Obstacle course activity in soldiers impedes recall but not learning of a read text

ABSTRACT

Research suggests that preceding physical activity benefits learning in the general population, however the impact of physical activity before testing has not been systematically explored. We examined how an acute, complex anaerobic activity impacted learning and retention of text. Soldiers read Special Operations or Pension texts and were tested on recall and comprehension at 1 and 48 hours after reading. Participants also performed a challenging physical course before learning, before testing, or outside of these two periods. Results were compared using a mixed Analysis of Covariance with a “fitness-effort” covariate factor. Testing complied with Human Research Ethics Committee guidelines. Participants performing physical activity before testing had lower mean first test scores than participants performing the activity outside of the learning or testing window. After 48 hours, there was no difference among groups. Equally, while participants who read the Special Operations materials performed better than those who read the Pension materials, there was no effect related to activity timing. Acute obstacle course activity impeded recall but not learning of a read text. Additional study is needed to determine whether metabolic or motoric demands of the physical activity caused the deficit.

KEYWORDS:

• Learning
• retention
• exercise
• cognition
• soldiers
• military

What is the public significance of this article?—We had soldiers perform a challenging obstacle course before learning or before testing, and found that the group performing activity immediately before testing had worse recall, but there was no difference when all groups were retested after 48 hours. Soldiers, police and other first responders make life and death decisions based on the recall of recently learned information. Because strenuous physical activity is also normal for these personnel, it is critical to characterize and improve learning when physical activity is also anticipated.

Introduction

Soldiers must learn and retrieve pertinent operational information in the midst of physical exertion. The question remains as to whether soldier fitness levels and activity types help or hinder their learning. Canadian Armed Forces (CAF) soldiers participate in daily physical training (PT) as part of the mandate to meet the CAF’s minimum physical fitness standard, as well as readiness requirements. The yearly qualification test includes sandbag lifts, sandbag drags, intermittent loaded shuttles, and 20-meter rushes that must be performed within a prescribed period of time (Canadian Forces Morale and Welfare Services, 2016), although dismounted soldier activity during operations is much more diverse. Soldiers are typically more active than the general population (Jacobs, 2015; Tharion et al., 2005); however, there is a lack of empirical evidence showing whether soldiers’ fitness levels impact learning or recall immediately following physical activity. Soldiers (mean age 22 years) performing routine patrol and reconnaissance duty with a peak heart rate of 160 beats per minute (BPM) over 15 minutes did not suffer any deficit in speed or accuracy in a test that requires quickly checking names and numbers (Amos, 2000). Further, it is unknown how the timing and the character of the specific physical activities performed by soldiers may themselves aid or interfere with learning or recall.

A growing body of research indicates that physical activity is a predictor of improved learning outcomes (Hillman et al., 2008), and typically, chronic physical activity (fitness) benefits cognitive performance (Etnier et al., 2006; Tomporowski & Ellis, 1986). More recently, researchers have demonstrated that acute bouts of exercise have immediate benefits for various cognitive functions (Chang et al., 2012; Coles & Tomporowski, 2008; Lambourne et al., 2010). Chang and colleagues performed a meta-analysis and found an interaction between the effect of overall fitness and acute exercise on performance: highly fit individuals show an advantage for cognitive performance during and immediately after acute exercise, while moderately fit individuals enjoy a modest benefit during exercise, but none immediately after. Low-fit individuals perform more poorly during – but improve immediately after – exercise. In that meta-analysis, “fitness” refers to cardiorespiratory fitness (Chang et al., 2012). The fitness levels were based on the American College of Sports Medicine 2010 Guidelines comparing VO_2max scores of adults of the same sex and age group against a national sample. Low fitness was that below the 20^th percentile of the sample, moderate was defined as between the 20th and 59^th percentile, and high was defined as being above the 60^th percentile. Fitness level does not impact performance after a delay, although timing between exercise and testing is a moderating factor on cognitive performance (Chang et al., 2012).

In the above meta-analysis by Chang and others, they noted that in experiments examining cognitive performance in general, the test may occur during exercise, immediately after exercise, or after a delay. They found the greatest positive effects for exercise when testing after a delay of 12 minutes, with diminishing positive effects after a delay of 20 minutes (Chang et al., 2012). A comprehensive review by Schmit and Brisswalter shows a sharp decline in executive function during prolonged exercise after a given period, depending on the duration and intensity of exercise. Most of the studies reviewed found a decline in executive function after exercise durations of longer than 60 minutes. In one study however, when subjects performed near maximal muscle contractions until fatigue, their error rate in a simultaneous choice reaction task increased as time elapsed (Schmit & Brisswalter, 2018). These authors and others (Marcora et al., 2009) have proposed that some neurocognitive resources are shared by mental and physical activity. However, some complex physical activities may actually benefit cognition.

