Edited by: Keita Kamijo, Chukyo University, Japan
Reviewed by: Hisashi Mori, University of Hyogo, Japan
Keishi Soga, Tohoku University, Japan
†These authors have contributed equally to this work
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
This study investigated which objective measures best predict cognitive performance in ultra-endurance athletes.
Forty-three recreational cyclists completed a race of ?341 km over 18 h, with cognitive function assessed using a three-stage Stroop Test before, immediately after, and 24 h post-race. For each stage, results from the three time points were combined and analyzed using regression analyses to examine their relationship with various anthropometric, metabolic, and performance-related parameters.
Higher average race speed was associated with improved simple reaction time, while longer race duration predicted better performance in more cognitively demanding tasks, such as inhibitory control. Muscle mass was the only anthropometric variable significantly linked to enhanced cognition. Other factors-including age, percentage of heart rate reserve, training volume, fat mass, lean mass, bone mass, and metabolic markers-showed no associations.
These findings suggest that race pacing and muscle mass play key roles in supporting cognitive resilience during prolonged physical exertion.
Ultra-endurance exercise presents a unique physical challenge that tests the limits of both physiological capacity and mental resilience. While its physical demands are well-documented, the cognitive implications of such extreme exertion remain less understood. Some studies have reported impairments in complex cognitive functions post-race, while performance on simpler tasks remained stable (
A self-planned ultra-endurance event offers a valuable framework to explore the bidirectional brain-body relationship. Cognitive capacity can affect pacing (
Beyond extreme endurance events, there is robust evidence that regular physical activity benefits cognitive functioning across all age groups, with even more advantages seen in older adults (
Current research, however, highlights that cognitive performance can vary pre- and post-ultra-endurance event. Some studies report a significant post-race decline in both simple and choice reaction times, possibly due to transient alterations in cardiovascular function following prolonged exertion in healthy adults (
Unsurprisingly, cognitive performance is linked to several factors reflecting participants' health and lifestyle prior to an ultra-endurance event. In this study, we focus on glycemic control, muscle mass, and age, as these variables have shown profound associations with cognitive functions in previous research (
To highlight their relevance, poor glycemic control—indicated by elevated glycated hemoglobin (HbA1c) levels—has been consistently associated with reduced executive functions and processing speed in both healthy and clinical populations (
In addition to the age-related decline in muscle mass and the corresponding rise in metabolic diseases, there is a well-documented decline in fluid intelligence beginning in the third decade of life (
The primary aim of this study is to examine how race performance metrics, prior training volume, age, body composition, and HbA1c levels are associated with cognitive function in the context of ultra-endurance exercise. We hypothesize that elevated HbA1c, older age, lower muscle mass, reduced resting metabolic rate (RMR), and lower training volume—either independently or in combination—will be associated with more pronounced cognitive impairment following ultra-endurance exercise.
The study was conducted during the ultra-distance cycling race, Veloreality. Held annually, the event took place from July 21–23, 2023, drawing recreational cyclists of all ages and fitness levels. Participants were required to reach all designated checkpoints by following a self-planned route, aiming to finish in the shortest time possible but no later than 42 h after the official start at 10 p.m. Consequently, the total distance covered varied, with an average of (341 ± 8 km). Forty-three of the recruited participants−39 men (91%) and 4 women (9%)–met the eligibility criteria for this study by completing the race before the 24 h mark. Participant characteristics are presented in
Descriptive statistics of the anthropo-physiological and race performance parameters.
| Age, years | 39 ± 8 |
| Finish time, min | 1,098 ± 169 |
| Average race speed, km/h | 21 ± 3 |
| Percentage of HRR, % | 56 ± 7 |
| Training volume, km | 3,278 ± 1884 |
| HbA1c, % | 5 ± 0 |
| Weight, kg | 84 ± 12 |
| Fat mass, kg | 13 ± 6 |
| Muscle mass, kg | 68 ± 8 |
| Lean mass, kg | 71 ± 9 |
| Bone mass, kg | 4 ± 0 |
| RMR, kcal | 2,066 ± 263 |
SD, standard deviation; HRR, heart rate reserve; HbA1c, glycated haemoglobin; RMR, resting metabolic rate.
Race regulations included the prohibition of crossing private property and cycling on highways outside populated areas, unless the highway had to be crossed. Participants were allowed to rest, refuel, and sleep during the race, if they completed the race within the designated time frame. Both the starting and finishing points were in Sigulda, Latvia, a city known for its varied terrain and elevation changes.
