Altered brain hemodynamic response and cognitive function after sleep deprivation: a functional near-infrared spectroscopy study

Abstract

Aim: Acute sleep deprivation has revealed altered executive cognitive functioning, including attention, working memory, and interference resolution. In the present study, we investigated brain hemodynamics and cognitive function measures using functional near-infrared spectroscopy (fNIRS).

Methods: fNIRS was used to record cortical brain activity in a resting state scan, flanker task, and n-Back tasks under well-rested (WR, >6.5 h sleep for ≥3 days) and sleep-deprived conditions (SD, 28 h awake) in a repeated-measures study in 20 participants (7 female, mean age 20.1 ± 0.413).

Results: Our results indicate decreased accuracy rates in the flanker task in the SD condition, suggesting reduced response inhibition capacity. In the flanker task, a significant increase of beta estimates was observed in the SD condition in the right dl-PFC, suggesting potential compensation (BH-FDR corrected p = 0.0418, t(19) = 3.51). There were no significant differences between WR and SD in n-back task accuracy rates or reaction times or in Flanker task reaction times. Otherwise, functional connectivity strengths were not significantly different between the WR and SD sessions on a group-level analysis for the resting state and all tasks (BH-FDR correction, p > 0.05).

Conclusion: Our results provide insight into how insufficient sleep patterns affect interference resolution, and that fNIRS can reveal functional connectivity differences inter- and intra-individually in SD conditions.

Keywords:

• Total sleep deprivation
• functional near-infrared spectroscopy (fNIRS)
• Flanker
• N-back
• resting-state
• functional connectivity

1. Introduction

Sleep is a natural biological process, necessary for brain function and other physiological processes such as metabolism, immune and cardiovascular function, and appetite regulation. Sleep deprivation is a sleep loss status with long-term risks that contribute to cardiovascular disease, type-2 diabetes mellitus, anxiety, depression, and gastrointestinal disorders [1]. Acute sleep deprivation also has effects on learning, memory, attention, and other executive functioning [2].

However, it is estimated that between 50 and 70 million Americans experience chronic sleep disorders that result in insufficient or inconsistent sleeping patterns that affect daily livelihoods. A trend in increasing sleep disruption has been largely caused by a wave of societal changes, including an increase in shift work, work hours, and the use of technology. Young adults and college-aged students especially report alarming rates of decreased quality and increased restrictions on sleep [3].

Research has consistently demonstrated that acute sleep deprivation impairs various aspects of executive function, inhibiting general attentional and mnemonic abilities, and influencing attention, working memory, focus, and self-regulation [4]. Studies using visuospatial N-back tasks have shown a decreased performance of working memory and decreased visuospatial capacity under conditions of sleep deprivation, including decreased accuracy and increased response time, as well as decreased task-related activation in the frontal and parietal regions [5–7]. Additional executive functions such as performance monitoring and filtering have also been shown to be impaired, with increased response times and decreased accuracy rates measured during Flanker tasks [8,9].

Decreased resting-state functional connectivity (RSFC) has been demonstrated in the right inferior parietal lobules and the left precuneus and posterior cingulate cortex after sleep deprivation [10]. Additionally, Horovitz and colleagues found significantly reduced RSFC interconnections between the medial prefrontal cortex (PFC) and the inferior parietal lobules (IPL) after acute sleep deprivation [11].

Functional near-infrared spectroscopy (fNIRS) is an optical imaging technique that utilizes light in the near-infrared spectrum (650–950 nm) to allow continuous, non-invasive measurements of oxygen-dependent responses due to neural activity [12]. As an emerging neuroimaging method, fNIRS operates on the idea of neurovascular coupling, in which brain activation during rest or tasks causes increased regional brain metabolism and oxygenation demands by neurons, altering regional cerebral blood flow (rCBF) and volume. fNIRS measures changes in both oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (HbR) by measuring variations in absorbance between an optode source and detector. Similar to fMRI, which estimates blood-oxygen-level-dependent (BOLD) signals from changes during blood de-oxygenation events using changes in blood magnetization, fNIRS provides an indirect measure of brain activity, with hemodynamic response peaks typically delayed between 5–10 s. Although fNIRS provides poor spatial resolution due to the nature of the loss of near-infrared light deeper into the scalp, fNIRS provides unique advantages because it is inexpensive, highly portable, and robust against head motion, making it especially advantageous in natural environments or certain populations such as children.

