Effects of Cognitive and Physical Load of Acute Exercise on Inhibitory Control and Prefrontal Cortex Hemodynamics in Children

ABSTRACT

Introduction

Evidence suggests that acute exercise benefits inhibitory control, but it remains unclear how physical and cognitive load influence this ability in preadolescent children, when faced with different types of distractors. We examined their moderating effects along with changes in prefrontal cortex hemodynamics.

Methods

Using a crossover design, 29 participants completed 10 min of exercise (with an interactive training wall) at 1) low physical and low cognitive load, 2) low physical and high cognitive load, 3) high physical and low cognitive load, and 4) high physical and high cognitive load in a randomized order. Before and after each condition, they performed a Standard and Affective Stroop task. During both tasks, the lateral oxygenation difference in the prefrontal cortex was measured by using functional near-infrared spectroscopy.

Results

For the Standard Stroop task, there was no effect of physical and cognitive load on performance. In contrast, exercise with low compared with high cognitive load was associated with a greater reduction of reaction time and increase in accuracy on the Affective Stroop task. This was accompanied by a decrease in lateral oxygenation difference on trials with low inhibitory control demands.

Conclusions

Acute exercise with low compared with high cognitive demand benefits the ability to resolve emotional conflict, but the control over inhibition of nonemotional information remains unaffected. This effect of cognitive load is complemented by an increased efficiency of the left prefrontal cortex, when no emotional conflict resolution is required.

Inhibitory control serves as a cognitive ability that is crucial in learning environments as it allows the suppression of a potent distractor while processing incoming information (1). The distraction can be caused by task-irrelevant information (e.g., color word–color interference on a Stroop task) and emotional interference (e.g., emotion word–face interference on a Stroop task). In the educational setting, preadolescent children typically face both attentional (e.g., distractions in the classroom) and emotional (e.g., problems in teacher–student relationships) conflicts (2,3). Given these challenges, it is reasonable that inhibitory control contributes to peer acceptance at school (4), on-task behavior in the classroom (5), and academic performance (6). In search of a strategy to promote inhibitory control in preadolescents, much potential for a temporary enhancement of this ability has been attributed to the implementation of exercise activities during school time.

In this respect, meta-analytical findings from all age groups suggest that after a single exercise session, small-to-moderate improvements for different components of executive function can be expected (7–9). Experimental evidence further suggests that the benefits for inhibitory control in particular are influenced by different exercise characteristics, such as intensity and cognitive demands (10,11). However, it should be noted that investigations of the aftereffects of exercise on this cognitive ability have been limited to tasks using nonemotional information as distractors. Studies that prescribed exercise sessions of varying intensity levels to children and adolescents found that higher intensities were more likely to elicit improvements for inhibitory control in a subsequent cognitive task than lower intensities (12,13). This pattern of results is in conflict with the direction of effects that has been proposed by an evolutionary model based on exercise-induced changes in cortical structures subserving inhibitory control. The reticular-activating hypofrontality (RAH) theory posits that exercise engages arousal mechanisms at low to moderate intensity, but disengages prefrontal cortex (PFC) functions, such as inhibitory control, to compensate for high demands on the motor cortex at high intensity (14).

Predictions of the RAH model can be tested with the Stroop task, which requires the selection of a correct response despite conflicting information (e.g., name the word “blue” although it is written in a different color). Activation likelihood estimation meta-analyses have shown that this task is prefrontal-dependent, with a laterality effect indicating a dominance of left dorsolateral PFC during processing of nonemotional and emotional interference (15,16). Findings generated with functional near-infrared spectroscopy (fNIRS) further support that processing of conflicting stimuli leads to a greater cerebral oxygenation of the left compared with right frontal lobe (17). Conversely, a greater lateral oxygenation difference (LOD) indicating dominance of the left hemisphere has been associated with better performance on the Stroop task (18) as well as postexercise improvements in task that tap into the same cognitive domain (19). In adults, both mild (30% V̇O_2peak) and moderate (50% V̇O_2peak) exercises were followed by an increase in oxygenated hemoglobin in the left compared with right PFC during a Stroop task, which contributed to a temporary improvement in inhibitory control (20,21). In contrast, high-intensity exercise (80% heart rate reserve) led to decreased bilateral PFC activation in response to the same task and lower inhibitory control than exercise at moderate intensity (50%–60% heart rate reserve) (22).

