Abstract
Effective stress management is crucial for optimal competition performance in athletes. Sleep deprivation (SD) can elevate physiological and psychological stress, and the SD-changed cognitive and emotion may reflect stress management capability and hold the predictive possibility for athletes’ performance in official competitions over some time; however, it lacks evidence. Here, we aim to increase stress level for athletes by 24 h SD and identify the predictive effects of cognitive and emotional changes after 24 h SD on sports performance in official competitions over ∼1.5 months. Sixty-five winter sports athletes (35 males) were recruited from college (test set) and professional athletes (validation set) separately. The anxiety and cortisol levels were assessed at baseline, after 24 h SD, and official competition. Athletes underwent cognitive tasks (Stroop, Go/No-Go, Competitive Reaction Time Task, and Iowa Gambling Task) and the event-related potential recording at baseline and after SD. Competition performance levels (supernormal, normal, and abnormal) were categorized based on a consensus of subjective and objective evaluations. We found anxiety and cortisol levels following 24 h SD were equaled with those observed in official competition. Notably, only the decreased incongruent Stroop response after 24 h SD was negatively associated with performance in official competition. The corresponding P3 component, particularly the delta frequency at the central lobe, largely mediated this effect. These findings highlight that athletes who effectively employ cognitive skills to manage stress under acute SD tend to exhibit superior performance.
- cognitive control
- competition performance
- event-related potential (ERP)
- predictive biomarker
- sleep deprivation
- winter sports athletes
Significance Statement
Predictive methods and biomarkers for athletic performance are currently lacking. Our study first confirmed that the changes in attention control after 24 h sleep deprivation (SD) held unique predictive effects for athletes’ competition performance over ∼1.5 months, and the delta frequency of the P3 EEG component at the central lobe may contribute to it. This study emphasized that athletes can harness additional cognitive resources to enhance stress management, which could lower the risk of abnormal performance in official competitions. Cognitive predictors after SD can assist athletes and coaches in monitoring the training state, enhancing stress management to optimize athletic performance, and adjusting athlete participation arrangements.
Introduction
Competition induces stress on athletes, and the effects that such stress may have on sports performance (Hardy, 1992). Stress management is crucial for achieving good competition performance (de Witte et al., 2016; Smyth et al., 2022). With the increasing emphasis on enhancing performance in sports, there has been a notable focus in sports psychology on predicting how athletes will perform in competition (Borresen and Lambert, 2009; Bonney et al., 2019). Therefore, it is essential to develop methods and biomarkers to forecast the likelihood of athletes performing poorly in competition ahead of time. This can assist coaches in monitoring athlete preparedness, enhancing training, and adjusting athlete participation arrangements.
Sports performance in official competitions requires a higher top-down executive control than daily training (Schweizer et al., 2017). There is a significant amount of evidence for the effects of stress-induced cognitive changes on sports performance (Beilock and Carr, 2001; Woodman and Hardy, 2003), and high performance is safeguarded against acute stress through the recruitment of compensatory control in cognition (Kalén et al., 2021). Therefore, the cognitive alterations that occur under acute stress may indicate the ability to manage stress. The stress vulnerability in the key cognitive components, including attention and inhibition control, decision-making, and reaction capacity (Baur et al., 2006; Furley and Wood, 2016; Mirela, 2020; Vaughan and Laborde, 2021), have the potential to predict athletes’ competition performance in a period of time. Besides, anxiety and stress of athletes have been recognized as key predictors of performance and injury in competition (Xanthopoulos et al., 2020). Successful football players tend to exhibit lower anxiety levels and higher self-confidence (La Fratta et al., 2021). However, empirical evidence for the effects of stress-induced cognitive and anxiety changes on sports performance is currently limited.
Factors contributing to sports competition stress include high-performance expectations, challenging opponents, and external pressures (such as coach attention and live audiences), leading to stress-induced sleep disturbances and cognitive and emotional changes (Malcata and Hopkins, 2014). Research showed that over 49% of Olympic athletes experience multiple sleep problems before competition, such as early awakening and difficulty falling asleep (Nédélec et al., 2015). Decreased sleep duration can heighten athletes’ levels of competition anxiety and fatigue, worsening overall stress during competition (Fullagar et al., 2015; Bonnar et al., 2018). Sleep deprivation (SD) is a common psychological method to increase physiological and psychological stressors (Matre et al., 2015). We suggest that the vulnerability to sleep loss is linked to a smaller cognitive buffer capacity, predictive of poor sports performance over a period of time.
