Journal Logo

APPLIED SCIENCES

High-Intensity Aerobic Exercise Enhances Motor Memory Retrieval

MANG, CAMERON S.; SNOW, NICHOLAS J.; WADDEN, KATIE P.; CAMPBELL, KRISTIN L.; BOYD, LARA A.

Author Information
Medicine & Science in Sports & Exercise: December 2016 - Volume 48 - Issue 12 - p 2477-2486
doi: 10.1249/MSS.0000000000001040
  • Free

Abstract

There is a growing body of evidence demonstrating that aerobic exercise can benefit memory processes (19,29). Interestingly, a recent meta-analysis concluded that the acute effects of a single bout of aerobic exercise on memory were larger than the chronic effects associated with aerobic exercise training over several weeks or months (29). Further, acute aerobic exercise effects were greatest in studies examining long-term memory processes (i.e., 2 min and up to 8 months after exposure) (29). However, the majority of this evidence was obtained from the evaluation of declarative learning processes, such as verbal/vocabulary learning and image recall tests (19,29). Less is known about the effects of acute aerobic exercise on motor learning.

Procedural motor learning is supported by memory systems that are unique from those involved in declarative tasks (21,36) but that are also specific to the nature of the motor task being learned (10,13). Two studies demonstrated the positive effects of a single bout of high-intensity aerobic exercise on motor learning, and notably, both used highly similar motor tasks (22,30). In these studies, high-intensity aerobic exercise, when paired closely in time to practice of a continuous motor sequence, reduced performance error at the end of skill acquisition (22) and at “no-exercise” retention tests conducted 24 h (22,30) and 7 d after practice (30). In the study conducted in our laboratory, the aerobic exercise effect was specific to improvements in temporal precision (22), a component of continuous motor sequence learning that is linked to cerebellar function (6). In addition, the tasks used in both studies involved a visuomotor rotation (22,30), a form of motor learning for which the cerebellum is implicated (4). Thus, evidence from these two studies (22,30) suggests that acute high-intensity exercise can benefit motor learning in terms of continuous motor sequence tasks and outcome measures that are highly cerebellum dependent.

In related work (28), the effect of acute high-intensity aerobic exercise on motor memory interference was examined using a discrete motor sequence task akin to the serial reaction time task. Performing moderate-to-vigorous aerobic exercise 2 h after practice of a motor sequence, but immediately before practice of an alternative sequence, reduced interference effects on the learning of the original sequence, assessed by a 24-h retention test (28). Although this result suggests a benefit of acute aerobic exercise on discrete motor sequence memory consolidation, the results partially contradict other findings (22,30) showing that aerobic exercise before motor practice enhanced continuous motor sequence acquisition and learning. For instance, if aerobic exercise before practice of the alternative sequence benefited learning, then a greater interference effect on the learning of the original sequence might have been expected. In another study, conducted by Statton et al. (37), acute moderate-intensity aerobic exercise immediately enhanced performance of a discrete motor sequence task (i.e., sequential visual isometric pinch task) but had no effect on task learning (37). The lack of a learning effect in this study (37) could potentially relate to the use of a moderate-intensity exercise bout; for example, we recently found that acute moderate-intensity aerobic exercise enhanced motor performance but not learning (35) of the same continuous motor sequence task for which we previously found learning benefits associated with high-intensity aerobic exercise (22). Nevertheless, given the findings of the motor memory interference study, which used high-intensity exercise (28), another potentially important consideration relates to differences in the principles that guide sequence learning of tasks that involve discrete versus continuous movements (10,20,33).

Discrete movements are characterized by a definite beginning and end and may be combined into a sequence or series, known as a serial movement (33). A well-studied feature of discrete motor sequence learning is the “chunking” of individual movements into subsequences that are then linked together (31). This chunking process is dependent on the function of the basal ganglia (5,15) and is not used when learning continuous movement sequences (10), such as in the previous studies of acute high-intensity aerobic exercise and motor learning (22,30). Consistent with the apparent difference in how sequences are learned, practice conditions can have disparate effects depending on whether a task is of a continuous or discrete nature (20,33). Thus, although our previous work indicated a benefit of acute high-intensity aerobic exercise for the temporal aspect of motor learning in a task that depended heavily on cerebellar function (22), the current literature does not address whether the positive effects of acute high-intensity aerobic exercise generalize to learning of sequences of discrete movements.

The main objective of the present work was to evaluate the effects of acute high-intensity aerobic exercise performed immediately before motor practice on the learning of a discrete motor sequence task, without visuomotor rotation. Although previous work demonstrated acute exercise-induced learning benefits on comparatively cerebellar-mediated tasks (22,30), learning of the task used in the present work likely involved a lesser contribution of cerebellar, relative to basal ganglia, circuits (5,15). Although the specific neurobiological mechanisms for acute aerobic exercise effects on human memory are not conclusively known, evidence from animal studies suggested that exercise-induced increases in neurochemicals (12,38), including dopaminergic effects in the basal ganglia (26), may play a role in facilitating memory processes. Given evidence that such neurochemical increases are elicited to a greater extent by higher-intensity exercise (40), it is plausible that the use of moderate-intensity aerobic exercise contributed to the lack of an exercise effect on discrete motor sequence learning in previous work (37). Thus, here we hypothesized that acute high-intensity aerobic exercise would benefit the acquisition and learning of a sequence of discrete movements. We examined change in performance over the practice period, which may reflect memory encoding (16). We also assessed change in performance from practice to a delayed (24 h) no-exercise retention test to evaluate motor learning after the evolution of consolidation processes (16). In line with previous evidence suggesting exercise effects on encoding (22) and consolidation (22,30), we expected positive effects of acute high-intensity aerobic exercise on these measures. In addition, we evaluated the rate of performance improvement during practice and during the retention test, which may reflect the speed of memory encoding and the rate of memory retrieval or relearning (i.e., “savings”), respectively (16,18). We anticipated that rates of skill improvement at both time points would be enhanced in the aerobic exercise condition, possibly due to exercise effects on the speed of information processing (19), which could facilitate discrete motor sequence learning and relearning.