The effect of exercise intensity might interact with its cognitive demands. The RAH theory mainly applies to exercise types with cyclic movements (14) as exercises requiring high executive function will not allow for the disengagement of the PFC. In this respect, the combination of physical exercise and cognitive effort led to a greater increase of oxygenated hemoglobin across different regions of the PFC and performance on the Stroop task than physical or cognitive exercise alone (23). Further experimental evidence from adults suggests exercise, which is complex or characterized by higher cognitive demands, to induce greater performance gains on inhibitory control tasks than simple exercise types (24,25).

Despite the indications that both physical and cognitive demands of acute exercise exert an immediate influence on inhibitory control, inconsistent results and a lack of studies investigating neural underpinnings complicate conclusions on how exercise should be designed to temporarily enhance this ability in children. This is further supported by developmental differences that suggest different reserves for changes in inhibitory control between children and adults. In this respect, children show a lower PFC oxygenation response to conflict tasks, such as the Stroop Color–Word, due to the immaturity of underlying cognitive functions (26). This immaturity is more pronounced for cognitive abilities required to perform task with affective content (27), suggesting that there is a greater reserve for exercise-induced improvements of inhibitory control of emotional compared with nonemotional information.

We compared the effects of cognitive and physical demands of exercise as well as their interaction on the inhibition of nonemotional and affective information. The examination of behavioral performance on standard and affective variants of the Stroop task was complemented by the investigation of task-related changes in cortical hemodynamics in the PFC. Based on the literature in adult populations (20–22,24,25), we expected both exercise with high cognitive demand and exercise with low physical demand to elicit greater benefits for inhibitory control than other combinations.

METHODS

Participants

Children aged 9 to 13 yr with right-hand dominance and corrected to or normal vision were recruited from a local school in Basel using written information. Participants receiving special educational services related to cognitive impairments or attentional disorders, undergoing a therapy for a mental disorder, or reporting color blindness or any medical condition, which poses an increased health risk during exercise activities, were excluded. The required sample size was calculated a priori using G*Power 3.1 (28). Based on previous experimental studies and meta-analyses, effect sizes reported for the acute benefits of exercise for inhibitory control ranged from moderate to large (7–9). Assuming a moderate effect size of f = 0.25, an alpha level of P = 0.05, a correlation between measures of r = 0.60, and the use of repeated-measures ANOVA, 28 participants were required to reach 80% power. Recruited participants expressed their willingness to take part in the study, and written informed consent was collected from their legal guardians. The study protocol followed the principles outlined in the Declaration of Helsinki, and it was approved by the local ethics committee.

Study design

Using a crossover design, participants completed four experimental conditions (5–7 d apart) in a randomized order in a school setting. Blocked randomization was used, with the list generated by sealed envelope. The conditions encompassed game-based exercise with variable physical and cognitive loads induced by an interactive training wall, and the allocation to the conditions was concealed. Before and after each exercise bout, participants completed a Standard and Affective Stroop task. Brain oxygenation responses were recorded simultaneously during the task by applying fNIRS. Moreover, participants filled out the Family Affluence Scale, the Strengths and Difficulties Questionnaire, and the Insomnia Severity Index. The assessments were performed between 8 and 11 am in dimly lit room, with different conditions within participants completed at the same time of the day. Participants were instructed to refrain from exercise within 24 h and to have their last meal 45 min before the assessments.