Neuroelectrophysiological brain activity, specifically the event-related electroencephalogram component, plays a critical role in understanding the psychobiological mechanisms that underlie the relationship between predictors and competition performance. Limited studies on cognitive testing in athletes have shown that professional athletes in fast ball sports have faster latencies of the P3 component during visual tasks compared with healthy controls (Isoglu-Alkac et al., 2018). Elite athletes, when engaged in a response inhibition task, exhibited higher P3 amplitudes in frontal and parietooccipital locations than amateur athletes (Park et al., 2015). Furthermore, skilled martial arts athletes showed significant amplitude differences between target and nontarget conditions in the P2 component compared with novices (Sanchez-Lopez et al., 2016). Moreover, table tennis athletes exhibited a decrease in P3 amplitude in the stop-signal task after 36 h sleep deprivation (Xu et al., 2022). While investigation of cognitive event-related potential components under stress shows promises in predicting neural correlates of performance in competitive scenarios, support evidence is still lacking.
To identify predictors of athletic performance level in advance and underlying neuroelectrophysiological mechanism, we first evaluated the predictive effects of anxiety, cortisol levels, and cognitive changes (attentional and inhibition control, decision-making, and reaction capacity) after 24 h SD on the sports performance at official competitions (∼1.5 months later) in both college athletes set and professional athletes set. Subsequently, we investigated the mediating neuroelectrophysiological mechanisms involving event-related potentials (ERPs) and time–frequency activities in relation to the predictive effects.
Materials and Methods
Study participants
In our study, there were 65 athletes participating in the winter games. The college athletes set comprised 35 college athletes in Hebei Sport University from June 2021 to December 2021, engaging in snowboarding, alpine skiing, cross-country skiing, curling, and speed skating. Additionally, the professional athletes set consisted of 30 professional athletes from the Hebei Province Sports Delegation, CHN, from July 2022 to October 2022, participating in cross-country skiing, ice hockey, and biathlon.
Experimental procedure
Participants were instructed to maintain a regular sleep schedule and wore actigraphy devices (Actiwatch Spectrum Pro, Philips) for a week before the experiment. Only participants who had slept for at least 6 h per night over a week before the experiment, went to bed no later than 12:00 midnight, and woke up no later than 10 A.M. were included in the study. The baseline information and questionnaires were evaluated the day before the experiment day. Subjects then underwent a 24 h sleep deprivation (SD) period, commencing at 8 A.M. on the first day and concluding at 8 A.M. on the following day. The baseline state (BL) was established at 8 A.M. on the onset day of sleep deprivation. Participants were subsequently followed up for their most recent official competition (ROC), with the time interval between BL and ROC averaging approximately 45.03 (±16.68) d. Athlete performance in the ROC was monitored, specifically the 3rd Hebei Ice and Snow Sports Games for college athletes in November 2020–December 2020 and the 16th Hebei Youth Cross-Country Skiing Competition during the 2021–2022 season for professional athletes in December 2021. Postmatch retrospective evaluations were conducted after the competition. State anxiety levels and saliva samples were collected at 8 A.M. on the first experiment day (BL), the day following a 24 h sleep deprivation (SD), and on the first day of the recent official competition (ROC). To ensure that saliva collection takes place at least 30 min after waking, the researchers would ensure that the subject has woken up by 7:30 A.M. on the morning of saliva collection. Event-related brain electrical signals during the cognition paradigms testing were collected at 8:00 on BL and SD (Fig. 1A).
Anxiety and stress state after 24 h sleep deprivation were equaled with those in competition. A, The experimental procedure. B, The scores of state anxiety inventory (SAI) at BL, SD, and ROC and the correlation of SAI between SD and ROC in the college athletes set (n = 35). C, The salivary cortisol levels at BL, SD, and ROC and the correlation of salivary cortisol between SD and ROC in the college athletes set (n = 35). D, The scores of state anxiety inventory (SAI) at BL, SD, and ROC and the correlation of SAI between SD and ROC in the professional athletes set (n = 30). E, The salivary cortisol levels at BL, SD, and ROC and the correlation of salivary cortisol between SD and ROC in the professional athletes set (n = 30). BL, in baseline; SD, after sleep deprivation; ROC, recent official competition; SAI, the scores of state anxiety inventory. *p < 0.05, **p < 0.01, ***p < 0.001.
