Executive function during exercise is diminished by prolonged cognitive effort in men

The speed and accuracy of decision-making (i.e., executive function (EF) domains) is an integral factor in many sports. At rest, prolonged cognitive load (pCL) impairs reaction time (RT). In contrast, exercise improves RT and EF. We hypothesized that RT and EF during exercise would be diminished by prolonged ‘dual tasking’ as a consequence of pCL. To test the hypothesis, twenty healthy male participants performed four conditions [resting control (Rest), pCL only (pCLRest), exercise only (EX), and pCL + exercise (pCLEX)] in a randomized-crossover design. Both exercise conditions utilized a 50-min cycling exercise protocol (60% VO2 peak) and the pCL was achieved via a 50-min colour-word Stroop task (CWST). Compared with Rest, pCLRest caused a slowed CWST RT (P < 0.05) and a large SD (i.e., intraindividual variability) of CWST RT (P < 0.01). Similarly, compared with EX, the slowed CWST RT (P < 0.05) and large SD of CWST RT (P < 0.01) were also observed in pCLEX. Whereas the reverse-Stroop interference was not affected in pCLRest (P = 0.46), it was larger (i.e., declined EF) in pCLEX than EX condition (P < 0.05). These observations provide evidence that the effort of pCL impairs RT and EF even during exercise.

Executive function. The number of errors for all CWST were not different between conditions (Table 1).

Discussion
The present study examined the impact of continuous pCL on RT and EF during acute moderate-intensity aerobic exercise. In line with our hypothesis, a slower RT and diminished EF was apparent during exercise with pCL, along with the increased RPE and mental fatigue. Namely, pCL enhanced cognitive fatigue even during aerobic exercise.
To avoid the speed-accuracy trade-off, all participants paid attention to prevent mistakes in CWST throughout all conditions (see Methods). Accordingly, the error of CWST was almost 0 in all conditions (see Table 1). Thus, we focused on the interpretation of changes in RT of CWST in the present study. Moreover, a previous study demonstrated that prolonged CWST (> 30 min) causes a slower averaged RT and large intraindividual RT variability 4 . At rest, we also confirmed that the 50-min CWST caused the same response. In addition, mood was impaired by the 50-min CWST at rest. Thus, our 50-min CWST is capable of inducing cognitive fatigue.
Acute moderate-intensity aerobic exercise increases state mental fatigue 13,14 . In addition, we have revealed for the first time that exercise-induced mental fatigue and perceived effort is further enhanced by pCL. These observations indicate that the subjective marker of internal load (i.e., RPE) was increased by the continuous cognitive effort, despite similar absolute (i.e., W) and physiological (i.e., HR) intensities. The increased RPE may be associated with the hyperventilation-induced suppression of cerebral blood flow (CBF) in response to aerobic exercise 21 or activation within the posterior cingulate cortex and precuneus, as demonstrated using fMRI 22 . Given that the dorsolateral prefrontal cortex is activated during EF performance 23 despite de-activation during aerobic exercise 24 , it is possible that the brain during a 50-min dual-task with motor and cognitive demands may need higher energy requirements (i.e., activate both anterior and posterior areas) whereas global CBF as energy supply may be decreased by moderate-intensity aerobic exercise-induced hyperventilation 12 . Hence, compared with exercise only, mental fatigue and RPE are further increased by the combined effort of exercise and pCL possibly due to cerebral energy deficiency.