METHODS

Participants

Six men and 10 women between ages 19 and 34 yr (mean ± SD = 25.7 ± 4.6 yr) participated in the present experiment. Participants had no known neurological diagnoses and were of adequate health to complete the exercise protocols. All participants provided written informed consent before testing. The Clinical Research Ethics Board at the University of British Columbia approved all testing procedures.

Experimental Overview

On a separate day, at least 2 d before all other experimental sessions, participants completed a graded maximal exercise test on a cycle ergometer. Each individual then participated in four sessions in a within-subjects experimental design to compare the effects of a 20-min period of rest and a 20-min bout of high-intensity interval cycling on the acquisition and learning of a discrete motor sequence task. The four experimental sessions were completed on separate days and included the following: 1) rest immediately before skilled motor practice, followed by 2) a no-exercise 24-h retention test, and 3) aerobic exercise immediately before motor practice, followed by 4) a no-exercise 24-h retention test. On all testing days, participants were instructed to refrain from any exercise other than that involved in the experimental sessions. There was a minimum washout period of 2 wk between motor practice under the different experimental conditions (i.e., rest or aerobic exercise). Participants were randomly allocated to an experimental order such that half of the sample completed the rest condition first and the other half the aerobic exercise condition. The experimental procedures are depicted in Figure 1.

FIGURE 1
FIGURE 1:
Experimental overview.

Graded Maximal Exercise Testing

A maximal exercise test was conducted on a cycle ergometer (Ergoselect 200; Ergoline GmbH, Bitz, Germany), beginning with a power output of 100 W for men and 50 W for women, and was increased by 30 W increments every 2 min until volitional exhaustion. Participants were instructed to maintain a pedaling cadence of 70–90 rpm and to remain seated throughout testing. The following measurements were monitored throughout exercise testing: expired O2 and CO2 concentrations and airflow via a metabolic cart (ParvoMedicsTrueOne 2400, Sandy, UT), HR via a HR monitor (Polar Electro, Oy, Kempele, Finland), and Borg’s 6–20 scale RPE. Finger-stick blood lactate (BLa) was determined immediately after the exercise test using an automated portable BLa analyzer and test strips (Lactate Pro; Arkray Inc., Kyoto, Japan). Peak O2 consumption (V˙O2peak) criteria included at least one of the following: a plateau in O2 uptake (V˙O2) and HR with further increase in workload, an RER greater than 1.1, an RPE greater than 17, a BLa greater than 10 mmol·L−1, an inability to maintain a cadence of 70 rpm, and volitional exhaustion. Exercise testing results for each individual are presented in Table 1.

TABLE 1
TABLE 1:
Participant characteristics.

Standardized Acute Aerobic Exercise Bout

Peak power output determined by the exercise test was used to inform the prescription of a standardized acute aerobic exercise bout in the same manner as our previous work (22). The bout lasted 20 min and included a 5-min warm-up at 50 W and self-selected cadence, followed by three sets of 3-min high-intensity cycling interspersed with 2 min of low-intensity cycling. The high-intensity intervals consisted of cycling at 90% of peak power output from the final fully completed stage of the graded maximal exercise test, and the low-intensity intervals involved cycling at 50 W, always maintaining a cadence greater than 70 rpm. The high-intensity exercise bout was used based on previous work, indicating a dose–response relationship between exercise intensity and increases in neurochemicals (17,40). Aerobic exercise prescriptions and physiological responses of each participant are presented in Table 1.

Serial Targeting Task Procedures

Practice of a discrete motor sequence task, herein called the serial targeting task, was completed after the rest and aerobic exercise conditions for each participant, with at least 2 wk between conditions. Serial targeting task practice involved manipulation of a computer mouse (Wheel Mouse Optical; Microsoft Corporation, Redmond, WA) housed in a custom frame that was held in a pronated grasp with the nondominant hand (Fig. 2). The mouse was used to move a cross-hair cursor between a series of discretely presented targets. To initiate the appearance of the next target, participants placed their cursor in the current target for 500 ms; this time was followed by an additional 500 ms intertarget interval. Each target could appear at one of nine locations. One location was central; the other eight locations formed an equidistant circular array. Cursor position sampling was presented at 200 Hz with custom software developed on Labview (v.8.1; National Instruments Corporation, Austin, TX). The target movement response time was defined as the time in seconds from target appearance to the presentation of the next target (corrected for the 500-ms stationary period and the 500-ms intertarget interval) and was extracted using custom Labview software (v.8.1, National Instruments Corporation).