Assessments
Demographic and sport characteristics
We collected demographic information, including age, sex, BMI, and education level. Sports information encompassed sports type, the length of time in the current type of sport, and sports grade, which was determined based on the Athletes Technical Level Certificate issued by the General Administration of Sport of China.
Questionnaires
We assessed state anxiety using the 20-item state anxiety inventory (Spielberger and Vagg, 1984) at BL, SD, and ROC. To gauge depression and anxiety levels, we utilized the Patient Health Questionnaire-9 (PHQ-9; Kroenke et al., 2001) and the Generalized Anxiety Disorder Scale (GAD-7; Spitzer et al., 2006). The severity of both nighttime and daytime components of insomnia was assessed using the insomnia severity index (ISI; Morin et al., 2011). These questionnaires were administered at BL assessment.
Postcompetition assessments for sleep problems and performance level
Following the ROC, we assessed the athletes’ performance levels and inquired about their prematch sleep quality (pre-ROC sleep problems). We asked the athletes if they experienced any sleep disturbances and requested them to specify the main types of sleep problems they encountered.
Competition performance evaluations were gathered from both athletes’ self-assessments and coaches’ evaluations to compare current performance with expectations. If there were any disparities between them, consultations with both parties were conducted to achieve consensus on the final result. Performance levels were categorized into three groups: high performance (HIGH), i.e., higher performance than expectation; normal performance (NOR), i.e., consistent performance with expectation; and low performance (LOW), i.e., lower performance than expectation.
Acute 24 h sleep deprivation (SD)
During the 24 h acute sleep deprivation phase, participants were required to stay awake from 8 A.M. on the first day to 8 A.M. on the following day. From 8 A.M. to 6 P.M. on the first day, participants were allowed to engage in their usual daily activities. Researchers maintained contact with the subjects every half an hour to ensure they remained awake and to provide support as needed. From 6 P.M. onwards to 8 A.M. the next day, subjects were required to stay within the laboratory under the continuous supervision of the research team.
To minimize variability and ensure consistency across the study, researchers provided a standardized set of activities for all subjects to follow, including nonstressful activities such as reading, watching videos, and chatting with other subjects. This protocol was designed to reduce the potential impact of different activities on cognitive measures and to control for additional cognitive load that might be introduced by more stressful or variable tasks. The researchers closely monitored the subjects’ status to ensure their well-being and to maintain the integrity of the sleep deprivation protocol, confirming that each subject adhered to the prescribed activities throughout the deprivation period.
Cognition paradigm testing
Stroop test was employed to assess attention control functions, cognitive flexibility, and processing speed (Kane and Engle, 2003). The Stroop color–word test included 100 trials (50% congruency, three colors: red, blue, and green). Each trial consisted of a white fixation point (500 ms), followed by a color word (500 ms). Stimuli were presented centrally and sequentially on a computer monitor, and all stimuli remained on the screen until a response was made. Reaction times were measured separately for congruent and incongruent trials. The primary focus was on collecting the response time of correct trials under color–word congruent stimuli (Con RT) and incongruent stimuli (Icon RT), as well as the correct response rate (accuracy).
The Iowa Gambling Task (IGT) was employed to assess decision-making (Toplak et al., 2010; Wemm and Wulfert, 2017) due to its real-world relevance (Bechara et al., 1997). The task involved participants who were presented with four virtual decks of cards (A, B, C, or D) on a computer screen. The participants started with ¥200, and over the course of 60 trials, participants had 3,000 ms to select one card from any of the four decks and received feedback immediately (1,000 ms). There were two “bad” decks (A or B) that offered larger immediate wins (+¥25) but also larger penalties (−¥50). Additionally, there were two “good” decks (C or D) that offered smaller wins (+¥5) and smaller penalties (−¥5). Each card has an equal chance of determining the outcome of the game, with a 50% probability of winning and a 50% probability of losing. College athletes made a total of 30 choices, while professional athletes made a total of 60 choices. The key parameters collected included the selection probability of “bad” decks (A or B solitaire) denoted as AB and the final winning amount (amount).