fNIRS has demonstrated a significant correlation with the BOLD signals that have also been measured by fMRI [13,14] and has revealed significant change between populations under executive tasks [15,16]. Significantly, fNIRS has also been instrumental in constructing human functional brain networks using resting-state data that were highly reproducible and consistent with findings from other neuroimaging techniques [17].

fNIRS has also been previously used to investigate phenomena of sleep deprivation on driving [18], time perception [19], and working memory tasks [20,21], consistently demonstrating altered oxygenation responses in the prefrontal cortex and impaired cognitive function. A decrease in prefrontal activity has been observed, which may explain cognitive decline [20,22], however, an increase in the region has also been documented [23], which may reflect compensatory mechanisms to overcome sleep deprivation when performing tasks. Resting-state data utilizing fNIRS following acute sleep deprivation is less explored. Ahn et al. [18], who tested drivers’ mental fatigue following sleep deprivation using fNIRS, proposed collecting resting state data to improve data normalization.

Thus, to further understand the relationships between sleep deprivation and neural metabolism underlying executive function, as well as to further explore the feasibility of fNIRS to reliably measure hemodynamic response during rest and tasks following sleep deprivation, the present study uses fNIRS to investigate neural processing during neurobehavioral tasks. An N-back task and Flanker task were designed to isolate neural mechanisms underlying working memory and performance monitoring. Existing fNIRS and fMRI literature suggest that compared to well-rested individuals, sleep-deprived individuals will experience reduced cognitive function (reduced accuracy rates and increased response time on tasks). Thus, we hypothesize that cognitive function and the hemodynamic response will both be decreased after sleep deprivation.

2. Methods

2.1. Participants

Twenty healthy participants [18–23 years old, 7 female, mean age 20.1 ± 0.413] were enrolled in this study through advertisements. All subjects were college students recruited from the community in the northern New Jersey, USA area between July 2019 and August 2019 such that classwork or work schedules would not conflict with the research study. Participants were recruited based on the following criteria: (1) being right-handed, (2) having no history of sleep, psychological, or neurological disorders, (3) no comorbidities, (4) no history of drug or alcohol abuse and no use of nicotine within the past year, (5) no long-term medication, and (6) healthy sleeping habits, including consistent sleep with a minimum of 6.5 h per night, (7) no shift work, and (8) none or minimal use of caffeine. All research study procedures were approved by the Institutional Review Board (IRB) of the New Jersey Institute of Technology (NJIT). All subjects provided written informed consent at each stage of their research trial according to approved guidelines and all participants received $100 in compensation for their participation in this study.

2.2. Neuroimaging methods

Neurobehavioral activity was measured using a continuous wave fNIRS system with 690 and 830 nm lasers. HbO, HbR, and total hemoglobin concentrations were measured at a sampling frequency of 50 Hz. A customized, 22-channel optode layout, was utilized to produce and position an optode cap on each subject (Figure 2). The optodes chosen were intended to collect hemodynamic data from regions of the brain associated with lowered executive function found in individuals after sleep deprivation. These included cortical regions of interest (ROIs) in the dorsolateral prefrontal cortex, the superior medial gyrus, the inferior parietal lobule, and the frontal eye fields. The locations of optodes were obtained using Brain Sight neural navigator (Rogue Research Inc.’s Neuronavigation System, Canada). Optode placement was guided by pre-defined ROI and literature. 18 of 22 detectors were placed at a distance of 30 mm from the source optode, recording hemoglobin concentration changes. The four remaining detectors (short-source separation detectors) were placed at a distance of 8.4 mm from the source optode, measuring physiological activity from the extracerebral layers.

2.3. Experimental design

A repeated-measures study was conducted to investigate the effects of sleep deprivation on short-term executive function. All subjects first underwent a screening session, in which all expectations and procedures were communicated, and eligibility was validated. During this screening session, participants were also fitted with a wrist fitness tracker (Letscom, USA) to monitor sleep patterns from the participants’ residences through the course of the study. Participants were provided instructions on how to utilize the corresponding app on their phones. The participants were also familiarized with the tasks they would be performing during their scanning sessions, in part in efforts to reduce order effects.

Each participant’s first neuroimaging scanning session under the rested wakefulness condition occurred after at least three days of consistent sleep (>6.5 h), validated by data from the wrist activity monitor. The first scans were performed in the morning and early afternoon at a time of the participants’ preference, however, times available were limited from 10 AM to 2 PM to reduce the effects of scan time throughout the day.