Intervention

Participants performed four experimental conditions, with each condition involving a 10-min exercise bout that varied in physical and cognitive load: low physical and low cognitive load (−P −C), low physical and high cognitive load (−P +C), high physical and low cognitive load (+P −C), and high physical and high cognitive load (+P +C). The exercise interventions were administered in a separate room within the school where participants were recruited from, using the SmartFit© Single (Ventura, CA) training device. This interactive stationary training wall is equipped with nine roughly hand-sized, computerized, touch-sensitive targets, which are able to show different shapes and colors. Different cognitive tasks were programmed to induce varying levels of cognitive demands. Conditions with low cognitive load mainly included visual tracking tasks (−P −C; +P −C), in which participants had to touch numbers, letters, emoticons, and symbols that appeared at different positions and in a randomized order (Supplemental Table 1, Supplemental Digital Content, http://links.lww.com/MSS/C958, Grand average of fNIRS waveforms at left and right PFC for compatible and incompatible trials in the Standard and Affective Stroop task). Tasks demanding higher-order cognition were used in conditions with high cognitive load (−P +C; +P +C). This included a variant of the Go/No-Go task, Paired Associates, a pattern recognition paradigm, and mental arithmetic, with each task requiring a motor response (Supplemental Table 2, Supplemental Digital Content, http://links.lww.com/MSS/C958, Comparison of changes in concentration of oxygenated and deoxygenated hemoglobin in response to compatible trials of the standard blocks of the Stroop task before and after the low and high cognitive load exercise condition displayed for the left and right PFC). The type of movement and motor response during the cognitive tasks was modulated depending on the condition. Participants were required to perform cognitive tasks in standing position or with a ball on the low physical demand conditions (−P −C; −P +C), whereas cognitive tasks were combined with jumps, sprints, and bodyweight exercises to induce high physical effort (−P −C; +P +C). At the end of each exercise session, participants rated their perceived physical exertion (RPE) and reported cognitive effort using the NASA Task Load Index (TLX).

Standard and Affective Stroop task

A modified, computer-based version of the Stroop task, which included both standard and affective blocks, was administered with E-Prime 3.0 (PST, Pittsburgh, PA). Visual stimuli used in standard blocks included color words written in the same (compatible trials) or different color of ink (incompatible trials). On affective blocks, faces showing angry, happy, and sad expressions were overlaid with adjectives expressing the same (compatible trials) or a different emotion (incompatible trials). Participants were instructed to press a button corresponding to the color the word is written in or the emotion expressed by the word. In both variants of the Stroop task, they completed two practice rounds, followed by four compatible and incompatible blocks that were alternating. Each block contained 15 trials and started with a fixation period of 18 to 20 s. Trials commenced with a fixation cross presented for 300 to 400 ms (random variation). Subsequently, visual stimuli were displayed against black background for 1000 ms or until a response has been made. The intertrial interval was adjusted based on the response time, so that the number of trials presented within a block was consistent across conditions and participants. Mean reaction time on response-correct trials and accuracy rates were extracted for both trial types and task variants.

fNIRS recordings

A continuous-wave, dual-wavelength (760 and 850 nm) fNIRS system (NIRSport; NIRx Medical Technologies, Berlin, Germany) was used. Fifteen optodes (8 illumination sources, 7 light detectors) were mounted in an individually adjustable headband and placed over the PFC. The headband was positioned so that source 2 was situated on the Fz position according to the 10:20 system. The montage resulted in 22 channels, with 7 representing the left and 7 representing the right hemisphere (Fig. 1). Rigid spacers ensured that neighboring sources and detectors were equidistant at 3 cm, which is considered the best trade-off between light penetration depth and signal-to-noise ratio (29). Participants wore a loosely attached, optically opaque overcap to prevent ambient light from affecting the measurement. Data were recorded at a sampling rate of 7.8125 Hz with the software NIRStar 15.2 (NIRx Medical Technologies). Before each measurement, signal level optimization was performed and optodes were adjusted until the signal quality was acceptable. The differential pathlength factor was modified depending on participants’ age (30).

FIGURE 1.

Placement of fNIRS optodes in relation to the 10:20 system. Notes: Red numbers indicate sources, and blue numbers indicate detectors. The LOD was calculated by subtracting oxygenated hemoglobin in the right PFC (average of pairs: 1–1, 1–3, 4–1, 4–3, 4–6, 6–3, 6–6) from oxygenated hemoglobin in the left PFC (average of pairs: 3–2, 3–5, 5–2, 5–5, 5–7, 8–5, 8–7).