Competitive Reaction Time Task (CRTT) was utilized to assess reaction capacity and aggression (Lobbestael et al., 2021). In this task, participants were informed that they were competing against an opponent and had to outperform them by clicking a mouse when a “*” turned from black to red. The time it took for the “*” to change from black to red was randomized, ranging from 500 to 1,500 ms. The number of win-or-lose trials and the volume and duration levels of tones administered by the opponent were preprogrammed in the same order for every participant. A valid reaction time range was set between 0 and 2,000 ms, and if participants reacted slower, they automatically lost the trial to ensure credibility. Participants were also informed that the winner of a trial could administer a loud noise to their opponent, which could potentially affect the opponent's performance in the next trial. Before each trial, participants were asked to choose the volume of the noise blast when a “?” turned from black to red which ranged from 1,000 ms, with options of 0, 90, 100, and 110 db. Participants completed a total of 50 trials. The primary parameters collected included the noise selection level (noise intensity) and reaction time (RT).
Go/No-Go Task (GNG) was administered to measure response inhibition (Schweizer et al., 2017). Participants were instructed to provide varying responses based on different picture stimuli. They were required to press a button when the “go” signal appeared and refrain from responding when the “no go” signal appeared. A total of 100 trials were administered, with each signal randomly appearing 50 times. Outcome measures included the correct response rate (accuracy) and reaction time (RT).
Cortisol analysis
Salivary cortisol is a well-established biomarker of stress (Noushad et al., 2021), and anxiety is closely linked to stress levels (Konstantopoulou et al., 2020). Both measurements can help assess the psychological and physiological state of stress. Saliva samples were collected through passive drool using Salivettes (Sarstedt) and subsequently stored at −80°C. These samples were utilized for the quantitative determination of human cortisol concentrations in saliva. Cortisol analysis was conducted using commercially available enzyme-linked immunosorbent assay kits (Salimetrics) according to the manufacturer's guidelines. The coefficient of variation for this ELISA kit was <7%, with an analytical sensitivity of <0.007 μg/dl.
EEG recording and analysis
EEG data were obtained using a 24-channel mBrainTrain Smarting EEG for whole-brain EEG acquisition. The 24-cade EEG cap adhered to the 10–20 system. EEG acquisition and recording commenced once the impedance of all electrode channels dropped below 5 kΩ. Prior to EEG data acquisition, subjects were told to keep their head and body still during the acquisition process and to control the blinking frequency under the instructions of the corresponding pictures and so on. The EEG signals were digitized online at a sampling rate of 500 Hz.
EEG data analysis was conducted using the EEGLAB and FieldTrip Toolbox in MATLAB R2021b. To ensure the quality of the data, any subject whose discarded segments exceeded 20% of the total was excluded from further analysis, and subjects whose PCA components removed exceeded 20% of the total were also not included in subsequent analyses. Event-related potential (ERP) and time–frequency analysis was performed on the preprocessed EEG data. The time range of the P3 component appearance was set at 300–450 ms, while the P2 was 150–250 ms. The regions of interest included frontal and central areas (F3, F4, Fz, C3, C4, Cz). Energy calculations were then performed for four classic frequency bands: delta (1–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), and beta (13–30 Hz).
Statistical analyses
To compare demographics between the college athletes set and professional athletes set, t tests were employed, while chi-square tests were used for between-group comparisons of discontinuous variables. One-way analysis of variance (ANOVA) was used to compare state anxiety and salivary cortisol levels between BL, SD, and ROC. Spearman's correlation was used to find the correlation between anxiety and salivary cortisol levels in SD and ROC, with age, sex, sports, and sports grade included as covariates.
t test was used to compare cognitive indicators between BL and SD. One-way ANOVA was used to analyze the differences in the 27 cognitive indicators from GNG, Stroop, CRTT, and IGT among the different performance groups (HIGH, NOR, and LOW) in college athletes set, where the original and FDR-corrected p-values were based on the false discovery rate. The ordinal logistic regression was then used to further assess whether the association between cognitive indicators remained significant after controlling for age, sex, sports type, sports grade, and pre-ROC sleep problem in performance athletes set. One-way ANOVA was used to analyze ΔIcon RT among the different performance groups (HIGH, NOR, and LOW), and further within-group post hoc analyses were conducted for the ΔIcon RT.