FIGURE 2
FIGURE 2:
Schematic of the serial targeting task practice. Performance of the task involved manipulation of a computer mouse housed in a custom frame that was held in a pronated grasp with the nondominant hand. Participants were instructed to use the mouse to move a cross-hair cursor between a series of discretely presented targets as quickly and accurately as possible while taking the most direct route.

Embedded within the target movements that were presented, there was a repeated six-target sequence that was flanked by seven-target random sequences. For the repeated sequence, the same sequence was repeated across familiarization, practice, and retention blocks. For each participant, the target movements within the sequences were reversed between conditions (rest and aerobic exercise), such that the sequences differed but shared equivalent difficulty. The order of presentation of conditions (rest and aerobic exercise) and the sequences (regular and reversed) assigned to each participant were pseudorandomized such that it was balanced across the sample. The random sequences were different at each presentation throughout a block. Participants were not informed of the existence of the repeated sequence but instructed to move the cursor to all targets as quickly and accurately as possible while taking the most direct route. The inclusion of repeated and random sequences allows the separation of change associated with implicit sequence-specific learning (repeated sequences) and those associated with generalized improvements in motor control (random sequences) (7,8,27). In the experimental sessions involving serial targeting task practice, participants completed three sequences (20 target movements; one repeated sequence and two random sequences) before the rest period or the aerobic exercise bout for task familiarization. For serial targeting task practice after rest or the aerobic exercise bout, participants completed three blocks of 110 target movements (8 × 6-target repeated sequences and 9 × 7-target random sequences). The following day (24 ± 2 h after motor practice), participants completed another single block of the serial targeting task (no-exercise delayed retention test).

After the final retention test, participants were tested for an explicit recognition of the repeated sequence within the serial targeting task from both conditions (rest and aerobic exercise). For the recognition testing, participants viewed a series of 20 target sequences and were asked to indicate whether they recognized any of the presented sequences by selecting “yes” or “no” with a computer mouse. Of the 20 sequences presented, 14 were random sequences, three were the repeated sequences from the rest session, and three were the repeated sequence from the aerobic exercise session. If explicit knowledge of the repeated sequence was acquired, we expected that, on average, the repeated sequence would be recognized correctly at a level higher than that associated with chance (i.e., >3/6 repeated sequence) (7,8).

Serial Targeting Task Analyses

Sequence response time, the sum of all target response times within an entire target sequence, was used to measure task performance. To obtain this measure, the sum response time of all target movements within a sequence was calculated for each movement sequence within a block. Response times were calculated separately for random and repeated sequences.

Baseline performance

To allow for comparison of serial targeting task performance at the outset of experimental sessions under each condition (rest and exercise), the response time for the three sequences (two random, one repeated sequence) presented in the familiarization period and the first three sequences in the practice period (initial practice) were averaged (two random, one repeated sequence).

Motor skill acquisition and learning

Response time from the first four sequences of the first practice block (early acquisition), the last four sequences of the third practice block (late acquisition), and the first four sequences of the retention block (retention) were averaged separately for random and repeated sequences. This method is consistent with past work investigating acute aerobic exercise effects on continuous motor sequence task learning (22) and other work using this serial targeting task (23). Change in response time was then calculated for each individual between early and late acquisition (acquisition change score [Δ-ACQ]) to assess practice-related changes in motor performance, and between early acquisition and retention (retention change score, Δ-RET) to assess motor learning (22).

Next, we used an exponential decay (decreasing form) curve fitting approach to obtain information about participants’ rate of improvement in response time for random and repeated sequences separately. Given the evolution of memory processes that contribute to motor performance during practice and at a delayed retention test (16), we fit curves to data obtained over practice (skill acquisition) and at the delayed retention test (skill retrieval) separately. Sequence response times from the practice blocks (blocks 1–3) and at retention were fit with a least squares regression analysis using the following equation (9):

E(RTN) is the expected value of the response time on sequence trial N. We were specifically interested in α, which represented the rate of motor skill acquisition (α-ACQ) during practice and the rate of retrieval of the motor memory (α-RET) during the retention test. All curve fitting was conducted using a custom MATLAB script (Mathworks, Natick, MA).

Statistical Analyses

Baseline performance and order effects

Data were evaluated for differences between conditions in serial targeting task performance (i.e., response time) at the outset of the experimental sessions. To achieve this, a two-way repeated-measures ANOVA (RM-ANOVA) was conducted with condition (rest and aerobic exercise) and time (familiarization and initial practice) as the within-subjects factors and response time as the dependent variable. In addition, all data were tested for differences between individuals’ first and second exposure to the experiment using multiple uncorrected paired t-tests, and for differences in outcomes between individuals experiencing the rest or exercise condition first using multiple uncorrected two-sample t-tests.

Motor skill acquisition and retention

To examine the effects of acute aerobic exercise on sequence-specific learning of the serial targeting task, separate two-way condition (rest or aerobic exercise) by sequence (repeated or random) RM-ANOVA were conducted for the following dependent variables: Δ-ACQ, Δ-RET, α-ACQ, and α-RET. For all RM-ANOVA, post hoc analyses (Fisher’s least significant difference tests) were conducted where appropriate.

Recognition

To examine whether there was recognition of the repeated sequence across the group, a one-sample one-tailed t-test was conducted to determine whether the number of correct identifications of the repeated sequence was greater than what would be expected by chance (>3/6). A paired t-test was then conducted to determine whether identification of the repeated sequence was different between the sequence presented for the rest and exercise conditions. Lastly, bivariate correlation analyses (Pearson’s r) were conducted to determine whether recognition scores were related to the difference between repeated and random sequence outcome measure values (Δ-ACQ, Δ-RET, α-ACQ, and α-RET) under each condition.