Participants completed their second scanning session under conditions of acute sleep deprivation for 28 h, which was also validated by the wrist activity monitor. All sleep deprivation scanning sessions were scheduled 28 h after participants normally woke up in order to maintain consistent sleep quality and control for each participant’s unique circadian rhythms. Behavioral and neuroimaging data were collected at both scans in order to fully elucidate neural mechanisms.

2.4. Neurobehavioral tasks

Subjects participated in three computerized neurobehavioral tasks for both the baseline rested scan and the sleep-deprived scan. The same task protocol was utilized for both scanning sessions. Each participant performed a resting state scan for 7 min, as well as an N-back task, and a Flanker task. All participants practiced each task before beginning fNIRS scanning. All behavioral tasks were generated and presented using behavioral research tools from E-Prime Psychological Software Tools.

2.4.1. Flanker task:

A Flanker task was utilized to measure response interference resolution and abilities to filter inputs of data [24]. Reaction time and accuracy percentage were measured while participants were asked to identify the direction of a central arrow stimulus with either the same arrows (congruent: ≪≤≪, ≫≥≫) on both sides, or arrows pointing in opposite directions (incongruent: ≪≥≪, ≫≤≫) flanking its sides. Each stimulus was presented for 2 s, with congruent and incongruent stimuli presented in a pseudo-random order. A block design was utilized, with 4 blocks consisting of 40 images each, with 30 s of rest in between blocks (Figure 1(A)).

Figure 1. (A) Flanker Task paradigm with alternating blocks of 30 s rest and 80 s task shown. The Flanker stimuli were further separated into congruent and incongruent stimuli, in a pseudo-random order. (B) Two N-back tasks were performed for each participant, one for 0-back and another for 2-back. All participants performed the Flanker and N-Back tasks by following instructions on a computer screen, through E-Prime psychological software tools.

2.4.2. N-back task:

Participants were presented with two N-back tasks: a 0-back task and a 2-back task. The N-back task activates working memory in the dorsolateral prefrontal cortex and the parietal cortex [25]. During the 0-back task, a new letter was presented every 2 s, and subjects were asked to identify whether the letter presented was either a “Z,” or a different letter, as quickly and as accurately as possible. During the 2-back task, a new letter was presented every 2 s, and subjects indicated whether the letter presented was the same as the letter displayed 2 letters back. A block design was utilized, with 6 blocks consisting of 22 letters that were presented in one block for 44 s, with 30 s of rest in between blocks (Figure 1(B)).

2.5. fNIRS Data quality

During the scanning procedure, the fNIRS data were checked for motion artifacts. Noisy channels were marked directly in HOMER2 after each participant’s scan run. In order to ensure adequate source-to-scalp coupling, the light sources were adjusted to provide the maximum signal-to-noise ratio, as assessed by the graphic user interface in HOMER2. The raw data were visually inspected for any motion artifacts. Due to the relatively small sample size, no channels were removed from analysis, except the short separation channels which were regressed out. Lastly, no participants were excluded from the study due to excessive head motion, since the heart rate signals were observable in the raw time series data. The percentage of head motion artifact was calculated for each channel and each subject’s raw time series data. An artifact was defined as an abrupt change in the signal greater than or less than 3 standard deviations from the mean of all the difference points (current time point – previous time point) in the raw data. Additionally, the percentage of head motion artifact was calculated as follows: (# of abrupt change points/total # of change points) * 100. For all rest and task conditions, and both rested and sleep-deprived sessions, the head motion artifact was less than 0.5% after averaging across each subject and channel (Supplementary Table 1). Therefore, no subject’s data were discarded due to head motion artifact, since minimal artifacts were observed and would be corrected for in the preprocessing step.