Spearman's correlation was used to identify which ERP components were correlated with behavioral indicators, and ANOVA was used to assess group differences in ERP components and time–frequency energy. All analyses were performed using SPSS27.0.
IRB approval
This study was approved by the Peking University Institutional Review Board and was performed following the relevant guidelines and regulations, and the reference number for approval was IRB00001052-21089. Each individual signed a written informed consent form.
Results
Basic characteristics and sports performance of athletes
A total of 65 subjects participated in the study, comprising a set of 35 college athletes and a set of 30 professional athletes, all of whom completed follow-up assessments. There were no statistically significant differences observed in age, sex, BMI, and sports grade between the two groups. Professional athletes spent more time in their current sport (t(63) = −2.67, p = 0.010) and had lower education levels (χ2(1) = 7.71, p = 0.006) compared with college athletes. Salivary cortisol levels were higher in college athletes than professional athletes at BL (t(63) = 3.74, p < 0.001), after SD (t(63) = 2.65, p = 0.010), and at ROC (t(63) = 2.32, p = 0.024). Professional athletes exhibited higher levels of state anxiety (t(63) = −2.31, p = 0.024) and depression (t(63) = 2.75, p = 0.008) at BL. No significant differences between the two sets were found in scores of GAD-7, ISI, total sleep duration, and pre-ROC sleep problems (Table 1).
Basic demographic and sleep characteristics
Based on the competition performance in the most recent official competitions (ROC) approximately 45.03 (±16.68) d after BL, athletes were categorized into three groups: high performance (HIGH), normal performance (NOR), and low performance (LOW) based on a consensus of both subjective and objective evaluations. In the college athletes set, there were 6 athletes in the HIGH group, 19 in the NOR group, and 10 in the LOW group. In the professional athletes set, there were 4 athletes in the HIGH group, 20 in the NOR group, and 6 in the LOW group. There were significant differences in anxiety (HIGH, 34.80 ± 1.63; NOR, 41.43 ± 1.78; LOW, 41.72 ± 1.36; F(2,62) = 4.06, p = 0.022, n = 65) and cortisol (HIGH, 39.45 ± 16.06; NOR, 83.13 ± 11.60; LOW, 125.27 ± 18.80; F(2,62) = 4.70, p = 0.013, n = 65) among the performance groups, with the HIGH group showing significantly lower anxiety and cortisol levels at ROC (Table 2).
The anxiety and cortisol levels among the three performance groups at BL, SD, and ROC
Anxiety and cortisol levels after 24 h SD were equaled with those of ROC
In the college athletes group (n = 35), there was no difference in the scores of state anxiety between SD and ROC; however, the anxiety scores in both SD and ROC were significantly higher compared with BL (F(2,32) = 7.80, p = 0.001; SD vs BL, p = 0.013; ROC vs BL, p < 0.001; Fig. 1B). This trend was also observed in the professional athletes’ group (n = 30; F(2,27) = 5.76, p = 0.007; Fig. 1D). Furthermore, the anxiety scores after SD were positively correlated with those of ROC (college athletes set, r = 0.39, p = 0.020; professional athletes set, r = 0.39, p = 0.035; Fig. 1B,D).
Additionally, salivary cortisol levels, an objective indicator of stress, also did not show any significant difference and were positively correlated between SD and ROC (college athletes set, r = 0.48, p = 0.004; professional athletes set, r = 0.77, p < 0.001) and both SD and ROC higher than BL (college athletes set, F(2,32) = 3.62, p = 0.031; professional athletes set, F(2,27) = 6.29, p = 0.003; Fig. 1C,E). These findings suggest that the 24 h sleep deprivation paradigm can effectively increase stress levels and equal with the stress experienced in ROC, for both college and professional athletes.
However, there were no differences in anxiety or cortisol levels at BL, after SD, and between SD changes among the HIGH, NOR, and LOW groups (Table 2). This suggests that competition performance cannot be predicted by anxiety and cortisol levels at BL or after SD.