A recent study examined the effects of different types of exercise on working memory in children. The research showed greater cognitive benefit for a varied exercise program featuring balancing, multi-limb coordination, spatial orientation, and ladders than a program of running and running-based games (Koutsandreou et al., 2016). In another study, executive function in 10 year-old children benefited more from an open-skilled, play-based approach to tennis instruction than from a technique-based approach. The cognitive testing in that instance was performed 15 minutes after the intervention (Ishihara et al., 2017). The present experiment features testing timing and activity similar to the more complex or play-based activities in these experiments with children, although at a much higher intensity.

In many instances of human research on this topic to date, the exercise intervention is placed before the learning period (Hillman et al., 2008), but not before the testing period. Equally, different types of exercise result in different physiological changes within the brain. Types of activity requiring diverse, often improvised movement appear to benefit cognition, possibly because they contribute to the development of new neural growth and connections (Black et al., 1990), and also result in different cognitive outcomes (Ishihara et al., 2017; Pesce, 2012). In addition to these changes and outcomes, distinct brain structures and processes are used in the encoding and retrieval of learned information (Kensinger et al., 2003; Prince et al., 2005). It is not unreasonable to expect that different types of physical activity may impact learning and recall differently.

Neuropsychological assessments typically used in such studies only explore one specific aspect of cognitive function, such as information processing, reaction time, attention, crystallized intelligence, executive function, or memory (Lezak, 2004). Chang’s meta-analytic review of research on exercise and cognitive task performance considered how performance was impacted by the timing and intensity of physical activity, as well as the nature of the cognitive task under consideration. The analysis revealed that visual short-term memory and free recall tests benefitted from exercise, in contrast to sequential memory and auditory verbal learning which were performed more poorly (Chang et al., 2012). While such discreet tasks are critical to understanding how the brain is impacted by the intervention, reading comprehension requires the concerted application of multiple cognitive functions such as attention, crystallized intelligence, executive function, and memory.

Reading is frequently used as a form of self-paced learning, giving the learner opportunities to review salient or challenging material, as well as to reduce focus on information they believe is already known or unimportant (Tullis & Benjamin, 2011). Furthermore, an older study showed that required literacy levels are higher in the military than in comparable civilian careers, in part because the complexity, cost, and potential danger of the military equipment and operations requires comprehensive training which is frequently text based (Duffy, 1985). Of closer concern to the present study, text and image-based mission briefings are used in-theater to reinforce the commander’s intent and concept of operation, provide target descriptions and mission details (United States Marine Corps, 2017), all of which must be recalled in a contested environment with life and death consequences, frequently during or after vigorous activity.

Defense Research and Development Canada – Toronto Research Center (DRDC-TRC) holds a yearly winter experimental campaign, with upward of 50 regular force and reserve soldiers participating in a wide variety of studies and experiments examining human performance, anthropometrics, and the integration of humans with equipment and systems. A regular feature of the campaign is the Canadian Load Effects Assessment Program (CANLEAP), a comprehensive physical challenge employing movements required of dismounted soldiers. To dovetail with an already scheduled experimental campaign, the present study was opportunistically implemented to address the question of how the timing of the CANLEAP activity in relationship to study and testing periods might impact learning and retention of information in a passage of text. We hypothesized that soldiers would have at least moderate cardio-respiratory fitness and consequently would enjoy no benefit from acute activity immediately before learning, in keeping with previous results. We also hypothesized that the duration of the CANLEAP activity would be too brief to negatively influence recall of the read material, but that the movement diversity of the CANLEAP could even benefit performance.

Material and methods

Participants

The participants were drawn from the 2015 Winter Experimentation campaign and had no previous experience on the CANLEAP course. The inclusion criteria for the campaign stipulated physically fit, combat arms soldiers with a Personal Weapons Test 3 qualification (high proficiency), functional English, and a preference for participants with combat or recent operational experience. A total of 51 male soldiers (43 infantry, 8 engineering trade) participated in the immediate study, aged 26–43, mean age 34 years, (SD 3.8). Forty-eight of these participated in all aspects of the experiment, and three others augmented a reading control group, participating only in the reading and testing. For the purpose of the entire campaign, the participants were divided into groups of 10 and were tested in 5-day blocks over a 6-week period. More than 80% of participants had completed high school or had some college education, and a bachelor’s degree was the highest level of education, held by two participants. The first language of all participants is unknown. All participants had normal or corrected to normal vision and where prescribed, used glasses during reading and testing. All participants provided informed consent and the research was done in compliance with DRDC’s Human Research Ethics Board.