Processing of fNIRS data

Data preprocessing were performed with Homer3 v1.52.0 (31). Missing values were first replaced by spline interpolation of the existing values. The function hmrR_PruneChannels removed channels with a dRange lower than 500 μV or higher than 1 V, and a quotient of mean(d)/SD(d) less than 6.67 from further analysis. After conversion from light intensity to optical density, motion artifacts were first detected using hmrR_MotionArtifactByChannel (time range, 0.5 s; masked data around motion artifact, 1 s; SD threshold, 25; amplification threshold, 5) and then corrected using spline interpolation (P = 0.99). Surviving artifacts were detected by repeating hmrR_MotionArtifactByChannel, and afterward rejected using hmrR_StimRejection. Subsequently, a bandpass filter with the thresholds of 0.01 Hz (high-pass filter) and 0.09 Hz (low-pass filter) was applied to remove instrumental and systemic noise from the signal (29). The low-pass filter was chosen to eliminate the influence of Mayer waves, which are located at 0.1 Hz, from the signal (32). The resulting filtered data were converted from optical density to concentration changes in oxygenated and deoxygenated hemoglobin using a partial path length factor of 1. Finally, a general linear model was applied to estimate the hemodynamic response function. Therein, ordinary least squares were used as a solving method, a consecutive sequence of Gaussian functions served as a basis function, and a polynomial drift correction of third order was applied. Block averages were calculated for the time range from −3 to 40 s relative to task onset. Hemodynamic response functions were exported, and mean values over the task period (0–25 s) for the right PFC were subtracted from the mean values for the left PFC to derive the LOD. Although LOD served as outcome, the lateral deoxygenation difference (LDD) was also extracted to comply with fNIRS reporting guidelines (33). Higher values indicate greater (de-)oxygenated hemoglobin in left compared with right PFC.

Statistical analysis

SPSS Statistics 28.0 for Windows (IBM, Armonk, NY) was used for statistical analyses. Change scores were calculated within each experimental condition by subtracting pretest values from posttest values for behavioral performance data (reaction time and accuracy) and the LOD. The subsequent analyses were restricted to participants who performed all conditions. For fNIRS data, nine participants had missing data on a single condition, which was estimated by the linear trend at that data point. Results derived from participants with complete fNIRS data on all conditions are compared with those derived from complete conditions analyses. For hypotheses testing, the change scores (pretest data are provided in Supplemental Table 3 (Supplemental Digital Content, http://links.lww.com/MSS/C958, Comparison of changes in concentration of oxygenated and deoxygenated hemoglobin in response to incompatible trials of the standard blocks of the Stroop task before and after the low and high cognitive load exercise condition displayed for the left and right PFC) were analyzed using a repeated-measures ANOVA with condition (−P −C, +P −C, −P +C, +P +C) and trial type (compatible, incompatible) included as within-subject factors. Preplanned mean contrasts were specified using the Mmatrix function of SPSS to examine the effect of cognitive load (−P −C and +P −C vs −P +C and +P +C), physical load (−P −C and −P +C vs +P −C and +P +C) as well as their interaction (−P −C vs +P +C). Reaction times and LOD on compatible and incompatible trials of the standard and affective blocks of the Stroop task served as main outcomes. Accuracy rates on both trial types were also examined to test whether there was a speed–accuracy trade-off for behavioral performance. LDD was analyzed to check differences in responses between oxygenated and deoxygenated hemoglobin. Moreover, the manipulation of RPE and NASA-TLX by the experimental conditions was examined from preplanned contrasts in ANOVA with repeated measures on condition. The level of statistical significance for preplanned contrasts was determined as P < 0.05.