SD-changed incongruent Stroop response predicts competition performance
Following sleep deprivation, all athletes showed decreased response times to color–word incongruent stimuli in Stroop (Icon RT; t(64) = 3.36, pFDR = 0.009). There were no significant changes observed in other cognitive indicators (Table 3). Then we compared differences between three performance groups for all cognitive indicators in college athletes set, and we found that among the different performance groups, a significant difference was observed in the changed Icon RT after SD (ΔIcon RT, F(2,32) = 3.78, p = 0.034) and the selection of noise intensities in CRTT (SD_Noise Intensity, F(2,32) = 6.03, p = 0.006; Table 4). After that, we performed the ordered logistic regression model to analyze professional athletes set on the cognitive indicators (SD_Noise Intensity and ΔIcon RT) that were significantly different between the three groups identified in the college athletes set. The logistic regression model just revealed a negative correlation between ΔIcon RT and competitive performance, where a decrease in ΔIcon RT was associated with better performance in ROC [HR (95% CI) = 5.29 × 10−4 (1.64 × 10−5–0.02), p = 0.015] after adjusting for sex, age, sports type, sports grade, and pre-ROC sleep issues (Fig. 2A). Among the different performance groups, there also had a significant difference observed in ΔIcon RT in ensemble set (F(2,62) = 7.88, p < 0.001; Fig. 2B). Compared with BL, Icon RT was significantly decreased after SD for both the HIGH (t(62) = 4.09, p < 0.001) and NOR (t(62) = 3.41, p = 0.003) groups, while no significant difference was observed in the LOW group (t(62) = 1.18, p = 0.563; Fig. 2C). Post hoc analysis showed ΔIcon RT was lower in the HIGH group than that in the LOW group (t(62) = 3.90, p < 0.001, Fig. 2B).
The SD-induced Stroop ΔIcon RT has predictive effects on athletes’ performance in competitions. A, The ordinal logistic regression analysis for assessing the association between cognitive indicators with athletes’ performances in professional athletes set (n = 35). B, The ΔIcon RT in Stroop were different among the three performance groups in ensemble set (n = 65). C, The Icon RT in Stroop at BL and SD in different performance groups in ensemble set (n = 65). BL, in baseline; SD, after sleep deprivation; ΔIcon RT, the change of response time under color–word incongruent stimuli in Stroop after 24 h of sleep deprivation; SD_Noise Intensity, the selection of noise intensities in CRTT after 24 h of sleep deprivation; HIGH, high-performance group; NOR, normal-performance group; LOW, low-performance group; Z, standardized score. **p < 0.01, ***p < 0.001.
The cognitive performance at baseline and sleep deprivation
Cognitive performance differences among the three performance groups
However, in the ensemble set, there were no significant correlations between ΔIcon RT and anxiety levels in both SD (r = −0.02, p = 0.909) and ROC (r = −0.02, p = 0.897), and neither did the cortisol levels (SD, r = −0.15, p = 0.270; ROC, r = −0.01, p = 0.966). The path analysis indicated that ΔIcon RT directly affected competition performance (β = −0.48, p < 0.001), while the anxiety level after SD indirectly impacted performance through ROC anxiety levels (mediating effect value, −0.13%; 95% CI: −0.25 to −0.05), as well as the cortisol levels (mediating effect value, −0.25%; 95% CI: −0.44 to −0.11; Fig. 3).
Path analysis for the predictive effects of cortisol, anxiety, and ΔIcon RT on the athletes’ performances in recent official competitions. The dotted line indicates that the path coefficients are not significant, the solid line indicates that the path coefficients are significant, and the path coefficient values are on the line. The model was adjusted for age, sex, sports type, and sports grade. ΔIcon RT, the change of response time under color–word incongruent stimuli in Stroop after 24 h of sleep deprivation. **p < 0.01, ***p < 0.001.
The mediating neuroelectrophysiological mechanism underlying the predictive effects of ΔIcon RT on competition performance
To increase the statistical power, the EEG analysis combined college athletes and professional athletes into a single set (n = 65). Results showed that under Stroop color–word incongruent stimuli, the mean amplitude of the P3 component in C3 channel (C3-icon-P3, r = −0.46, pFDR = 0.036; Fig. 4A), and the mean amplitude of the P3 component in Fz channel (Fz-icon-P3, r = −0.37, pFDR = 0.048) were significantly correlated with ΔIcon RT (Table 5). C3-icon-P3 exhibited significant differences (F(2,62) = 4.17, p = 0.025) among the HIGH, NOR, and LOW groups, but Fz-icon-P3 did not reach significance. Specifically, the mean amplitude of C3-icon-P3 was significantly higher in the HIGH (t(62) = 2.61, p = 0.026) and NOR (t(62) = 2.36, p = 0.046) groups than in the LOW group (Fig. 4B).