This study was conducted in accordance with the Declaration of Helsinki and was approved by the Riga Stradinš University Research Ethics Committee (reference number: 2-PEK-4/332/2023). All participants provided written informed consent after receiving information about the study's objectives and potential risks.
Anthropometric data—weight, fat mass, muscle mass, lean mass, and bone mass—were collected 1 to 2 days prior to the race. Body composition and RMR were assessed using a Tanita Multi-Frequency Body Composition Analyzer (MC-180MA) based on bioelectrical impedance analysis. Participants were instructed to follow the manufacturer's recommendations for pre-assessment preparation to ensure accurate and reliable measurements: (1) proper hydration should be maintained and alcohol avoided for 24 h prior to the measurement; (2) caffeine should be avoided for 12 h; (3) food intake should be refrained from for 3 to 4 h; (4) exercise should be avoided for 6–12 h; (5) showers and saunas should be avoided for 24 h; and (6) the application of creams or lotions to the palms and soles should be avoided on the day of the measurement. All participants self-reported adherence to these guidelines.
Individual finish time was obtained from the organizing team of Veloreality. Race performance metrics—including average race speed and average heart rate during the race—were collected from participants' personal wearable devices.
Additional data extracted from these devices included resting heart rate, maximum heart rate, and total cycling distance covered in the 6 months before the race (training volume), which served as an indicator of each participant's physical conditioning.
The average percent of heart rate reserve (%HRR) was calculated using the formula:
To evaluate participants' long-term blood glucose metabolism, a blood sample was collected before the race. The samples were sent to E. Gulbis Laboratory (Riga, Latvia) for analysis, where HbA1c levels were measured.
To assess participants' cognitive functions, the Stroop Test (ST) was performed at three different time points during the study: 1–2 days before the race (pre-race measurement), within 5–10 min after crossing the finish line (post-race measurement), and ~24 h after each participant's finish time (recovery measurement).
Each assessment was conducted under standardized conditions: participants were tested in groups of 4–6 people, seated in a quiet room, each at an individual computer provided by the research team. Researchers instructed participants on the procedure and provided one practice trial to familiarize them with the test format. The Stroop Test (ST) was administered using the SIA Exploro platform (exploro.lv).
The ST was used to evaluate participants' executive functions, such as inhibitory control and cognitive flexibility, as well as basic processes like attention and processing speed. The ST consisted of three stages: stage one—Stroop word-naming, stage two—Stroop color-naming, and stage three—Stroop color-word (
Stage I assessed the speed of visual search by testing how quickly participants can respond to a simple stimulus. In this stage, participants were instructed to press the spacebar as soon as any word appeared on the screen.
Stage II introduced additional cognitive challenges, including selective attention and working memory (
Stage III focused on assessing inhibitory control (
To account for random variability introduced by repeated measurements of the ST across three time points, Linear Mixed Models were used to examine the relationship between performance on each stage of the ST (dependent variables) and a range of predictors: age, finish time, average race speed, %HRR, training volume, fat mass, muscle mass, HbA1c, weight, lean mass, bone mass, and RMR.
To develop the regression models, both forward and backward stepwise selection methods were applied, evaluating all possible combinations and interactions. Model selection was guided by the Akaike Information Criterion. A Spearman's correlation matrix of all dependent and independent variables was also constructed to examine potential associations and inform the regression model-building process. To assess model fit, key assumptions—linearity, homoscedasticity, absence of multicollinearity, and normality of residuals—were tested. When the assumption of the residual normality was violated, robust p-values and confidence intervals were computed using a bootstrapping procedure with 1000 replications. All statistical analyses were conducted in Jamovi (version 2.5), with statistical significance set at
Results of the final regression model illustrating the relationship between ST performance and anthropo-physiological and race performance parameters.
| Stage I (Stroop word-naming) | Average time for correct reactions (ms) | Intercept | 294.6 | 281.0 to 308.2 | <0.001 |
| Average race speed | −2.5 | −4.5 to −0.4 | 0.021 | ||
| Stage II (Stroop color-naming) | Average time for correct reactions (ms) | Intercept | 658.4 | 623.4 – 693.4 | <0.001 |
| Muscle mass | −3.0 | −1.0 to −5.0 | 0.003 | ||
| Stage III (Stroop color-word) | Average time for correct reactions (ms) | Intercept | 757.9 | 714.7 to 801.1 | <0.001 |
| Finish time | −0.1 | −0.3 to −0.01 | 0.038 |
CI, confidence interval.