Normality of data

All data were visually inspected for skewness and kurtosis and objectively tested for normality with the Shapiro–Wilk test with a significance level set at P < 0.001 (14). The following data were found to be nonnormal (W16 ≤ 0.760, P < 0.001): α-ACQ for the repeated sequence under both rest (W16 = 0.744, P < 0.001) and exercise (W16 = 0.691, P < 0.001) conditions and the random sequence under the rest condition (W16 = 0.475, P < 0.001), as well as α-RET for both repeated (W16 = 0.745, P < 0.001) and random sequences (W16 = 0.760, P < 0.001) under the rest condition. For statistical analyses, square root transformations were applied to all α-ACQ and α-RET data. After square root transformation, all α-ACQ and α-RET data were found to be normal, based on visual inspection and objective testing with the Shapiro–Wilk test (W16 range = 0.774–0.918, P range = 0.0012–0.160). All baseline performance and change score data were found to be normally distributed (W16 range = 0.784–0.983, P range = 0.0017–0.985) and were subsequently analyzed in their raw form. All descriptive statistics are reported as mean ± SD. For all statistical tests, significance level was set at P < 0.05. Statistical analyses were conducted using SPSS software (SPSS 21.0; IBM Corporation, Armonk, NY).

RESULTS

Response Time

Baseline performance and order effects

The RM-ANOVA examining initial performance yielded a significant main effect of time (F1,15 = 24.14, P < 0.001, η2partial = 0.62), with an average response time of 9.03 ± 1.66 s at familiarization and 8.06 ± 1.40 s at initial practice across conditions. There was no effect of condition (F1,15 = 3.01, P = 0.10, η2partial = 0.17) and no interaction effect (F1,15 = 0.63, P = 0.44, η2partial = 0.04), indicating that, across the group, baseline serial targeting task performance improved after familiarization but was not significantly different between conditions at either time point. Multiple paired t-tests to evaluate performance between individuals’ first and second exposure to the task indicated that there was a trend for better performance (lower response time) during familiarization on the second exposure (t15 = 1.99, P = 0.07); however, this trend was no longer present at initial practice (t15 = 0.83, P = 0.42). There was a trend for a lower Δ-ACQ for the random sequence on the second exposure (t15 = −1.88, P = 0.09); all other outcomes were not significantly different between the first and the second exposure to the task (t15 < |1.39|, P > 0.18). Comparing outcomes between individuals with different order of exposure to experimental conditions (i.e., rest first or exercise first), we found trends that individuals who performed the exercise condition first demonstrated a higher α-RET for the repeated sequence under the rest condition (t14 = −2.10, P = 0.06) and a greater △-ACQ for the random sequence under the rest condition (t14 = −1.94, P = 0.07). All other comparisons between these groups were nonsignificant (t14 < |1.76|, P > 0.10). Overall, these results suggest that baseline performance was not different between experimental conditions and that any potential order effects on the results were largely mitigated by the experimental design.

Motor skill acquisition

Group average response times for each movement sequence trial over the course of motor practice are presented in Figure 3A. A RM-ANOVA indicated that there was a significant main effect of sequence on α-ACQ (F1,15 = 6.26, P = 0.02, η2partial = 0.29), with a faster rate of acquisition for the repeated sequence relative to the random sequence, regardless of condition (Fig. 4A). There was no effect of condition (F1,15 = 0.28, P = 0.61, η2partial = 0.02) and no interaction effect (F1,15 = 0.28, P = 0.61, η2partial = 0.02) on α-ACQ. When considering ΔACQ, there was no effect of condition (F1,15 = 0.02, P = 0.89, η2partial < 0.01), sequence (F1,15 < 0.01, P = 0.97, η2partial < 0.01), or their interaction (F1,15 = 0.34, P = 0.57, η2partial = 0.02). These results suggest that there was a sequence-specific increase in the rate (α-ACQ), but not the extent (ΔACQ), of skill acquisition over practice; however, this effect was not different between rest and aerobic exercise conditions.

FIGURE 3
FIGURE 3:
Average response time across the group for each movement sequence trial during practice (A) and retention (B). Trial-by-trial data points are depicted for the rest (red circles) and aerobic exercise (green squares) conditions for the repeated (filled) and random (unfilled) sequences. Rest condition data points (red circles) are shifted by 0.25 U on the x-axis for ease of viewing. During the practice session (A), individuals completed three blocks, each composed of eight repeated sequences (8 × 3 = 24 sequence trials) and nine random sequences (9 × 3 = 27 sequence trials). During the retention session (B), participants completed one block composed of eight repeated sequences and nine random sequences.
FIGURE 4
FIGURE 4:
Rate of change parameter (α), obtained from exponential decay curves fit to trial-by-trial practice (A) and retention (B) data, averaged across the group. Red and green bars denote α obtained from data collected under rest and aerobic exercise conditions, respectively. Filled and patterned bars show data for the repeated and random sequences. Error bars depict 1 SD. α-ACQ, rate of acquisition; α-RET, rate of retrieval.