RESULTS

Of 35 recruited participants, 3 were not willing to complete the scheduled sessions after the first experimental condition because of personal reasons and three dropped out because of illness. The characteristics of the remaining 16 girls and 13 boys who completed the study are shown in Table 1. Ratings on NASA-TLX were lower in conditions with low compared with high cognitive load (M_Diff = −1.49; SE_Diff = 0.25; F(1,28) = 38.21, P < 0.001, η² = 0.556), and low compared with high physical load (M_Diff = −0.75; SE_Diff = 0.28; F(1,28) = 7.44, P = 0.010, η² = 0.210). Similarly, RPE was lower in conditions with low compared with high physical load (M_Diff = −4.79; SE_Diff = 0.44; F(1,28) = 119.77, P < 0.001, η² = 0.811). However, conditions with low cognitive load yielded a higher RPE than those with high cognitive load (M_Diff = 0.97; SE_Diff = 0.35; F(1,28) = 7.73, P = 0.10, η² = 0.216).

TABLE 1.

Overview of participants’ characteristics (N = 29).

	Female Participants		Male Participants		All Participants
	M	SD	M	SD	M	SD
Age, yr	11.3	1.1	11.0	1.4	11.1	1.2
BMI, kg·m−2	17.0	5.5	17.5	2.2	17.3	4.2
FAS	7.6	0.9	7.8	1.2	7.7	1.0
ISI	7.1	3.5	6.2	1.9	6.7	2.9
SDQ	12.1	4.3	9.7	3.8	11.0	4.2

BMI, body mass index; FAS, Family Affluence Scale; ISI, Insomnia Severity Index; SDQ, Strengths and Difficulties Questionnaire.

Cognitive performance

With regard to performance on the Standard Stroop task, there was no effect of physical and cognitive load as well as no interaction for reaction time and accuracy, irrespective of the trial type (Table 2). In contrast, participants showed a change toward shorter reaction time on compatible (M_Diff = −45.0 ms; SE_Diff = 20.7 ms; P = 0.039) and incompatible trials of the Affective Stroop task (M_Diff = −60.1 ms; SE_Diff = 23.0 ms; P = 0.014) in conditions with low compared with high cognitive load (Fig. 2). Similarly, exercise with low cognitive load increased accuracy on incompatible trials to a greater extent than exercise with high cognitive load (M_Diff = 4.1%; SE_Diff = 1.9%; P = 0.043), whereas no difference in changes was observed between conditions for accuracy on compatible trials (Table 2). Beyond the effects of cognitive load on reaction time, physical load and the interaction of physical and cognitive load did not influence changes in any behavioral performance outcomes of the Affective Stroop task.

TABLE 2.

Effects of physical load, cognitive load, and their interaction on behavioral performance on the Stroop task and lateral (de-)oxygenation difference measured with fNIRS (N = 29).

Outcome	Task	Trial Type	Low vs High Cognitive Loada		Low vs High Physical Loadb		Low vs High Physical and Cognitive Loadc
			F	η 2	F	η 2	F	η 2
∆Reaction time	Standard Stroop	Compatible	0.05	0.002	0.37	0.013	0.53	0.002
		Incompatible	1.87	0.063	0.02	0.001	1.42	0.048
	Affective Stroop	Compatible	4.71*	0.144	0.28	0.010	1.06	0.036
		Incompatible	6.82*	0.196	0.15	0.005	3.17	0.102
∆Accuracy	Standard Stroop	Compatible	0.14	0.005	0.01	<0.001	0.03	0.001
		Incompatible	0.15	0.005	0.60	0.021	0.06	0.002
	Affective Stroop	Compatible	0.72	0.025	1.81	0.061	0.25	0.009
		Incompatible	4.50*	0.139	0.21	0.008	0.72	0.025
∆Lateral oxygenation difference	Standard Stroop	Compatible	3.76*	0.145	0.01	<0.001	3.67	0.116
		Incompatible	5.04*	0.153	0.42	0.015	2.68	0.087
	Affective Stroop	Compatible	6.70*	0.193	1.96	0.065	0.37	0.013
		Incompatible	2.04	0.068	0.03	0.001	1.42	0.048
∆Lateral deoxygenation difference	Standard Stroop	Compatible	4.91*	0.149	0.94	0.033	0.81	0.028
		Incompatible	1.78	0.060	3.70	0.117	0.32	0.011
	Affective Stroop	Compatible	0.11	0.004	0.12	0.004	0.31	0.011
		Incompatible	0.02	0.001	1.05	0.036	0.59	0.020

∆ = change calculated as posttest − pretest values.