P3 component mediates the predictive effects of Stroop ΔIcon RT on competitive performance. A, The correlation between the change of response time (ΔIcon RT) and the P3 component in the C3 channel (C3-icon-P3) under color–word incongruent stimuli in Stroop (n = 65). B, The mean amplitude C3-icon-P3 in Stroop were different among the three performance groups [the gray-shaded regions represent the time window for P3 component measurement (300–450 ms), and the white circle indicates the location of Channel C3]. C, The mediation model showed that C3-icon-P3 largely mediates the effects of the ΔIcon RT on competition performance. HIGH, high-performance group; NOR, normal-performance group; LOW, low-performance group; ΔIcon RT, the increased response time to color–word incongruent stimuli in the Stroop after sleep deprivation; C3-icon-P3, the mean amplitude of the P3 component in the C3 channel under the color–word incongruent stimuli in the Stroop; performance, the athletes’ performance in recent official competition. *p < 0.05.
The correlation between the P3 component and ΔIcon RT under Stroop color–word incongruent stimuli
Using C3-icon-P3 as the mediator (M), we analyzed the mediating effect of ΔIcon RT in Stroop (X) on competition performance (Y). The findings revealed a significant total effect (c = 0.48, p = 0.006), along with a significant indirect effect (ab = 0.21%, 95% CI: 0.033–0.47, p = 0.026), while the direct effect (c' = 0.27, p = 0.143) was found to be insignificant. Therefore, C3-icon-P3 demonstrated a complete mediating role, explaining 43.8% of the total variance in the relationship between ΔIcon RT and competition performance (Fig. 4C).
Additionally, time–frequency analysis identified the frequency spectrum (delta, theta, alpha, sigma, beta) of C3-icon-P3 (300–450 ms) under color–word incongruent stimuli in Stroop. It showed a significant difference only in the delta (1–4 Hz) power among the three performance groups (F(2,62) = 4.21, p = 0.031). Post hoc tests revealed that the delta power in the HIGH group was significantly higher than that in the LOW group (t(62) = 4.44, p = 0.021; Fig. 5). Consequently, the mediation effects of ΔIcon RT, relating C3-icon-P3 to sports performance, were predominantly linked to delta oscillation.
There was a significant difference in the delta power of the P3 component among the three performance groups. A, The time–frequency power in Stroop between different performance groups. The rectangular part indicates the EEG delta band (1–4 Hz) in the time window for the P3 component (300–450 ms). B, The averaged power of different frequencies of the P3 component between different performance groups. HIGH, high-performance group; NOR, normal-performance group; LOW, low-performance group; C3-icon-P3, the mean amplitude of the P3 component in the C3 channel under the color–word incongruent stimuli in the Stroop. *p < 0.05.
Discussion
This study aimed to identify the predictor with practical value for the athletic competition performance level from cognitive and anxiety changes after sleep deprivation. We found that 24 h SD could effectively increase stress levels and equal with those of ROC; however, that competition performance cannot be predicted by anxiety and cortisol levels at BL or after SD. The decreased response time for Stroop incongruent stimuli following SD could specifically and negatively predict possible better performance in near future competitions among both college and professional athletes. Notably, the corresponding P3 component at the C3 channel, specifically the delta oscillation, was found to mediate this effect.
Numerous studies have consistently reported higher levels of psychophysiological stress indices before competition compared with training (Papacosta et al., 2016; Souza et al., 2019). Our present findings, demonstrating an increase in cortisol levels and anxiety ratings after 24 h SD, align with previous research (Wright et al., 2015; Nollet et al., 2020). Furthermore, a significant correlation was observed between 24 h SD and competition in state anxiety and cortisol levels. It suggests that the physiological and psychological stress level of sleep deprivation may equate with the those experienced just before competition. However, there are much more complex stressors [e.g., expectations (Lagamon et al., 2019), competitors (Hettinga et al., 2017), and competition environment (Arruda et al., 2014)] than sleep deprivation. Previous studies have often suggested that better emotional regulation leads to better performance (Myall et al., 2023). Although the stress state was highly correlated with sports performances at the ROC, the stress state after SD could not predict the ROC performance in this study, which may be due to the stress state not reflecting the coping and management ability under competition stress.