CANLEAP exercise activity

The Canadian Load Effects Assessment Program (Human Systems Incorporated, Guelph, Canada) is a calibrated series of instrumented physical obstacles and challenges derived from actual combat tasks. Activities on the course include crawling through tunnels, sprinting, stair and ladder climbs, agility running around and over low obstacles, clearing window and wall obstacles, bounding rushes, navigation on a raised, angled balance beam with integrated obstacles, low crawling, and a casualty drag. In addition to the obstacles, there are lifting and transfer tasks, and a test of marksmanship (Figure 1). CANLEAP is used to gain an understanding of how the soldier’s ensemble impacts movement, but is also very physically demanding, with a brief duration, typically completed in under 6 minutes. The CANLEAP presents occupationally relevant obstacles, through which participants are instructed to give 100% effort. The course has been used in numerous studies because of its relevance (Brown & Mitchell, 2019; Heinrich, 2016; Karakolis et al., 2017; Mitchell et al., 2016). For the present experiment, we only used performance data from the obstacles, and not the weight transfer or marksmanship challenges.

Figure 1. The layout of the CANLEAP obstacle course.

The troops in the 2015 winter campaign wore CAF-issue clothing, helmet and boots weighing nearly 5 kg, as well as full fighting order: a tactical vest with 4 magazines, a full canteen, 2 fragmentation and 2 smoke grenades, a personal radio, and ballistic eyewear. They also carried a C7 rifle with a sling and magazine. The total weight of all worn and carried gear was more than 23 kg. Participants in the present experiment performed the CANLEAP task up to 10 times throughout their 5 days in the campaign.

CANLEAP measures

During each CANLEAP attempt, heart rate (HR) data was collected using Polar RS 400 Heart Rate Monitors sampling at 1 Hz, and kept for analysis off-line. When the reliability of the heart-rate data for a CANLEAP run was lower than 70%, the data was excluded from the mean in Table 1, Table 2. For each attempt, time to completion was measured using photoelectric sensors on the course. Self-assessed level of effort was measured using the Borg Rate of Perceived Exertion (RPE), a Likert-like scale from 6 to 20, where 6 is equal to no exertion and 20 is equal to maximal exertion, or exhaustion (Borg, 1982).

Table 1. Comparison of readability scores for learning material. The scale graphic is provided in the online test to show the rating relative to the scale of the readability measure

Test/Measure	SOF	Pensions
Flesch Reading Ease	50.8	59.3
Automated Readability	11.1	10.8
Flesch-Kincaid Grade Level	9.8	9.8
Coleman-Liau Index	13.7	10.7
Gunning fog Index	14.7	13.9
SMOG Index	12.8	12.5
Characters	18,998	16,849
Words	2893	2795
Complex Words	640	427
Syllables	4832	4220
Sentences	199	143
Syllables/word	1.67	1.51
Words/sentence	14.54	19.55

Table 2. Comparison between course completion values for the 2015 Winter Campaign and results for Mitchell et al. 2016 ǂ Standard deviations were estimated from graphed SEM values

	Weight worn	Weight born	Mean Peak HR on course	Mean RPE (SD)	Mean Completion Time (SD)
Winter Campaign	4.8 kg	18.3 kg	183.67 (9.5)	13.39 (2.27)	243.5 (49.07)
Mitchell et al. (2016)	4.8 kg	0 kg	N/A	15 (2.17ǂ)	222.45 (27.54ǂ)

Demographics, anthropometry, and MHR and estimated V0_2max

In a separate portion of the experimental campaign, participants provided demographic information including age, sex, and education level, and were measured for height, weight, waist circumference, and body composition using both skin-fold calipers and ultrasound measurements. Additionally, maximum heart rate (MHR) was measured using a Polar RS 400 Heart Rate Monitor in a separate step test procedure specified by the modified Canadian Aerobics Fitness Test (mCAFT), while VO_2max values were estimated for each participant based on those values (Weller et al., 1993). The estimated VO_2max measures for all participants were within 20^th – 59^th percentile of a comparison population, meaning that all participants were of moderate cardiorespiratory fitness according to the criteria employed by Chang and colleagues (Chang et al., 2012). Height, weight, body mass index (BMI), fat percentage by ultrasonography, MHR and VO_2max were data of interest in the current experiment.