In relation to mental fatigue and RPE, a slower averaged RT and large intraindividual RT variability were observed during aerobic exercise with pCL in the same way as the rest conditions. These observations indicated that the maintained cognitive effort causes a greater delayed RT and the unstable RT during the last 5-min of 50-min exercise. Incidentally, pCL did not alter the SD of RT for the congruent task during exercise, implying that the pCL-induced intraindividual simple RT variability (i.e., stability of performance for easy task) may be masked by the effect of exercise. Importantly, EF (i.e., interference) during exercise was diminished by the 50-min pCL (i.e., dual-task), despite no effect of the single-task of pCL (i.e., rest). It is likely that the delayed decisionmaking speed during the prolonged dual-task may be partly evoked by the impaired cognitive process of EF. Given that the speed of decision-making and EF may contribute to superior sports performance 1,25 , cognitive central fatigue may adversely impact sports performance. www.nature.com/scientificreports/ The physiological mechanism(s) of this finding remains unknown. However, it may be linked to psychological arousal. Psychological arousal is associated with the exercise-induced improvements in EF 26 . Correspondingly, a lower level of psychological arousal may be associated with the pCL-induced impairment in EF during exercise (see Fig. 6). Therefore, the results of the present study may be explained by biomarkers of physiological arousal. Alternatively, we previously reported that dynamic cerebral autoregulation (dCA) may be maintained/ improved by exercise/diet stress 27,28 which is capable of facilitating speed of decision-making 29,30 . The dCA acts to maintain relatively constant CBF against rapid fluctuations in perfusion pressure for brain homeostasis 31 , but Ogoh et al. 32 demonstrated that a 5 min cognitive performance-induced cerebrovasodilation (i.e., brain activation) impaired dCA. Future studies should examine whether pCL impairs the dynamic CBF regulation at rest and during exercise, thereby contributing to cognitive decline. www.nature.com/scientificreports/ Implications for brain health. EF has an important role in brain health and deteriorations have been observed in neurological conditions such as dementia 33 . It is widely known that regular and chronic exercise training improves EF 34 . For instance, evidence exists suggesting that a 3-12 month dance program as a simultaneous motor-cognitive exercise with an 'incorporated' cognitive task 35 enhances EF 36 . Interestingly, a 3-month intervention of 40-min dual-task "Cognicise (Cognition + Exercise of low-intensity)" may improve cognitive function for the elderly who have a cognitive impairment although to what extent aerobic exercise alone could have improved cognition in this previous study was not investigated 37 .
The exercise training-induced improvements in cognition may be predicted by the acute effects of exercise 38 . Of note, the effect size of a single bout of moderate-intensity exercise on EF improvement is greater than lowintensity exercise 13 . Nevertheless, the present study implies that the inhibitory control in EF may be diminished when cognitive performance is maintained during moderate-intensity aerobic exercise prescription, indicating the strong negative effect of cognitive fatigue due to simultaneous motor-cognitive exercise with an 'additional' cognitive task 35 . Moreover, Kamijo and Abe 39 demonstrated that working memory in EF is improved after 20-min moderate-intensity aerobic exercise but this aftereffect is cancelled by a 20-min additional cognitive performance during moderate-intensity aerobic exercise. In addition to EF, pCL increased RPE in the present study. The higher level of RPE may be associated with a lower level of psychological self-efficacy 40 , which predicts lower exercise adherence 41 . In light of these findings, there is, on the one hand, evidence that chronic physical exercise with an additional or incorporated cognitive task (e.g., dancing) can improve cognitive performance, whereas, on the other hand, our findings imply that acute physical exercises with an additional cognitive task can cause cognitive fatigue which, in turn, negatively influence cognitive performance with respect to shorter time frames. To optimize the prescription of physical exercise for cognitive health, future research on physical exercise with additional or incorporated cognitive tasks is needed.
Perspective. In football, an increase in goal-scoring rate is observed in the latter stages of the match along with fatigue 42 , which may partly be due to the impairment in cognitive perception 3 . However, Wiśnik et al. 43 reported that the RT of cognitive task is shortened during treadmill exercise of changing intensity simulating a sport game. To the best of our knowledge, previous research could not detect why the cognitive-perceptive skills may often be poorer in the latter stage of the match. For example, Wang et al. 44 demonstrated that EF is diminished during exhaustive high-intensity continuous exercise (e.g., 80%VO 2 peak for 40-min), but the continuous exercise may not reflect the sporting performance and training situation, except for marathon etc. The present findings provide an alternative explanation for physical fatigue as to why sports performance may be impaired towards the end of the game 3 . Although the cognitive-perceptive skills in sports may be altered by confounding factors such as contacts [45][46][47] , we proposed a new experimental design, which can examine the impact on cognitive fatigue during prolonged aerobic exercise. Future studies are necessary to reveal training/nutritional  www.nature.com/scientificreports/ strategies that may improve EF and psychological variables during this dual-task, which in turn could mitigate the negative effects and subsequently improve sports performance.