2.6. Preprocessing of fNIRS data

All preprocessing of fNIRS data was performed using HOMER2, an open-source MATLAB-based toolbox [26]. The preprocessing pipeline consists of the following steps: conversion of intensity to optical density values, wavelet-based head motion correction [27], band-pass filtering, conversion of optical density to HbO, HbR, and HbT concentrations, physiological noise removal, and 3rd-order polynomial drift correction (Figure 2). For the fNIRS task data, a band-pass filter of 0.007 to 0.15 Hz was applied, while for resting-state data, a band-pass filter of 0.01 to 0.1 Hz was applied. Resting-state signals are commonly reported to be within 0.01 to 0.1 Hz [28]. The filter frequency range is different for task data, since the fundamental frequencies of the task data are 0.009 Hz and 0.0135 Hz, for Flanker and N-Back tasks, respectively. Generally, a cut-off frequency value is chosen to include as much information derived from the fundamental task frequency as possible, while accounting for very low-frequency drifts, so we chose a high-pass frequency cut-off at 0.007 Hz [29]. We chose a low-pass cut-off value of 0.15 Hz, to minimize physiological artifacts from respiration (∼0.2 Hz). The optical density values were converted to HbO, HbR, and HbT concentrations using the modified Beer–Lambert Law [30]. Although optical density was converted to HbO, HbR, and HbT concentrations, only HbO concentrations were used for subsequent analyses since they yielded the highest signal-to-noise ratio. Physiological noise removal was performed using the short source-detector separation channels, in which extracerebral noise is regressed out using the short source-detector channel with the highest correlation to the long source-detector separation channel [31].

Figure 2. Preprocessing pipeline for fNIRS data. The same pipeline is used for all data, with the only difference between preprocessing resting-state and task data is in the band-pass filtering step. Channels are indicated in red numbers. Regions of interest: DLPFC (±30, 36, 42), SMG: (−2, 28, 42), IPL: (±42, –42, 54), FEF (±30, 4, 46).

2.7. Resting-state functional connectivity analysis

Resting-state functional connectivity (RSFC) analyses were performed using in-house MATLAB scripts. Following preprocessing of resting-state data, the first two minutes of data were removed to allow the resting-state signal to reach steady-state [32]. The resting-state time-series data from each of the 18 long source-detector separation channels were correlated with each other using Pearson’s r correlation, yielding an 18x18 correlation matrix. The mean RSFC for both rested and sleep-deprived sessions were calculated by averaging each channel pair’s Pearson’s R-score across all subjects. Additionally, the r-scores in each subject’s correlation matrix were transformed to Fisher’s Z-scores, and then a paired t-test was carried out between the rested and sleep-deprived sessions, for all channel pairs. Following the paired t-test, a correction was performed for multiple comparisons using Benjamini-Hochberg’s False Discovery Rate (BH-FDR) correction [33].

2.8. Task-based functional connectivity analysis

After preprocessing the fNIRS task data, baseline corrections were performed to account for intra-individual variability of the fNIRS signal in each session. The baseline was defined as the average HbO concentration during the −5 s time period before the trial stimulus onset. This average baseline value was subtracted from the rest of the block, which consisted of the task and rest duration, for each block trial in the tasks. After baseline correction was performed for each block, the blocks were concatenated together to yield continuous HbO concentration time series data. Subsequently, the time-series data for each channel was correlated in a pair-wise manner with all the other channels, yielding an 18 × 18 Pearson’s r correlation matrix. Similar to the RSFC analyses, Pearson’s R-scores were converted to Fisher’s Z-scores, and paired T-tests were performed comparing sleep-deprived and rested sessions, along with BH-FDR correction. The functional connectivity strengths (Z-scores) were then correlated with behavioral task measures such as response time and accuracy.

2.9. Task activation from general linear model

The preprocessed fNIRS task data were block averaged and baseline corrected to −5 s before the stimulus onset. For the Flanker task, the window of interest was from −5 s to 110 s, and for the N-back tasks, the window of interest was from −5 s to 74 s; this includes both task and rest (30 s) periods. A general linear model (GLM) was then fitted to the data using an ordinary least squares method in HOMER2 with the temporal basis function set as a gamma function [idx Basis = 2]. Beta estimates were then compared between rest and sleep-deprived sessions, for all tasks.

3. Results

3.1. Behavioral results – Flanker task

Accuracy scores for the Flanker task were significantly greater in individuals during the rested condition compared to the sleep-deprived condition (rm-ANOVA, F(1,16) = 6.02, p = 0.024, partial η² = 0.273). Additionally, significantly lower accuracy scores were observed during the incongruent stimulus compared to the congruent stimulus for both sessions (F(1,16) = 35.18, p = 1.039*10⁻⁵, partial η² = 0.687) (Figure 3(A)). No significant differences were observed in the response time between the rested and sleep-deprived sessions (F(1,16) = 0.76, p = 0.40, partial η² = 0.045); however, the response time was significantly greater for the incongruent stimuli than congruent stimuli (F(1,16) = 111.61, p = 2.16*10⁻⁹, partial η² = 0.875) (Figure 3(B)).