Reading materials

All reading and testing materials were presented on an Internet Explorer web browser on an internet-connected laptop computer. These materials were presented using FluidSurvey online questionnaire software (http://fluidsurveys.com). Participants were provided with a web link on the computer. A student engagement questionnaire, also presented on FluidSurvey, was provided following the completion of the final test.

The reading materials for this study were taken directly from the Canadian Armed Forces website (http://www.forces.ca). Readings were matched for overall length within 100 words (about 3% difference in length) but differed in subject matter, with one reading taken from the web page for Special Operations Forces Command (SOFCOM) and the other taken from the page for pension (Pension) information. We anticipated that participants would find the SOFCOM subject matter to be more engaging. Varying the salience of the reading materials provided a means to determine whether the activity intervention was responsible for any difference in performance, or whether the reading material itself was not sufficiently engaging to be retained after the activity. If the tests for both subject matters suffered or improved, then the activity could be construed as having an effect, but if only one subject matter showed an advantage, then the effect may be attributable to the subject matter.

An online readability analysis (http://www.editcentral.com/gwt1/EditCentral.html) revealed that the text for both groups is comparable for Flesch reading ease, Automated Readability Index, Flesch-Kincaid grade level, the Gunning Fog Index and the Smog Index, but the Coleman-Liau index is higher for the Special Forces reading (Table 1), most likely because of the frequent use of acronyms in that text.

Reading tests

For each passage of text, a 25 question, multiple-choice test was developed to measure participants’ ability to retrieve the read information. Some questions required more than one correct answer, and each test had the same number of distractors, for a total of 35 correct answers out of 120 possible answers (85 distractors). The tests were matched for structure and complexity of questions, with equal numbers of questions testing for knowledge or understanding levels (Bloom, 1956) of the material. Five additional general reading comprehension questions about the material were shared between the two tests, with only minor variations in wording.

Measures of student engagement

To test student engagement we used the Motivated Strategies for Learning Questionnaire (MSLQ) (Pintrich, 1991). The version of the questionnaire used in this study is a 44-item survey using a 7-point Likert rating system. The five subscales in the MSLQ are Self-Efficacy, which measures the rater’s self-assessment of their ability to master the material; Intrinsic Value, which measures the rater’s internal motivation to learn as an end in itself; Test Anxiety, which measures the rater’s affective self-assessment of their worry and emotional state with respect to performance; Cognitive Strategies, which measures the rater’s deliberate approach to understanding, rehearsing and retaining information, and Self-Regulation which measures the rater’s combined metacognition and effort (Pintrich & De Groot, 1990).

Procedure

Participants provided informed consent and were pseudorandomly assigned to one of the three timing groups in terms of the scheduling of the physical activity. Within each timing group participants were also pseudorandomly assigned to Pensions or SOFCOM subject matter groups (discussed below). Each was seated at a laptop computer and shown a passage of text on the FluidSurvey website and told that they would later be tested on the information in the passage. One hour after reading, the participants were given an online multiple-choice test on the same computer (early retention). The test assessed participant recall, understanding, and ability to make inferences from the text they had read. The software would not permit the participant to continue until each question was answered. Forty-eight hours after the beginning of the first test participants were retested using the same questions in the same order (late retention). The maximum time allowed for each reading or testing session was 25 minutes, with an additional 5 minutes of administrative time. Participants who finished before the maximum time were asked to wait quietly. (Reading time mean: 15:52, max. 23:29; Test 1 mean: 11:59, max. 18:49; Test 2 mean: 7:49, max. 14:20). Participants were given no feedback on test success.

Additionally, participants performed the CANLEAP course, an acute physical challenge taking a mean time of 4.8 minutes for CAF combat arms soldiers to complete. The course raises participant heart rate to approximately 80% the estimated MHR even a for 20 year-old participant. The physical activity was scheduled to occur in one of three time periods: within a one hour window before the learning block (Before Learning; BL), within a one hour window between the learning block and the first testing block (Before Testing; BT), or outside of a time window spanning from 24 hours before learning until one hour after the early retention test. (Outside Window; OW). Additionally the BL and BT groups performed at least one instance of the CANLEAP in the 48 hour period between the early and late retention tests (Figure 2).