Motor skill retention

Group average response times for each movement sequence trial over the course of the retention test are presented in Figure 3B. An RM-ANOVA demonstrated a significant condition by sequence interaction effect on α-RET (F1,15 = 7.13, P = 0.02, η2partial = 0.32). There was no main effect of condition (F1,15 = 3.14, P = 0.10, η2partial = 0.17) or sequence (F1,15 = 1.35, P = 0.26, η2partial = 0.08) on α-RET. Post hoc analyses of the significant interaction indicated a significantly greater α-RET for the repeated sequence relative to the random sequence for the aerobic exercise condition (P = 0.01, η2partial = 0.32). By contrast, there was no difference in α-RET values between sequences for the rest condition (P = 0.33, η2partial = 0.06). Further, α-RET for the repeated sequence under the aerobic exercise condition was also greater than α-RET for both the repeated (P < 0.01, η2partial = 0.38) and random sequences (P = 0.01, η2partial = 0.18) under the rest condition (Fig. 4B). On the other hand, the RM-ANOVA evaluating △-RET demonstrated no effect of condition (F1,15 = 0.09, P = 0.77, η2partial < 0.01), sequence (F1,15 = 0.20, P = 0.66, η2partial = 0.01), or their interaction (F1,15 = 0.07, P = 0.79, η2partial < 0.01). Altogether, these results indicate that there was a sequence-specific increase in the rate of retrieval or relearning of the motor skill in the aerobic exercise compared with the rest condition, but that the retention change score was not different between sequences or conditions.

Recognition

The repeated sequence was correctly identified at a level consistent with chance (3.0 ± 2.3/6 or 50.0% ± 38.0% correct; correct identification ≯3/6, t15 = 0.0, P = 0.5), indicating that across the group, participants did not demonstrate explicit knowledge of a repeated sequence. Also, the identification of the repeated sequence presented during the rest and aerobic exercise sessions was not significantly different (rest: 1.5 ± 1.2/3 or 50.0% ± 40.0% correct; aerobic exercise: 1.5 ± 1.3/3 or 50.0% ± 43.0%; t15 = 0.0, P = 1.0). Lastly, recognition scores did not correlate with the difference between repeated and random sequence outcome measures under either experimental condition (r < |0.24|, P > 0.36). Taken together, these results suggest that explicit memory processes likely did not influence our findings related to aerobic exercise effects on implicit sequence-specific motor learning.

DISCUSSION

This study was designed to evaluate the effect of a single bout of high-intensity aerobic exercise, performed immediately before practice of a discrete motor sequence task (i.e., serial targeting task), on the extent and rate of implicit sequence-specific skill acquisition and learning. There was a positive effect of acute aerobic exercise compared with a period of seated rest, on the rate of improvement in performance at a 24-h no-exercise delayed retention test. By contrast, there were no significant effects of acute aerobic exercise on the rate of improvement during practice (i.e., acquisition), or the overall change in performance across practice and from practice to retention. Thus, our main finding was that acute high-intensity aerobic exercise enhanced implicit sequence-specific motor learning of a discrete motor sequence task, expressly by increasing the rate of motor memory retrieval or relearning (i.e., savings).

Motor learning can involve memories that are accessed implicitly (i.e., without conscious awareness) (33,36) and rely on multiple brain regions, including primary sensory and motor cortices, premotor cortex, supplementary motor area, prefrontal cortex, cerebellum, and basal ganglia (13,21). Within this overarching implicit motor learning brain network, the relative contributions of specific brain regions vary based on the characteristics of the learned task (10,13). The only studies to demonstrate the positive effects of acute aerobic exercise specifically on motor learning used tasks of a very similar nature, which involved learning of a continuous motor sequence (22,30). Given the use of visuomotor rotations (22,30) and preferential benefits in the learning of temporal precision (22), the previously observed acute aerobic exercise effects were likely dependent on cerebellar function (4,6). In the present study, we found that acute aerobic exercise performed before motor task practice also benefited the learning of a discrete motor sequence task with no visuomotor rotation. The task used here, the serial targeting task, was similar in nature to the commonly used serial reaction time (1) and discrete sequence production motor tasks (2), but instead of fine movements involving a finger key press, gross movements of the wrist and arm, including cross-lateral movements for certain targets, were used to move a computer mouse. Although regional brain activity was not measured here, learning of discrete motor sequence tasks, such as our serial targeting task, may be more dependent on basal ganglia circuits (5,15) than the tasks used in the aforementioned work (22,30).

In the current experiments, there was an enhanced rate of acquisition of the repeated, relative to the random, sequence during practice that was independent of condition. This result suggests that either implicit sequence-specific encoding processes occurred to a similar degree under the rest and aerobic exercise conditions or, as previously suggested by Roig et al. (30), that enhanced encoding under the exercise condition was masked by fatigue-related decreases in performance. When evaluating learning at retention, the rate of retrieval or relearning for the repeated sequence under the aerobic exercise condition was greater than for the random sequence and for both sequences under the rest condition. Also, this rate parameter did not differ between sequences at retention for the rest condition. These findings indicate that implicit sequence-specific learning occurred when practice was preceded by aerobic exercise, but not rest. The enhanced rate of retrieval or relearning, but not acquisition, under the exercise condition is consistent with the concept of savings (18) and suggests a potential role for acute aerobic exercise to improve consolidation processes. Nevertheless, given that the exercise bout was performed before motor practice, we cannot discount the idea that enhanced encoding, potentially masked by exercise-induced fatigue, may also contribute to the observed learning benefits. Regardless, the observation of an increased rate of memory retrieval at a delayed retention test, when practice was preceded by exercise, demonstrates a significant benefit of acute high-intensity aerobic exercise for implicit learning of discrete movement sequences.