In Stage I, each 1 km/h increase in average race speed was associated with a 2.5 ms decrease in correct reaction time. This finding suggests that higher cycling speed is positively associated with cognitive functioning on Stage I of the ST. Given the participants' average race speed of (20.7 ± 3.1 km/h), this translates to an approximate 16-ms difference in reaction time across one standard deviation.
In Stage II, each additional kilogram of muscle mass predicted a 3.0 ms reduction in correct reaction time. This suggests that individuals with greater muscle mass performed better on Stage II of the ST. With an average muscle mass of (67.6 ± 8.3 kg), this corresponds to an estimated 50 ms difference in reaction time across one standard deviation, which represents a substantial effect.
In Stage III, each additional minute required to complete the race was associated with a 0.1 ms decrease in correct reaction time. Although the effect size per minute is small, it becomes functionally relevant when considering the natural variability among participants. For instance, a difference of ~169 min (the standard deviation of finish time) could correspond to a nearly 17 ms improvement in reaction time.
Other parameters examined in this study—including age, %HRR, training volume, HbA1c, weight, fat mass, lean mass, bone mass, and RMR—did not show statistically significant associations with endurance exercise mediated cognitive performance.
This study aimed to examine relationships between anthropometric, metabolic, and performance parameters and cognitive functions, as measured by the ST before, immediately after, and during recovery from an ultra-endurance cycling event. The key findings reveal that: (1) increased average race speed was associated with shorter reaction time in the Stroop word-naming task, reflecting improved simple reaction time; (2) greater muscle mass was associated with shorter correct reaction time in the Stroop color-naming task, which engages both working memory and simple reaction processes; (3) longer race completion time was unexpectedly positively associated with better cognitive performance in the Stroop color-word task, which involves conflict monitoring and executive function (
Our findings suggest that cognitive functions following an ultra-endurance event are more closely linked to actual race-day performance metrics—such as average race speed and finish time—than to relative physical load during the race (%HRR) or prior conditioning (training volume). Average speed and finish time reflect the athlete's real-time exertion and pacing strategy, while measures like %HRR and previous training volume are relative to the individual. %HRR, a widely used measure of cardiovascular exertion, can also be influenced by various factors—such as dehydration, ambient temperature, and body weight—making it a less precise indicator of the internal conditions affecting cognitive performance (
The direct influence of average race speed on simple reaction time may in fact be negative, as faster cycling presumably results in greater physical fatigue, which can slow reactions and reduce their consistency (
Another key factor influencing race speed is participants' pacing strategy, which is at least partially determined by cognitive skill level. Research in intellectually impaired individuals showed that optimal pacing relies on a complex interplay between executive functions and adaptive behaviours (
Lastly, average race speed serves as an indicator of participants' competitiveness. Indeed, performance on agility tasks improves when assessed under simulated competition conditions (
Previous research has predominantly found a positive association between inhibitory control and athletic performance. For example, studies using the Stop-Signal Task have shown that superior inhibitory control correlates with better sporting outcomes and tends to improve with increasing levels of athletic expertise (
In contrast, our findings reveal a different pattern. Comparing the cognitive demands of Stage I and Stage III of the ST suggests that they may tap into opposing abilities. Stage I measures simple processing speed, whereas Stage III evaluates the capacity to inhibit automatic responses. This divergence may imply that individuals with faster simple reaction times might struggle more with inhibitory control, and vice versa. In the context of race performance, this could mean that participants who cycled faster—potentially favoring a more intuitive, automatic style of decision-making—excelled in tasks requiring quick reactions but not necessarily in those demanding cognitive restraint. Conversely, participants who adopted a more deliberate and cautious pacing strategy may have demonstrated enhanced inhibitory control.