Conclusion.
A 50-min cognitive task impaired RT not only at rest, but also during moderate-intensity aerobic exercise. In addition, EF during aerobic exercise was diminished when the cognitive task was continuously performed. These findings provide empirical evidence that cognitive fatigue appears even whilst aerobic exercising and impairs cognitive-perceptive skills during aerobic exercise.  www.nature.com/scientificreports/ Methods Ethics and participants. All procedures were approved by the Ethics Committee of Ritsumeikan University (BKC-2020-085) and conformed to the Declaration of Helsinki, except for registration in a database. A priori sample size calculation (G*Power) suggested that assuming the variation in reaction time following the completion of a prolonged modified-CWST 6 , a sample size of 21 would provide a statistical power of 80% at an α level of 0.05. Accordingly, twenty-one healthy males participated in the study after providing written informed consent. All participants were free of neurologic, cardiovascular or pulmonary disorders, were not taking any medication, and were non-smokers. Participants were instructed to avoid alcohol, caffeine intake and strenuous physical activity in the 24-h preceding each experimental visit. Each participant was also asked to abstain from food for 12-h before each experiment. Experiments were performed at 22-24 ℃.  www.nature.com/scientificreports/ Study design (Fig. 7). Participants initially completed a preliminary session where they were familirised with the cognitive assessment (i.e., CWST) and their cardiorespiratory fitness was determined. Thereafter, 4 separate conditions were completed on separate days in a randomized, counterbalanced order: Rest, pCL Rest , EX, and pCL EX . Participants attended the experimental days between 08:00-10:00 am and each visit was separated by a minimum of 72-h.
Experimental conditions (Fig. 7). During all conditions, participants performed CWST after resting for 10-min to determine the speed and accuracy of the decision-making and EF at baseline. HR and psychological state were also measured prior to each intervention. After confirming a slight day-to-day difference at baseline (Supplementary Table 1), we performed the experiment using within-subjects crossover posttest comparison, the most robust study design to investigate the acute effects of physical exercise on cognitive performance 15 . In the Rest condition, the participants performed 5-min CWST after sitting at rest on the bike for 45-min. In the EX condition, the participants performed moderate-intensity cycling exercise for 50-min, and CWST was carried out during the final 5-min. In the pCL Rest condition, the participants continuously performed CWST for 50-min in a sitting position on the bike. In the pCL EX condition, the participants continuously performed both CWST and moderate-intensity cycling exercise for 50-min. In both pCL Rest and pCL EX conditions, the speed and accuracy of the decision-making, including EF, was determined from the score of CWST at the final 5-min of such interventions.
Experimental procedures and measurements. Cardiorespiratory fitness. VO 2 peak was determined on a bike (95C Inspire Upright Lifecycle, Life Fitness Japan, Tokyo, Japan). The exercise test began at a power of 60 W for 1 min. Thereafter, the workload was increased by 15 W/min until the participants could not maintain a cadence of 60 rpm (task failure of a pedaling rate of at least 55 rpm over 5 s despite maximal effort) 27,48 . Throughout the test, expired gas fractions were measured continuously using online breath-by-breath gas analysis (AE-310S; Minato Medical Science, Osaka, Japan). VO 2 peak was determined as the highest 30-s value attained prior to exhaustion 13,14,30,49 and used to calculate moderate-intensity exercise for the relevant experimental conditions (60% VO 2 peak; 151 ± 18 W).
Heart rate. HR was monitored at rest and during exercise using telemetry (RS400 and H10; Polar Electro, Kempele, Finland).
