potential confound of performing learning paradigms in a painful context while testing for retention in a pain-free context. Another issue of concern in the literature is that using musculoskeletal experimental pain models introduces the confound of movement-evoked pain. Here, we addressed both these issues in a well-controlled study and found that experiencing a painful stimulus during acquisition of a novel locomotor task resulted in reduced retention 24-hours later, regardless of whether retention was assessed in a painful or pain-free context. This finding has potentially important implications for rehabilitation clinicians, as motor learning-based interventions may result in impaired retention for those experiencing pain. While this work has implications for patients experiencing acute pain, future work should focus on whether this effect is also observed in clinical populations with chronic pain, especially older adults, who experience chronic pain at a much higher prevalence than young adults.

Data Availability:

Data will be made available upon reasonable request.