Figure 3. Mean behavioral task outcomes, for both rested and sleep deprived sessions, for (A) Flanker task accuracy, (B) Flanker task response time, (C) N-Back task accuracy, and (D) N-Back task response time. Error bars indicate standard error of the mean (SEM). Significant codes: ***p < 0.001, **p < 0.01, *p < 0.05.

3.2. Behavioral results – N-back task

Individuals performed significantly worse on the 2-back task than 0-back task, both in accuracy (F (1,15) = 37.93, p = 8.15*10⁻⁶, partial η² = 0.717) and response time (F(1,15) = 41.50, p = 4.61*10⁻⁶, partial η² = 0.735) for both sessions. However, no significant differences were observed between the rested and sleep-deprived sessions for both accuracies (F(1,15) = 0.038, p = 0.85, partial η² = 0.003) and response time (F(1,15) = 1.72, p = 0.206, partial η² = 0.103) in both 0-back and 2-back conditions (Figure 3 (C and D)). We found an interaction effect in response time between the N-back task (0-Back vs. 2-Back) and sleep condition (rested vs. sleep deprived) (F(1,15) = 12.54, p = 0.002, partial η² = 0.455). However, post-hoc paired Sample t-tests with Bonferroni’s multiple comparison correction at α = 0.05/2 = 0.025 yielded no significant differences (0-Back: t(18) = −1.428, p = 0.1705; 2-Back: t(18) = 2.403, p = 0.0273).

3.3. Resting-state functional connectivity

Although intraindividual variabilities in RSFC are present between the rested and sleep-deprived sessions, these differences are not significant (paired-sample t-test, BH-FDR correction, p > 0.05) (Figure 4). In particular, subject #18 saw a large decrease in RSFC in the sleep-deprived session compared to the rested session (Figure 4). On a group-level, RSFC is higher overall within the right dorsolateral prefrontal cortex compared to other brain regions (Figure 5).

Figure 4. Individual variabilities in RSFC maps shown for all subjects, during the (A) rested and (B) sleep deprived sessions.

Figure 5. Group-averaged RSFC for both (A) rested and (B) sleep deprived sessions. No significant differences are observed between the two sessions (paired-samples t-test, BH-FDR corrected p > 0.05).

3.4. Task-based functional connectivity

For Flanker, 0-Back, and 2-Back tasks, activation is seen primarily in the right and left dorsolateral prefrontal cortices, shown in the task-based functional connectivity maps (Figure 8). The functional connectivity strengths were not significantly different between the rested and sleep-deprived sessions, for all tasks (paired-samples t-test, BH-FDR correction, p > 0.05). However, correlation of the functional connectivity strength with behavioral task measures showed significant differences in the 2-Back task mean response time, for the rested session only (Figure 6(A)) (paired-Samples t-test, BH-FDR corrected p = 0.01956). This was observed in channel pair 11–16, located within the left inferior parietal lobule. This relationship is not significant in the 0-back task for the same brain region (Figure 6(B)).

Figure 6. (A) Significant correlation between functional connectivity strength (z) and mean response time for the 2-back task during the rested session, in channel 11–16 (inferior parietal lobule), but not observed in (B) the 0-back task. BH-FDR correction was performed for all channel pairs.

3.5. Task activation – GLM

The beta estimates from the GLM are of greater magnitudes in the Flanker task, compared to the N-back tasks (Figure 7). In the Flanker task, significant differences between the rested and sleep-deprived sessions were observed in channel 5, corresponding to the right dlPFC (Paired-sample t-test, BH-FDR corrected p = 0.0418, t(19) = 3.51). No significant differences were observed in the 0-Back and 2-Back tasks, between rested vs. sleep-deprived sessions (p > 0.05).

Figure 7. (A) Flanker Task beta estimates with significant difference between rested and sleep-deprived condition in channel 5 (right dlPFC) (paired-sample t-test, BH-FDR corrected p = 0.0418, t(19) = 3.51) (B) 0-Back task beta estimates, no significant differences (p > 0.05) (C) 2-Back task beta estimates, no significant differences (p > 0.05).