The results partly align with our previous work that evaluated implicit sequence-specific motor learning with a continuous tracking task (22). In the previous study (22) and the current experiment, we attempted to minimize participants’ exposure to the task to reduce the potential of carryover effects between conditions, but this also resulted in the use of a relatively small dose of practice. As in the previous study (22), it appears that the practice dose was not sufficient for implicit sequence-specific learning to occur under the rest condition, but that acute high-intensity aerobic exercise facilitated the rapid formation of an implicit motor memory. Thus, the results suggest that acute high-intensity aerobic exercise can promote implicit motor learning, regardless of whether a task is continuous or discrete in nature.

The results of two recent studies (28,37) raised uncertainty as to whether the positive effects of acute aerobic exercise on continuous motor sequence learning (22,30) generalize to discrete motor sequence tasks. Rhee et al. (28) found that moderate-to-vigorous cycling performed 2 h after practice of a discrete movement sequence “A,” but immediately before practice of an alternative discrete movement sequence “B,” protected the learning of sequence “A” from interference effects. The lack of change in sequence “B” performance suggested that the acute exercise before motor practice did not affect its acquisition (28). Also, had the exercise benefited sequence “B” learning, one might have expected exercise to elicit a greater interference effect on sequence “A” learning, although sequence “B” learning was not tested at retention (28). Importantly, these results reported by Rhee et al. (28) are likely related to the specific experimental protocol that was used to examine motor memory interference and are not necessarily at odds with our current finding that acute aerobic exercise immediately before motor practice benefited learning of a discrete motor sequence. In another study by Statton et al. (37), the acquisition of a discrete motor sequence task, but not learning, was enhanced when practice was preceded by a bout of moderate-intensity aerobic exercise. By contrast, we presently found that acute high-intensity aerobic exercise enhanced learning, but not acquisition, of a discrete motor sequence task. Thus, given that acute aerobic exercise benefits on continuous motor sequence learning were demonstrated with high-intensity exercise bouts (22,30), it seems that the lack of a learning effect in the recent study by Statton et al. (37) was more likely related to the exercise prescription than to the nature of the task. Although moderate-intensity aerobic exercise may have the capacity to promote immediate motor performance gains (37,38), perhaps high-intensity aerobic exercise is needed to enhance later memory processes, such as consolidation and retrieval.

Although our current findings suggest that the positive effects of acute high-intensity aerobic exercise on motor learning generalize to discrete motor sequence tasks, the results are still somewhat contradictory to our previous study using a continuous motor tracking task (22). Previously, we observed an effect of acute aerobic exercise on the overall change in motor performance of a repeated sequence across practice and from practice to retention (22), whereas presently we observed no effect of exercise on the change score measures with the serial targeting task. The preferential effect of acute aerobic exercise on rate of improvement rather than change in performance for the serial targeting task used here may relate to the specific processes that underpin discrete motor sequence learning (3,32). For example, past work suggests that individuals commonly show a greater degree of forgetting of discrete, rather than continuous, motor tasks over delayed retention intervals (3,32). Perhaps, the initial retention test performance of the discrete movement sequence task in the present study was compromised by such forgetting, and instead, the speed at which the movement sequences were reprocessed, remembered, or relearned over multiple trials as a measure of savings provided a better indicator of the learning process. Thus, we speculate that the benefits of aerobic exercise may have been most amenable to an influence on motor memory savings, in part because of the discrete nature of the movements. Although an increased rate of improvement at retention could occur as a statistical artifact (individuals who perform poorly at the outset of the retention test have greater room for improvement), the lack of difference in retention change score between conditions suggests that this was not the case for the observed effects.

It is also important to note that the main findings in the current study were found when considering the rate of improvement in serial targeting task performance, although consideration of overall change in performance did not yield any significant effects. Commonly, motor skill acquisition and learning is assessed by performance measurements taken at discrete points in time (16,33), as we did here when calculating change scores. Although this approach has value for evaluating motor learning, it is also susceptible to the dilution of effects due to within-subject variability in motor performance (25). A curve fitting approach that uses measures obtained across multiple time points is less vulnerable to misrepresentation of effects due to unsystematic within-subject variability (9,25). Thus, under the present experimental conditions, the curve fitting analysis approach may have provided a more sensitive measure (i.e., rate of improvement) than the change scores (derived from measures of performance at discrete time points) for detecting interactions between acute aerobic exercise and implicit sequence-specific motor learning.

Regardless of how the learning benefit manifests (i.e., change score or rate of improvement), a positive effect of acute aerobic exercise on motor learning across varying motor tasks is consistent with the general hypothesis that the effects of acute aerobic exercise on memory are driven by transient increases in neurochemicals (29,34). Neurochemicals upregulated by aerobic exercise are expressed across multiple brain regions (24) and, thus, may be most used by the brain regions that are activated by subsequent experiences, such as specific types of motor practice. Another consideration relates to whether acute aerobic exercise might exert its effects partly through interaction with other factors known to affect motor sequence learning. For example, given evidence that sleep can affect motor memory consolidation (39), it is possible that engagement in aerobic exercise might interact with individuals’ quality of sleep to affect motor learning. Also, past work suggests that motor sequence consolidation occurring over periods of wakefulness and sleep differentially contribute to movement- and goal-based learning (11). Roig et al. (30) also previously found that acute aerobic exercise had a greater effect on motor learning as assessed with 24-h and 7-d retention intervals that involved sleep compared with a 1-h retention interval with no sleep. Thus, it may be interesting for future work to consider how acute aerobic exercise interacts with different phases of consolidation.