Figure 2. Experimental schedule. Three groups learned then were tested on material. The before learning group also performed the CANLEAP obstacle course within the hour preceding learning, while the before testing performed CANLEAP within the hour preceding early retention testing. The outside window group performed CANLEAP at least 24 hours before learning.

To control for the possibility that reading the material did not improve test scores, a subgroup of six participants was tested on both reading topics, but had read only one. Scheduling of the entire campaign allowed only three of our initial 48 participants to be included in this group, so we augmented with three additional participants who had not performed the CANLEAP course, but met all other inclusion criteria for the experiment. The control group read either the SOFCOM or Pension material, and were given early and late retention tests on the material they read. Additionally, immediately following the early retention test, they were also tested on the material they had not read, evenly divided between SOFCOM and Pension.

Data analysis

Test results were exported from the FluidSurveys website to comma-separated values files. Microsoft Excel was used to calculate a total test score and percentage correct for each participant. These values were then imported into SPSS statistical software for further analysis. For the MSLQ, survey scores were exported from the FluidSurveys website directly to an SPSS file where a custom script was generated to flip reversed scales and to calculate a value for each subscale of the MSLQ for each participant.

The data was organized to compare within participant performance on the early and late retention tests as a repeated measure, and between groups for timing of activity (Before Learning, Before Testing, Outside Window); as well as grouping for the subject matter read by each group (pensions or SOFCOM). We performed a mixed Analysis of Covariance (ANCOVA) with a “fitness-effort” covariate factor, described below.

V0_2max, rate of perceived exertion and time on course

To control for the possibility that the level of participant fitness in comparison to their effort on the CANLEAP course indirectly influenced test outcomes, the participants’ dimensions of estimated VO_{2 max} score, Borg RPE, and amount of time spent on the CANLEAP course were reduced to a single “fitness-effort” variable using the principal component analysis feature of SPSS. The Kaiser-Meyer-Olkin measure for sampling adequacy was .53, and Barlett’s test of sphericity was < .001. VO_2max alone accounts for 43% of variability in the model. Although it would have been optimal to use the MHR reached on the CANLEAP in this analysis, more than one-third of the heart-rate samples collected did not meet the minimum 70% data confidence level for inclusion in the analysis.

Results

Anthropometry and CANLEAP

MHR and VO_2max performance

A reliable, separately measured MHR for each participant (mean 192.1, S.D. 3.8 BPM) was used to estimate VO_2max. Using an independent samples, two-tailed t-test, we compared the estimated aerobic performance of the experimental group against a measured VO_2max in a population of 118 male CAF members using a TrueOne 2400 Metabolic System (Sandy, Utah). Levene’s test for equality of variances was not violated. The mean estimated VO_2max for the experimental group was 45.23 (SD 5.3) ml/kg/min in contrast to the larger population which was 47.08 (SD 7.5) ml/kg/min. The t-test revealed no reliable difference between the experimental group and the larger population. Additionally for the experimental group, a Pearson’s correlation test showed no relationship between VO_2max and early retention test performance. Finally, for the experimental group, a one-way ANOVA with activity group as factor showed no difference between the VO_2max scores of any group.

CANLEAP performance

For the available data, the mean peak HR on the course was 184 BPM, the mean Borg RPE was 13.4, and the mean time to completion was 243.5 seconds. The results are compared above with a different experiment where soldiers did not wear the full fighting order, but only the tactical clothing (Table 2).

Reading test performance

Test performance and activity timing

The scheduling of the overall experimental campaign precluded precise control of the timing of the physical activity relative to exposure to the material or testing. We calculated the offset from the start of the activity to the beginning of reading for both the reading and Outside Window groups, and to the beginning of the testing for the before testing group (Table 3). An independent samples t-test revealed no differences between the offset for the BL and BT groups, and there was no correlation between the offset and score on the first test for those groups.