Figure 8. Task-based group-level functional connectivity maps demonstrate activation primarily seen in right and left dorsolateral prefrontal cortices (channels 1–8). Functional connectivity strengths not significantly different between rested and sleep deprived conditions: (A) Flanker Task (B) 0-Back task (C) 2-Back task.

4. Discussion

In the more challenging behavioral tasks such as the 2-back task and Flanker-incongruent tasks, we observed decreased performance when evaluating response time and accuracy. No significant differences were observed after sleep deprivation for both the 0-Back and 2-Back tasks. Our results suggest that working memory is unimpaired after sleep deprivation in this cohort of students. However, significant differences were observed after sleep deprivation, on measures of accuracy for the Flanker task (both incongruent and congruent stimuli), but not on measures of response time. Brain activation also significantly increased in the sleep-deprived condition during the Flanker task in channel 5, associated with the right dorsolateral prefrontal cortex. Interestingly, Honma et al. [23] have demonstrated an increased right prefrontal cortex activity reflect an ability to overcome sleepiness to maintain cognitive performance. Since the Flanker task measures response inhibition, our results demonstrate response inhibition capacity is decreased after total sleep deprivation, also reported in several other studies [34,35]. However, this study only investigated one session of total sleep deprivation and a more longitudinal study would be required to further investigate the effect of lowered response inhibition due to sleep deprivation on obesity or other eating disorders. Furthermore, response time during the Flanker task is not significantly different after sleep deprivation.

Intraindividual and interindividual differences are present in the RSFC maps between the rested and sleep-deprived sessions. However, these differences are not significant in a group-level analysis. This supports previous fNIRS findings of large inter- and intra-individual variabilities in RSFC derived from fNIRS [36]. Although confounds such as sleep deprivation may enhance the intraindividual differences, it appears that sleep deprivation does not affect RSFC when group-averaging is performed. There were no significant differences in RSFC between the rested group and sleep-deprived group when paired-samples t-tests were performed (FDR-corrected p > 0.05 for all channel pairs). This suggests that group-level analysis of RSFC is robust to the effects of sleep deprivation, however, a larger sample size and the investigation of a larger number of brain regions are still needed.

Similar to the RSFC maps, the task-based functional connectivity maps showed no significant differences between the rested vs. sleep-deprived sessions (paired-samples t-test, FDR-corrected p > 0.05). Regions that had high levels of functional connectivity in the resting-state condition appear to be apparent in the task-based functional connectivity maps, as seen in Channels 1–8, which cover the right and left prefrontal cortices. It is important to note that for the Flanker task, the task-based functional connectivity map shows the activity due to the whole task (80 s) and does not separate out the incongruent vs. congruent stimuli, since the stimuli were presented closely in time to each other (2 s).

Several limitations of this study should be addressed. First, subjects performed their first neuroimaging session under the rested condition and returned for the second session under sleep-deprived conditions. Since the rested and sleep-deprived sessions were not counterbalanced, it is possible that confounds due to order effects may be present in this study. This order was ultimately chosen because the amount of “sleep recovery” necessary after sleep deprivation to recover normal cognition is not well understood or agreed upon in the literature, and would have been both individualized and difficult to control for in a countered sample [37, 38]. Additionally, in efforts to reduce order-effects, participants were trained in tasks until they were familiarized before scans were performed. It is also possible that scan-time affected the results reported; however, we accounted for the scan-time as best as possible by limiting the first scan session between the daylight hours of 10AM to 2PM and scheduling the second scan session 28 h after normal wake-times for each participant.

Further, while significant differences were discovered between the rested and sleep-deprived group during Flanker task in Channel 5, little could be extracted from other tasks. It is possible that the N-back task (0-back and 2-back) and Flanker task were not complicated enough to extract significant differences in hemodynamic response between conditions.

In summary, fNIRS is an inexpensive and portable imaging method that can be used to provide insight into how sleep deprivation affects cognitive function and hemodynamics, and that it may be reliably used to show functional connectivity differences inter- and intra-individually in well-rested and sleep deprivation conditions. Our results also provide insight into how acute sleep deprivation may cause a decline in interference resolution ability, with potential compensation in the right dorsolateral prefrontal cortex. Future studies with larger samples and more difficult tasks are required to further elucidate the neurobiological mechanisms involved in sleep deprivation.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Reasonable requests regarding the data can be made to the corresponding author in writing. All the codes used in this manuscript will also be made available upon written request to the authors.