Implications

Humans have a remarkable ability to learn and perform movement that is crucial for participation in everyday life. However, the memory processes supporting motor skill learning are highly dependent on the nature of the learned task, and as such, practice conditions that benefit one type of skill learning do not necessarily extend to other tasks (10,13,20,33). The present findings suggest that acute high-intensity aerobic exercise can benefit the learning of a discrete movement sequence, in addition to its positive effects on continuous motor sequence learning, which have been shown in past work. Although replications of this effect and further work on the topic is necessary, the generalization of an acute aerobic exercise effect across different movement tasks suggests that it may provide a means to facilitate varying movement tasks in sport or rehabilitation settings (33). Nevertheless, the past and the present studies demonstrating these learning effects used high-intensity aerobic exercise bouts, based on indications of a dose-dependent effect of exercise intensity on neurochemical production (17,40). Thus, further investigation of how the exercise prescription affects motor learning will be important for progress on this topic of study.

CONCLUSIONS

Our results indicate that a single bout of high-intensity aerobic exercise enhances discrete motor sequence task learning, preferentially through an increase in the rate of motor memory retrieval or relearning (i.e., savings) rather than an overall change in performance, as shown previously with continuous motor sequence tasks. Thus, the positive effects of acute aerobic exercise on motor learning appear to occur for both continuous and discrete tasks, but the specific learning benefits of acute aerobic exercise may be somewhat task specific. In conclusion, these data add to a growing body of evidence suggesting that acute bouts of high-intensity aerobic exercise may be used to facilitate motor skill learning in sport and rehabilitation contexts.

This work was funded by awards from the Natural Sciences and Engineering Research Council of Canada (RGPIN 401890-11 to L. A. B.) and the Peter Wall Institute for Interdisciplinary Studies at the University of British Columbia (awards to L. A. B.). C. S. M., N. J. S., and K. P. W. were supported by the Natural Sciences and Engineering Research Council of Canada. C. S. M. and K. P. W. received support from the University of British Columbia Four Year Fellowship Program. N. J. S. received support from the University of British Columbia Li Tzi Fong Memorial Fellowship. L. A. B. receives salary support from the Canada Research Chairs and the Michael Smith Foundation for Health Research.

The authors thank Kris Deasis for technical assistance.The authors have no conflicts of interest.

The results of the present study do not constitute endorsement by the American College of Sports Medicine.