Table 3. Time offset from commencement of the obstacle course relative to commencement group-dependent reading or testing

Condition	Mean	Std. Dev.	Minimum	Maximum
Before Learning	0:36:41	0:14:33	0:13:50	1:03:28
Before Testing	0:32:27	0:14:26	0:14:06	1:04:23
Outside Window	19:29:32	1:18:17	17:28:23	21:49:15

A comparison of 2 subject matter (SOFCOM or Pensions) × 3 physical activity timings (BL, BT, or OW) × 2 testing periods (repeated 1 and 48 hours after the learning period) mixed ANCOVA was performed on test scores. Levene’s test for homogeneity of variances was not significant. The above derived fitness-effort score was used as a covariate factor. We found a test period by physical activity timing interaction (F(2,42) = 3.396, p = .049, η² = .14). We performed post hoc tests with a modified Bonferroni correction (adjusted p of .016). In the early retention test, the performance of the BT group was worse (mean, 61.1%) than the OW group (mean, 72.5% p = .012), and trended worse (p = .025) than the BL group (Figure 3a). These differences were not present in the late retention test. The covariance measure was not significant: in other words, the prior fitness of the participants did not significantly factor in any differences of performance among participants on the tests in this study.

Figure 3. (a) The effect of physical activity timing on early and late retention test scores: Mean early retention scores: BL 68.4%, BT 61.1%, OW 72.5%. (b) The effect of subject matter on test scores: Pension 55.9%, SOFCOM 76.5%. Error bars represent ±1SD.

Test performance and subject matter

There was also a main effect of subject matter (F(1,41) = 46.4, p < .001, η² = .12) with the test scores of learners in the SOFCOM group being higher (76.5, SD 10.24) than those in the pensions group (55.9, SD 10.3, Figure 3b).

We considered the possibility that participants may not have learned anything at all from the reading, and that any degree of success was based on prior knowledge. Using a paired t-test, we compared the performance of our control sample between the material they did not read (naïve) and did read (exposure). Their performance on the naïve condition was worse (51.4%, SD 10.38) than the exposure condition (74.7%, SD 12.70) t(5) = 3.48, p = .018.

Motivated strategies for learning questionnaire

We performed bivariate correlations between each subscale score and early retention test scores for each participant. Test Anxiety correlated negatively with the test score, and all other measures were positively correlated (Self-Efficacy r(47) = .432, p = .002; Intrinsic Value r(47) = .411, p = .004; Test Anxiety r(47) = −.350, p = .013; Cognitive Strategies r(47) = .303, p = .039; Self-Regulation r(47) = .357, p = .014).

Finally, to determine if subject matter or physical activity timing influenced student engagement measures, we performed separate two-way ANOVAs on each of the MSLQ subscale measures with subject matter and physical activity timing as between-groups factors. Participants in the SOFCOM reading group rated Self-Efficacy F(1,41) = 10.99, p = .002, Intrinsic Value F(1,41) = 10.76, p = .002, and Cognitive Strategy F(1,41) = 5.27, p = .038 higher than the Pension reading group, but there was no difference for the Test Anxiety or Self-Regulation Measures. In no instance did physical activity timing impact MSLQ scores or interact with subject matter to impact MSLQ scores (all p > .1).

Discussion

Soldiers studied a portion of text, and then were tested on the material up to one hour after, and again 48 hours after that. They also performed the CANLEAP obstacle course before learning, before testing, or the day before the learning. As predicted by our first hypothesis, there was no benefit for the group that performed the high intensity physical challenge just before learning. Soldiers who performed the CANLEAP activity immediately before the first test performed more poorly on that test than those who did not did perform the CANLEAP activity in the preceding 24 hours. This result contradicted our second hypothesis, that the CANLEAP activity would have no detrimental effect during testing. However, all three groups performed equally well after 48 hours, with no main difference between performance on the early and late test. The group that performed the activity prior to learning did not show improved performance on the test. The fitness of the participants, the timing of the physical activity, and the kind of physical activity might have all contributed to the current results.

Fitness and performance

Aerobic fitness and cognitive performance

The subjects in this study all fell within the “moderately fit” range (Chang et al., 2012), with an estimated VO_2max between the 20^th and 59^th percentile of a larger sample population. The fitness-effort variable in the ANCOVA analysis composed of estimated VO_2max, Borg RPE, and time on course did not show a relationship to the test scores, possibly because all participants were at the same fitness level. In other words, participant fitness as measured by VO_2max, in combination with the effort and time spent on course does not explain the detriment or lack of benefit provided by physical activity. Additionally, VO_2max scores, which accounted for most of the variance of the fitness-effort measure, showed no correlation with the time needed to complete the course. Taken together, the analysis indicates the test scores are related to when the activity occurred, rather than fitness levels. These results support our first hypothesis, that soldiers with a moderate fitness level would not enjoy improved learning after an acute bout of activity.