^a −P −C and +P −C versus −P +C and +P +C.

^b −P −C and −P +C versus +P −C and +P +C.

^c −P −C versus +P +C.

*P < 0.05.

FIGURE 2.

Means and confidence intervals of changes in behavioral performance (posttest − pretest) on compatible and incompatible trials of the Standard (left) and Affective (right) Stroop tasks displayed for all experimental conditions. Notes: Positive values denote improvements on accuracy and impairments in reaction time. −C −P = low cognitive, low physical load; −C + P = low cognitive, high physical load; +C −P = high cognitive, low physical load; +C +P = high cognitive, high physical load.

Cortical hemodynamics

The grand averaged waveforms showing the prefrontal oxygenation response to the Stroop tasks are provided in Supplemental Figure 1 (Supplemental Digital Content, http://links.lww.com/MSS/C958, Grand average of fNIRS waveforms at left and right PFC for compatible and incompatible trials in the Standard and Affective Stroop task). The results of the repeated-measures ANOVA (Table 2) revealed that on the Standard Stroop task, conditions with low cognitive load led to a greater increase of LOD (M_Diff = 0.018 μmol·L⁻¹; SE_Diff = 0.008 μmol·L⁻¹; P = 0.038) and LDD (M_Diff = 0.006 μmol·L⁻¹; SE_Diff = 0.003 μmol·L⁻¹; P = 0.035; i.e., increase of oxygenated and deoxygenated hemoglobin in the left relative to right PFC) on compatible trials (Supplemental Fig. 2, Supplemental Digital Content, http://links.lww.com/MSS/C958, Comparison of changes in concentration of oxygenated and deoxygenated hemoglobin in response to compatible trials of the standard blocks of the Stroop task before and after the low and high cognitive load exercise condition displayed for the left and right PFC), but a decrease of LOD on incompatible trials (M_Diff = −0.014 μmol·L⁻¹; SE_Diff = 0.006 μmol·L⁻¹; P = 0.033; Supplemental Fig. 3, Supplemental Digital Content, http://links.lww.com/MSS/C958, Comparison of changes in concentration of oxygenated and deoxygenated hemoglobin in response to incompatible trials of the standard blocks of the Stroop task before and after the low and high cognitive load exercise condition displayed for the left and right PFC). For the Affective Stroop task, the pattern of results was reversed for LOD, meaning that a more pronounced decrease (i.e., increase of oxygenated hemoglobin in the right relative to left PFC) was found in response to compatible trials during conditions with low compared with high cognitive load (M_Diff = −0.021 μmol·L⁻¹; SE_Diff = 0.008 μmol·L⁻¹; P = 0.015; Fig. 3; Supplemental Figure 4, Supplemental Digital Content, http://links.lww.com/MSS/C958, Comparison of changes in concentration of oxygenated and deoxygenated hemoglobin in response to compatible trials of the Affective Stroop task before and after the low and high cognitive load exercise condition displayed for the left and right PFC). In contrast, there was no effect of cognitive load on changes in LOD and LDD on incompatible trials. Moreover, changes in LOD and LDD on both the Standard and Affective Stroop task were not influenced by physical load and its interaction with cognitive load. The pattern of the main results remained unchanged, when the analysis was restricted to participants with complete data on all conditions (see Supplemental Digital Content, http://links.lww.com/MSS/C958).

FIGURE 3.

Means and confidence intervals of changes (posttest − pretest) in the LOD on compatible and incompatible trials of the Standard (left) and Affective (right) Stroop tasks displayed for all experimental conditions (N = 29). Notes: Positive values denote increases in LOD. −C −P = low cognitive, low physical load; −C + P = low cognitive, high physical load; +C −P = high cognitive, low physical load; +C +P = high cognitive, high physical load.