To our knowledge, no studies have directly compared simple reaction time and inhibitory control in an ultra-endurance setting. It is also important to note that inhibitory control encompasses more than just response suppression; it involves self-regulation and discipline, focused attention (e.g., ignoring interference), and cognitive inhibition (e.g., filtering intrusive thoughts) (
Recent studies suggest that muscle quality—not merely quantity—plays a key role in cognitive performance. By analyzing event-related potentials—small voltage fluctuations in response to specific stimuli recorded via electroencephalography—researchers have been able to more precisely and objectively assess neural activity underlying attention, cognitive inhibition, working memory, and related processes (
Given that muscle mass has been connected to interference suppression, (
An alternative perspective involves the endocrine role of skeletal muscle. During contraction, muscle tissue releases several myokines, including brain-derived neurotrophic factor (BDNF) and insulin-like growth factor 1 (IGF-1). BDNF has been strongly implicated in exercise-induced improvements in memory and learning (
A potentially more direct mechanism involves IGF-1. Exercise enhances IGF-1 expression, which supports neurogenesis and neuroplasticity while also exerting anti-inflammatory effects—mechanisms particularly relevant for protecting against cognitive dysfunction, especially in individuals with obesity (
As it relates to the current setting, the interplay between ultra-endurance cycling, muscle mass, and cognitive function has not yet been studied, though research on associations between any two of these factors does exist. For example, compared to the sedentary population, ultra-endurance athletes demonstrate increased total volumes of grey and white matter, while also showing reduced regional volumes in several brain areas involved in executive control, among other functions (
Among the additional variables considered in this study, HbA1c levels were also hypothesized to be associated with cognitive performance. Since impaired glucose metabolism also affects the brain, it can lead to reduced synthesis of neurotransmitters, impaired synaptic plasticity, and eventual neuronal damage, all of which contribute to poor cognitive performance (
As exercise is a vital tool in glycemic control, the majority of our study participants were normoglycemic and maintained well-regulated glucose levels. This likely eliminates the risk of cognitive decline due to glucose dysregulation, providing a plausible explanation for why HbA1c did not emerge as a significant variable in the analysis.
Age was another variable explored for its relationship to cognitive function, as it is a well-established predictor of cognitive performance at rest (
Given that age-related cognitive decline is typically gradual in early adulthood and accelerates later in life, the subtle changes expected in our relatively young and middle-aged sample may have been too small to detect. It can also be hypothesized that age has a more pronounced effect on chronic cognitive changes, whereas acute impairments—as measured in this study—may be less sensitive to age-related effects (
Coaches and athletes can leverage insights into pacing strategies to tailor training and race-day approaches according to the cognitive demands of their sport, whether emphasizing reaction time or inhibitory control. Since muscle mass emerged as a key anthropometric predictor of cognitive performance, incorporating resistance training may help support cognitive resilience during endurance events. Given the limited influence of physiological measures like %HRR and HbA1c, future research should explore the relationships between cognitive capacity and these markers in diverse athletic populations and conditions.
Limitations of this study include a relatively small sample size, which may have reduced the ability to detect subtle cognitive changes and limited the strength of observed associations. The use of dual-energy X-ray absorptiometry (DXA) could have improved the precision of anthropometric measurements.
Additionally, variability in HbA1c appeared to be very limited (SD not exceeding 0.5%), with mean value falling within the healthy range—likely due to the inclusion criterion requiring participants to complete the grueling race within 24 h, which is improbable for individuals in poor metabolic health.
The current study revealed that a higher average race speed during an ultra-endurance cycling event was associated with shorter simple reaction time, whereas a lower race speed was linked to better inhibitory control. Among the anthropometric measures, only muscle mass demonstrated a significant association with cognitive performance: participants with greater muscle mass exhibited superior control of working memory and selective attention.
Other variables—including age, exercise intensity (measured by %HRR), previous training volume, HbA1c, weight, fat mass, lean mass, bone mass, and RMR—did not show statistically significant associations with any of the assessed cognitive domains, as measured by the ST.
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
The study involving humans was approved by Riga Stradinš University Research Ethics Committee (reference number: 2-PEK4/332/2023). The study was conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.
DV: Investigation, Writing – original draft. DL: Investigation, Writing – original draft. JR: Investigation, Writing – review & editing, Writing – original draft. LL: Writing – review & editing, Investigation. GG: Writing – review & editing, Investigation, Methodology. IB: Investigation, Writing – review & editing. IJ: Investigation, Writing – review & editing. KK: Investigation, Writing – review & editing. BR: Methodology, Writing – review & editing, Conceptualization. MZ: Data curation, Formal analysis, Writing – review & editing. DR: Funding acquisition, Writing – review & editing, Methodology, Conceptualization, Supervision, Project administration.
The author(s) declare that financial support was received for the research and/or publication of this article. This study was funded by an internal grant from Riga Stradinš University (Reg. No. 6-ZD-24/2/2023) titled “Effects of acute, long-term endurance exercise on blood plasma sestrin concentrations.” The funding source had no involvement in the study design, data collection, analysis, interpretation, or manuscript preparation.
Bart Roelands is a Collen-Francqui research professor and a member of the Strategic Research Program
Edmunds Vanags, PhD, is a leading expert in Psychology at the University of Latvia. He contributed to this study with his extensive knowledge in cognitive psychology and neurophysiology.
The authors declare that the results of the current study do not constitute an endorsement of any product by the authors or the journal.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The author(s) declare that no Gen AI was used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.