Timing and type of activity

Timing of physical activity relative to cognitive activity

In the present study, using the measured participant MHR (mean 192 BPM) and their peak heart rate on the course (mean 184 BPM), almost every participant was operating at over 90% of their MHR. The mean level of exertion for the current experiment was 83% MHR using Tanaka’s age-predicted heart rate formula (MHR = BPM/(208-age*0.7)) (Tanaka et al., 2001). In the present instance, the BT group performed the cognitive test as early as 13 minutes after exercise, with a mean delay of 32.5 minutes, however there was no net benefit in performance for that group.

Constraints of the overall experimental campaign required that the learning block and early retention test block be approximately 1 hour apart. The physical activity for the “before testing” group was performed during this interval, meaning the activity was also “after learning.” The question may be posed whether the physical activity impeded encoding of the studied material, or whether the physical activity inhibited retrieval of the already learned material. The mean offset between the physical activity and cognitive activity was the same (just over 30 minutes) for both the Before Learning and Before Testing groups. The minimum offset for both groups was greater than 13 minutes. While the BT group performed more poorly than the OW group on the first test, their performance was equal to the other two groups 48 hours later, suggesting that they did not suffer from poorer encoding of the information.

Many studies directly exploring learning place the exercise block in relationship to the learning block, but not in relationship to recall or retention blocks (Hillman et al., 2008). An experiment by van Dongen and others showed acute improvements for associative learning when 35 minutes of exercise (~80% MHR) was performed four hours after learning, but not when performed immediately after learning. In a condition where the participants exercised immediately after, there was no benefit to learning, but also no learning decrement when tested 43 hours after the last exercise period. (Van Dongen et al., 2016). Different researchers compared a learning block after either a “relaxed” aerobic condition, or an “intense” exercise condition. The intense condition featured two sprints of three minutes, each increasing in speed to the point of exhaustion. Participant heart rates were in the range of 163–202 BPM (median 184 BPM). After the intense condition participants had a 20% better retention rate for vocabulary in an invented language when they were tested at one and two weeks after the initial learning (Winter et al., 2007). With a longer delay post-activity than the present experiment, learning improves. One possible explanation of the present results is the proximity of the intervention to the exposure period which may impact blood lactate levels, known to affect cognitive performance in some cases (Coco et al., 2009).

Conversely, a fatigue-based perspective suggests that prolonged exercise engages cognitive and physical effort in a top-down fashion, and bodily sensory processes in a bottom-up fashion. Taken together, these processes mean that executive function has a dynamic relationship to exercise (Schmit & Brisswalter, 2018). While the authors of that review limit their discussion to prolonged exercise, it is conceivable that any exercise providing sufficient cognitive challenge could also impair recall.

Type of physical activity and type of cognitive activity

The specific type of cognitive performance relative to the activity might play a moderating role in the rejection of our second hypothesis. The physical activity in this experiment was explicitly designed to replicate the challenges faced by soldiers, and differs considerably from effort on a treadmill or cycle ergometer, regardless of the intensity level of the exercise. In completing all CANLEAP obstacles, each participant faced 25 distinct physical challenges with a maximum of only 51 seconds on any given task, even for the slowest participant. Some research suggests that exercise type – not just intensity – exerts influence over exercise-related cognitive gains (Pesce, 2012), a difference that has been demonstrated at a cortical level in animals (Black et al., 1990). In human experiments, behavioral differences based on exercise type have also been observed.

While no measures of cognitive load have been examined for the CANLEAP course specifically, recent research has demonstrated that walking on a path with 15 cm-high traffic cone obstacles is considerably more cognitively demanding than walking on either narrow or wide paths (Lin & Lin, 2016). Given the agility challenge of the CANLEAP course, the imposed level of cognitive load and need for sustaining attention would likely be quite high. The process of sustaining attention is stressful in itself (Szalma et al., 2004), and it is possible that the stress level imposed by the activity on the obstacle course causes neurophysiological effects greater than elevation of blood lactate levels. High-intensity exercise elevates concentrations of norepinephrine and dopamine, and particularly in the presence of corticotropin releasing hormone, cause a cascade effect of reduced signal and more noise in the pre-frontal cortex which impedes performance of central executive tasks (McMorris, 2016).