DISCUSSION

The key finding of our experimental study is that the cognitive rather than the physical load of exercise influences inhibitory control, but only if the distraction results from conflicting emotional information (e.g., emotion word–face interference). Exercise with a low compared with a high cognitive demand in particular led to improved performance on the Affective Stroop task, whereas we found no influence of this exercise characteristic for reaction time and accuracy on the standard blocks of the same task.

This pattern of results was not expected, as previous findings on the role of cognitive and physical load for exercise-induced benefits for inhibitory control have solely been generated by studies using tasks that require the resolution of a nonemotional conflict. With regard to the influence of physical load, higher exercise intensity was associated with increased inhibitory control in a subsequent cognitive task in previous experiments with adolescents (12,13), whereas the direction of effects was reversed in adults (20,21). Our results do not support these patterns, as independent of the type of the Stroop task, physical load did not moderate postexercise changes in inhibitory control. This discrepancy might be due to a limited comparability of the exercise programs. We compared 10-min exercise sessions with low and high physical load at matched cognitive load, but previous studies with adolescents contrasted different intensities using either simple (e.g., aerobic exercise) or more complex exercises (e.g., dancing to a choreography) along with different durations (ranging from 5 to 30 min) of the individual bouts (12,13). This provides an indication that the effect of physical load on inhibitory control might vary, when intensity interacts with other dose parameters and exercise characteristics. Moreover, the absence of performance decrements on the Standard Stroop task at high physical load is in conflict with predictions made by the RAH model (14). Participants maintained their level of inhibitory control irrespective of the physical load, meaning that from behavioral performance and (de-)oxygenated hemoglobin changes, there was no evidence for a disengagement of PFC functions following exercise. Consequently, there was no need to compensate for additional cortical resources required to perform movements at high intensity or the PFC has been reengaged quickly after exercise cessation. The specific exercise sessions prescribed to the participants may offer an explanation for this deviation from the predictions of the RAH model (14). Training with the interactive wall was externally paced and did not include cyclic movements. This type of exercise might restrict the downregulation of higher-order cognitive functions, including inhibitory control, to ensure that the nonautomatic movements required to complete the task are maintained.

In contrast to physical load, the manipulation of cognitive demand affected postexercise behavioral performance on the Stroop task, but only if the conflict was due to emotional interference. This effect was independent of exercise intensity, recalling that there was no interaction of cognitive and physical load for behavioral performance and for changes in cortical hemodynamics during the cognitive task. The observation of decreased reaction time and increased accuracy following exercise at low compared with high might be explained by the association between emotion and cognition. Neuroimaging findings have shown that negative emotional priming impairs subsequent performance on the Stroop task, whereas high cognitive demand during the task leads to a blunted neural response to emotional stimuli in emotion-related networks, including the PFC, superior temporal lobe, and insula (34). Because of this interrelation, the cognitively challenging training session could not induce a temporary improvement of the ability to solve emotional conflict. However, exercise at low cognitive demand might have improved reaction time and accuracy on the subsequent Affective Stroop task because of a mild stimulation of networks subserving inhibition of emotional interference. A review of studies applying transcranial magnetic stimulation to the PFC and its dorsolateral region in particular found benefits for cognitive control processes involved in emotion regulation (35). Similar to traditional aerobic exercise (21), a short exergaming bout with low movement complexity has been found to increase the oxygenated hemoglobin concentration in the dorsolateral PFC in young adults (36). Using activation likelihood estimation, left dorsolateral and inferior PFC along with medial frontal gyrus, and left precuneus have been identified as network underlying emotional interference in the Affective Stroop task (16). Consequently, low-cognitively demanding exercise with the interactive training wall might have induced benefits in behavioral performance by a favorable alteration of activation patterns. However, the fNIRS findings indicate that LOD and LDD remained unchanged on incompatible trials, whereas on compatible trials, the oxygenated hemoglobin concentration decreased in the left relative to the right PFC following exercise at low cognitive load. This change could be interpreted as a shift toward a more efficient activation, given that the left-lateralized dominance is mainly required to resolve the conflict induced by incompatible trials (37). The pattern of changes was reversed for the standard blocks as indicated by a greater relative increase of left PFC oxygenation on compatible trials and right PFC oxygenation on incompatible trials. Based on the observation that the dominance of the left PFC is associated with a better resolution of the stimulus conflict (18), the participants’ oxygenation pattern could be deemed ineffective for the standard blocks of the Stroop task. This discrepancy of LOD findings between task variants might be explained by the cognitive tasks that were implemented into the exercise session to induce differing levels of cognitive load. The exergaming tasks did not involve any affective information, so that there was a preload on functions required for processing of nonemotional (and conflicting) information. The prolonged stimulation of these functions by exercise at high cognitive load might have led to a limited ability to upregulate the left PFC in response to incompatible trials. This may partly be due to an immaturity of the PFC, which has been shown by less left-lateralized changes in oxygenated hemoglobin in children compared with adults (38).