Psychological parameters. The visual analog scale (VAS) included questions of 4 psychologic types that assessed mental fatigue, ability to concentrate for CWST, motivation for CWST, and comfort. Each VAS was labeled from 0 (not at all) to 100 mm (extremely). The participants drew lines to indicate their responses 13,14,30,48 . Moreover, the Felt Arousal Scale (FAS) assessed arousal level, ranging from 1 (low arousal) to 6 (high arousal) and the participants were asked to report how they felt after performing the CWST. A high arousal level represents excitement, and a low arousal level represents relaxation 50 . Immediately after exercise, participants were asked to provide RPE using the Borg 6-20 scale, ranging from 6 (no exertion) to 20 (maximal exertion) 51 to estimate the effort expended during exercise.
Reaction time and accuracy of executive function task. Evaluation of the speed and accuracy of the decisionmaking and EF was completed using CWST 52 , a paradigm for investigating aspects of cognitive performance. This specifically provides selective attention to specific information and inhibits prepotent responses during decision-making tasks involving stimuli and responses 53 . The CWST was adopted in an event-related design and both 5-min and 50-min CWST were programmed in SuperLab (Cedrus, San Pedro, CA, United States). The words stimuli were 4-colour names; red, blue, yellow, and green in Japanese characters. The participants performed 3 types of CWST: 1, the congruent task as a facilitated easy task displaying the colour names presented in the same-coloured text; 2, the neutral task as the control displaying the color names presented in black text; and 3, the incongruent task as an interference task displaying the color names in a different coloured text. The words for each type of task were presented for 1500 ms following a 500 ms fixation cross, in a counter-balanced random order and random interval (1, 3, and 5 s) to avoid prediction of the timing of the subsequent task 4 ; in other words, 20 congruent, 20 neutral and 20 incongruent stimuli were presented per 5-min CWST. A colour-labeled keyboard (RB-540, Cedrus, San Pedro, CA, United States) was prepared and the participants were asked to press the colour-labeled key that corresponded to the text meaning of the stimulus word. The following instruction was given to the subjects; "Please do not make a mistake" and "You must perform the task as accurately and quickly as possible". The averaged and SD of RT, and the number of errors were assessed using 5-min data. To evaluate EF, reverse-Stroop interference scores were calculated as the averaged RT as [(RT of incongruent task− RT of neutral task) / RT of neutral task)] × 100 because a higher and stable reverse-Stroop interference scores can be observed when CWST is performed using manual response 54 . In a preliminary session, CWST was performed during cycling exercise for approximately 10-min (i.e., a dual-task) until the participants obtained consistent scores to avoid learning effect.
Statistical analysis. If  www.nature.com/scientificreports/ was confirmed, whereas the Friedman test was used to analyze the baseline data if normal data distribution was not confirmed (IBM SPSS Statistics version 27, Chicago, IL, United States). In addition, whereas we did not directly compare the data between the rest and exercise conditions because of the influence of different test conditions due to the dual-task (e.g., the percentage of effort to CWST test, gaze destabilization, and sweating, etc.), a paired t-test (for normal data distribution) and the Wilcoxon signed-rank test (for non-normal data distribution) were used to compare the data between Rest and pCL Rest , and between EX and pCL EX , respectively (IBM SPSS Statistics version 27, Chicago, IL, United States). The statistical significance level was defined at P < 0.05. Moreover, Cohen's d effect size using the means and pooled SD were calculated, along with the 95% confidence interval to determine the magnitude of differences for normal data distribution. For non-normal data distribution, the effect size was estimated as r using Z-score for the Wilcoxon signed-rank test. The strength of effect size of Cohen's d was interpreted as weak (0.20 ≤ d < 0.50), medium (0.50 ≤ d < 0.80), and large (0.80 ≤ d), www.nature.com/scientificreports/ while the strength of effect size of r was interpreted as weak (0.10 ≤ r < 0.30), medium (0.30 ≤ r < 0.50), and large (0.50 ≤ r) 56 . Finally, we performed repeated measures correlation (rmcorr) between the reverse-Stroop interference and arousal 57 .

Data availability
The data that support the findings of this study are available from the corresponding author, H.Tsukamoto, upon reasonable request. www.nature.com/scientificreports/