Additionally, the hypofrontality hypothesis proposes the considerable increase in motor-specific brain metabolism and widespread activity during exercise consumes finite resources, thus reducing activity in the frontal lobe. There is evidence that no additional metabolic resources are available to the brain during exercise, since the total volume of blood to the brain remains constant with a greater percentage/heartbeat of blood being shunted to the skeletal muscles (Dietrich, 2006). Researchers suggest this is why persons of lower fitness have poorer cognitive performance during exercise, and no improvement after (Chang et al., 2012). In the same way, the level of intensity may also explain why the soldiers did not see the kinds of gains experienced by children who engaged in play-like activity prior to learning (Ishihara et al., 2017; Koutsandreou et al., 2016).

The learning and testing blocks both required reading, known to activate the left middle and superior temporal areas (Price et al., 1994). By contrast, successful encoding (learning) and retrieval (testing) of learned information are known to activate different areas of the brain including the hippocampus and hippocampal gyrus, while specific areas of the prefrontal cortex are involved in aspects of both processes (Kensinger et al., 2003; Prince et al., 2005). It is possible that in concert with the physical load, these factors may cause a performance decrement in the activity before testing group due to the specific nature of the cognitive structures involved in retrieval, as well as exercise-induced hypofrontality.

In a design similar to the present experiment, other researchers tested the effect of acute exercise on both learning and recall for long-term memory of a heard passage of text. Participants used a cycle ergometer at moderate intensity for 30 minutes immediately prior to exposure or recall. The group exercising prior to exposure improved in contrast to either the group exercising prior to recall, or a control group (Labban & Etnier, 2011). It is important to note that in the Laban and Etnier investigation participants were not required to recall cohesive portions of information. While the intensity and type of physical activity was different in that experiment from the present, the researchers found no advantage for the recall group. The findings are consistent with the present results, and may be explained as an interaction of types, intensities, and timings of physical activity relative to specific types and timings of cognitive activity. Further explorations of these factors within soldier and general populations are needed to understand their specific effects and how they may vary among different groups.

Student engagement and subject matter

All measures of engagement except for test anxiety correlated positively with test performance. In other words, the more engaged the participant was with the learning, the better their performance. The participants who studied the SOFCOM subject matter performed better, even though the text was more difficult according to one index of readability. While every soldier contributes to a pension plan and is eligible to receive pension benefits, very few will ever be members of the special operations forces. One might argue that information about pensions “matters” more, but that information about special operations is “more interesting.” This was borne out in the MSLQ test. While measures of Test Anxiety and Self-Regulation (the student’s thinking about their thinking, and effort to learn the material) were the same for both groups, Self-Efficacy, Intrinsic Value and Cognitive Strategies were rated as more positive for the SOFCOM group. In this instance, participant engagement was directly linked to the subject matter, however the timing of the physical activity did not change the degree to which participants were engaged with either topic. In other words, the subject matter did not increase or decrease in salience in relationship to the activity, so any affective component conferred by the physical activity did not impact the appetite to learn the material. Importantly for the present experiment, poorer results for the BT group cannot simply be attributed to a failure to find the information sufficiently interesting to retain following the activity.

Limitations and future work

Because soldiers often follow intense activity with life and death decision-making based on recently acquired information, the links between activity and cognition are not trivial. Additional research is needed to elucidate the relationships among types and timings of activity in relationship to types of cognition.

The design of this study was constrained because it was scheduled within a larger group of experiments using the same participants, which restricted options for control and timing of physical activity, and the type of physiological tests that could be performed in compliance with an omnibus ethics protocol. Future work should include explicit testing of anaerobic capacity, particularly of lactate threshold, and salivary cortisol, as a measure of stress. The lactate and cortisol levels should be measured before and after the physical activity, and before and after learning and testing. Additionally, a comparison between the attentionally demanding CANLEAP course should be made against a metabolically equivalent treadmill or ergometer activity. With the knowledge that subject matter does not alter the impact of the physical activity, only one subject matter should be presented and tested, thus improving statistical power. Finally, the material to be learned should be a video recording of a pre-mission brief, with recall of the details as the test condition. This would better mimic operational conditions and improve ecological validity.

Conclusion

The effects of physical activity on cognition are complex, and depend on many factors. Standard measures of fitness such as BMI and VO_2max may not be adequate predictors of cognitive performance for Canadian infantry soldiers, since BMI correlates poorly with body fat for that group, and many activities for soldiers require anaerobic rather than aerobic fitness. Furthermore, the complexity of movement in diverse, contested environments may impose cognitive costs. Retrieval of a read passage of text was impaired immediately following vigorous activity, but not 48 hours later. Test performance was better for a group learning more salient subject matter, regardless of the timing of the physical activity.

Disclosure statement

No potential conflict of interest was reported by the author.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.