Our experimental findings provide new insights into the moderating role of cognitive demands and type of distractions on the acute effects of exercise on inhibitory control, but some limitations need to be considered. First, our study was designed to examine the influence of these moderators, but did not compare exercise-induced changes in behavioral performance with a nonexercise control condition. Although the repeated testing must have reduced a learning effect, it cannot be determined with our design. Second, we only contrasted two levels of physical and cognitive load, with a successful manipulation of the effort indicated by ratings on the NASA-TLX and perceived exertion. However, this approach only allows for an examination of linear associations, so that possible nonlinear trends are masked. In addition, both RPE and NASA-TLX are subjective ratings, meaning that it remains unclear whether objective measures of cognitive and physical effort would have indicated similar differences between the experimental conditions. Third, the tasks implemented in the exercise bouts were identical, meaning that external cognitive and physical load was identical across participants. However, cognitive load was rated higher in experimental conditions inducing high compared with low physical load. This may indicate that with increasing physical effort, participants experienced the cognitive component of the motor tasks as more demanding. Recalling that our results support improvements in inhibitory control of emotional information after exercise with low cognitive demands, the perception of a higher cognitive challenge (irrespective of the prescribed cognitive load) may cancel any benefits that are expected at exercise with high physical load. Consequently, an individualized approach for the determination of physical and cognitive load might be more appropriate to account for the immature cognitive and motor development in preadolescents. Fourth, our understanding of the mechanisms underlying the moderating role of cognitive load on exercise-induced benefits for inhibitory control is limited, because we focused on cortical hemodynamics in the PFC. Among other potential candidates, transient changes in autonomic function state and neurotransmitter levels contribute to Stroop performance (39,40) and may also be sensitive to acute exercise. Fifth, we calculated LOD for the PFC, with no further distinction made for the dorsolateral part, although this region is particularly activated during the Stroop task (16). This was due to a preliminary analysis showing a high cross-correlation between neighboring fNIRS channels and recommendation to examine cortical hemodynamic changes at the cluster level to increase test–retest reliability (41). In addition, the PFC was determined as the region of interest, meaning that it remains unclear whether changes in oxygenated hemoglobin in other functional units of the brain further contribute to improved inhibitory control following exercise at low cognitive load. Sixth, our fNIRS setup did not include short channels that allow the control of the signal for systemic physiological changes, such as scalp blood flow. However, these changes may have a low impact on LOD, because a similar blood flow in the left and right frontal region is neutralized during the calculation of the hemispheric difference. Lastly, our sample consisted of healthy preadolescent children, which potentially limits generalizability to other age groups due to the developmental differences in cortical hemodynamics (38).

CONCLUSIONS

In contrast to physical load, the cognitive demand of a single exercise session seems to influence inhibitory control in preadolescent children. However, this effect is limited to a specific type of interference. Although acute exercise with low compared with high cognitive demand benefits the ability to resolve emotional conflict, the control over inhibition of nonemotional information remains unaffected. The temporary improvement behavioral performance is complimented by a change in cortical hemodynamics promoting the efficiency of the left PFC, when no emotional conflict resolution is required. These benefits of acute exercise with low cognitive demands may be relevant in learning environments, where processing of information needs to be prioritized despite the presence of emotional distractions.