REFERENCES

1. Abrahamse EL, Jiménez L, Verwey WB, Clegg BA. Representing serial action and perception. Psychon Bull Rev. 2010;17(5):603–23.
2. Abrahamse EL, Ruitenberg MF, De Kleine E, Verwey WB. Control of automated behavior: insights from the discrete sequence production task. Front Hum Neurosci. 2013;7:82.
3. Bailey CD. Forgetting and the learning curve: a laboratory study. Manag Sci. 1989;35(3):340–52.
4. Bernard JA, Seidler RD. Cerebellar contributions to visuomotor adaptation and motor sequence learning: an ALE meta-analysis. Front Hum Neurosci. 2013;7:27.
5. Boyd LA, Edwards JD, Siengsukon CS, Vidoni ED, Wessel BD, Linsdell MA. Motor sequence chunking is impaired by basal ganglia stroke. Neurobiol Learn Mem. 2009;92(1):35–44.
6. Boyd LA, Winstein CJ. Cerebellar stroke impairs temporal but not spatial accuracy during implicit motor learning. Neurorehabil Neural Repair. 2004;18(3):134–43.
7. Boyd L, Winstein C. Explicit information interferes with implicit motor learning of both continuous and discrete movement tasks after stroke. J Neurol Phys Ther. 2006;30(2):46–57.
8. Boyd LA, Winstein CJ. Impact of explicit information on implicit motor-sequence learning following middle cerebral artery stroke. Phys Ther. 2003;83(11):976–89.
9. Brown S, Heathcote A. Averaging learning curves across and within participants. Behav Res Methods Instrum Comput. 2003;35(1):11–21.
10. Chambaron S, Berberian B, Delbecque L, Ginhac D, Cleeremans A. Implicit motor learning in discrete and continuous tasks: toward a possible explanation of discrepant results. In: Pelligrino LT, editor. Handbook of Motor Skills: Development, Impairment, and Therapy; 2009. pp. 139–55.
11. Cohen DA, Pascual-Leone A, Press DZ, Robertson EM. Off-line learning of motor skill memory: a double dissociation of goal and movement. Proc Natl Acad Sci U S A. 2005;102(50):18237–41.
12. Cotman CW, Berchtold NC, Christie LA. Exercise builds brain health: key roles of growth factor cascades and inflammation. Trends Neurosci. 2007;30(9):464–72.
13. Doyon J, Penhune V, Ungerleider LG. Distinct contribution of the cortico-striatal and cortico-cerebellar systems to motor skill learning. Neuropsychologia. 2003;41(3):252–62.
14. Gamst G, Meyers LS, Guarino AJ. Analysis of Variance Designs. New York (NY): Cambridge University Press; 2008. pp. 56–7.
15. Graybiel AM. The basal ganglia and chunking of action repertoires. Neurobiol Learn Mem. 1998;70(1):119–36.
16. Kantak SS, Winstein CJ. Learning-performance distinction and memory processes for motor skills: a focused review and perspective. Behav Brain Res. 2012;228(1):219–31.
17. Knaepen K, Goekint M, Heyman EM, Meeusen R. Neuroplasticity—exercise-induced response of peripheral brain-derived neurotrophic factor: a systematic review of experimental studies in human subjects. Sports Med. 2010;40(9):765–801.
18. Krakauer JW, Shadmehr R. Consolidation of motor memory. Trends Neurosci. 2006;29(1):58–64.
19. Lambourne K, Tomporowski P. The effect of exercise-induced arousal on cognitive task performance: a meta-regression analysis. Brain Res. 2010;1341:12–24.
20. Lee TD, Genovese ED. Distribution of practice in motor skill acquisition: different effects for discrete and continuous tasks. Res Q Exerc Sport. 1989;60(1):59–65.
21. Lohse KR, Wadden K, Boyd LA, Hodges NJ. Motor skill acquisition across short and long time scales: a meta-analysis of neuroimaging data. Neuropsychologia. 2014;59:130–41.
22. Mang CS, Snow NJ, Campbell KL, Ross CJ, Boyd LA. A single bout of high-intensity aerobic exercise facilitates response to paired associative stimulation and promotes sequence-specific implicit motor learning. J Appl Physiol (1985). 2014;117(11):1325–36.
23. Meehan SK, Dao E, Linsdell MA, Boyd LA. Continuous theta burst stimulation over the contralesional sensory and motor cortex enhances motor learning post-stroke. Neurosci Lett. 2011;500(1):26–30.
24. Murer MG, Yan Q, Raisman-Vozari R. Brain-derived neurotrophic factor in the control human brain, and in Alzheimer’s disease and Parkinson’s disease. Prog Neurobiol. 2001;63(1):71–124.
25. Nesselroade JR, Salthouse TA. Methodological and theoretical implications of intraindividual variability in perceptual-motor performance. J Gerontol B Psychol Sci Soc Sci. 2004;59(2):P49–55.
26. Petzinger GM, Fisher BE, Van Leeuwen JE, et al. Enhancing neuroplasticity in the basal ganglia: the role of exercise in Parkinson’s disease. Mov Disord. 2010;25(1 Suppl):S141–5.
27. Pew RW. Levels of analysis in motor control. Brain Res. 1974;71(2):393–400.
28. Rhee J, Chen J, Riechman SM, Handa A, Bhatia S, Wright DL. An acute bout of aerobic exercise can protect immediate offline motor sequence gains. Psychol Res. 2016;80(4):518–31.
29. Roig M, Nordbrandt S, Geertsen SS, Nielsen JB. The effects of cardiovascular exercise on human memory: a review with meta-analysis. Neurosci Biobehav Rev. 2013;37(8):1645–66.
30. Roig M, Skriver K, Lundbye-Jensen J, Kiens B, Nielsen JB. A single bout of exercise improves motor memory. PLoS One. 2012;7(9): e44594.
31. Sakai K, Kitaguchi K, Hikosaka O. Chunking during human visuomotor sequence learning. Exp Brain Res. 2003;152(2):229–42.
32. Schendel JD, Hagman JD. On sustaining procedural skills over a prolonged retention interval. J Appl Psychol. 1982;67(5):605.
33. Schmidt RA, Lee TD. Motor Control and Learning: A Behavioral Emphasis. 4th ed. Champaign (IL): Human Kinetics Publishers; 2005. pp. 301–459.
34. Skriver K, Roig M, Lundbye-Jensen J, et al. Acute exercise improves motor memory: exploring potential biomarkers. Neurobiol Learn Mem. 2014;116:46–58.
35. Snow NJ, Mang CS, Roig M, McDonnell MN, Campbell KL, Boyd LA. The effect of an acute bout of moderate-intensity aerobic exercise on motor learning of a continuous tracking task. PLoS One. 2016;11(2): e0150039.
36. Squire LR. The organization and neural substrates of human memory. Int J Neurol. 1987;21–22:218–22.
37. Statton MA, Encarnacion M, Celnik P, Bastian AJ. A single bout of moderate aerobic exercise improves motor skill acquisition. PLoS One. 2015;10(10): e0141393.
38. Vaynman S, Ying Z, Gomez-Pinilla F. Hippocampal BDNF mediates the efficacy of exercise on synaptic plasticity and cognition. Eur J Neurosci. 2004;20(10):2580–90.
39. Walker MP, Brakefield T, Morgan A, Hobson JA, Stickgold R. Practice with sleep makes perfect: sleep-dependent motor skill learning. Neuron. 2002;35(1):205–11.
40. Winter B, Breitenstein C, Mooren FC, et al. High impact running improves learning. Neurobiol Learn Mem. 2007;87(4):597–609.
Keywords:

MOTOR LEARNING; IMPLICIT MEMORY; RETRIEVAL; DISCRETE; SAVINGS

© 2016 American College of Sports Medicine