Some studies have introduced the Attentional Control Theory-Sport, which emphasizes the important role of attentional processes for sports performance under stress (Albaladejo-García et al., 2023). Changes in attention control after one night of SD may reflect the ability to cope with stress in a temporary manner (Kalanthroff et al., 2016). Our findings showed that SD-changed attention control for conflict processing, as indicated by ΔIcon RT, could be used as an early indicator and potentially predict upcoming performance in athletes. Although most previous studies show the Stroop response time was increased after SD (Kim et al., 2011; Floros et al., 2021), a study involving healthy, young, and physically active adult males observed a decrease in reaction time after sleep deprivation, with a more pronounced decrease in the control group (Skurvydas et al., 2021). This suggests that the Stroop response time could be trained in physically active adults just for 1 d, which may be hindered by SD. Our study further supports that higher resilience for conflict attentional control after SD may serve as an index of excellent stress coping and can more reliably predict good performance of the most recent official competitions (ROC).
Studies on the neural oscillations of athletes and their relation to performance are still lacking. It is theorized that the amplitude of P3 reflects the level of attention allocated to cognitive tasks (Kok, 2001). Larger P3 amplitudes suggest that the brain can effectively manage conflict to enhance processing efficiency (Polich, 2007). Previous studies have shown a correlation between shorter Stroop response times and larger P3 amplitudes in athletes (Wang et al., 2022). Our results indicate that more significant reductions in reaction time of icon-stimulate in the Stroop task following sleep deprivation are correlated with increased P3 amplitudes, as well as better performance in competition. This suggests that high-performance athletes can harness additional cognitive resources for stress management to enhance their performance in the Stroop task after sleep deprivation, enabling them to perform at a higher level of the competitions. Specifically, delta oscillations within the P3 component are crucial for attention (Schürmann et al., 2001; Knyazev, 2012), and the increase in delta power of P3 in the central region was negatively correlated with decreased executive function (Bong et al., 2020). Our results highlight the significant role of delta oscillations in attention control under stress among athletes.
Several limitations need to be acknowledged in our study. Firstly, the cognitive effects of sleep deprivation (SD) may vary among different types of athletes (Maruo et al., 2018). Unfortunately, we only controlled for sport type as a covariate and did not separately analyze skill versus strength or individual versus team sports in this study due to the limited sample size. All athletes included were in winter sports, as this study was supported by the Beijing Winter Olympics Special Fund, and the main findings may require further validation in summer sports. Secondly, the applicability of the research results needs to be verified in a broader athlete population, including elite athletes and the general population engaged in sports and exercise. Due to practical constraints, no measures were taken in competition. The state anxiety levels and saliva samples were corrected at 8 A.M. on competition day (before the competition). Further in-competition assessments are desired to validate the predictive effects. Lastly, our follow-up only extended to the official competition after 1–2 months, and the long-term predictive effects on subsequent competition performance were not evaluated.
In summary, anxiety and stress state after 24 h SD were equaled with those of official competition, but there were no differences between the three performance groups. The 24 h SD-decreased response time for Stroop incongruent stimuli has specific predictive effects on the higher performance of both professional and college athletes in the upcoming official competitions. This predictive effect was mainly mediated by parietal delta oscillations when processing incongruent stimulation. Our findings offer practical applications for athletic training and competition strategies, and coaches can identify athletes who had smaller cognitive buffer capacity after SD, informing poor pre-event preparations. Neurophysiological insights may guide interventions to boost cognitive function and performance, making our research valuable for enhancing athletic readiness and success.
Footnotes
This work was supported in part by the STI2030-Major Projects (2022ZD0211200), National Key Research and Development Program of China (2021YFF0306502), and National Natural Science Foundation of China (82171488 and 82471515).
The authors declare no competing financial interests.
- Correspondence should be addressed to Yan Sun at sunyan{at}bjmu.edu.cn or Jie Shi at shijie{at}bjmu.edu.cn.
