Introduction

10% of school-age children and 2%−5% of adults are affected by attention-deficit/hyperactivity disorder (ADHD) worldwide1,2,3. ADHD is a condition characterized by deficits in cognitive domains related to attention, working memory, and inhibitory control4. Children and adults with ADHD experience substantial sleep problems5,6,7. Large studies have found that 40% of adults with ADHD reported insomnia symptoms and short sleep duration8. It is estimated that nearly twice as many adults with ADHD met the cutoff criteria for insomnia as their healthy counterparts9. Adults with ADHD show poorer sleep patterns than the general population, particularly in measures such as sleep onset latency (the time it takes to fall asleep) and sleep efficiency (the percentage of time in bed actually spent asleep)10.

Shortened or restricted sleep is known to impair short-term executive function and sustained attention in healthy individuals, negatively affect long-term memory, and lead to cognitive deficits11,12. Sleep disruption in individuals with ADHD has been shown to lead to even more pronounced deficits in these areas13,14,15,16,17,18,19. In children with ADHD, the effects of sleep restriction appear predominantly in the domains of attention13. For instance, moderate sleep restriction slowed reaction times, increased the number of failures to respond to target stimuli (i.e., omission errors) on the continuous performance test (CPT)14, and impaired inhibitory control (the ability to stop a pre-planned motor response), as assessed by a Go/No-go task15. By contrast, adolescents with ADHD showed no impairment in CPT performance following sleep restriction16. However, parental reports from that study indicated an increase in inattentive symptoms following sleep restriction.

Recently, the negative effects of sleep disruption on the cognitive abilities of young adults with ADHD have also been demonstrated. A series of studies17,18,19 examined the impact of sleep deprivation (25 h of sustained wakefulness) on the behavioral and cognitive performance in young adults with and without ADHD. In both groups, sleep deprivation impaired performance on all CPT indices, resulting in increased omission and commission errors and slower reaction times17. Participants were also tested using a visual emotional oddball task requiring them to respond to targets (an angry face or a geometric shape with a cross) while withholding responses to non-targets (a neutral face or a shape without a cross). At baseline, both groups performed worse on facial stimuli than on non-facial stimuli across all task indices. Notably, sleep deprivation did not significantly alter the performance of the control group; in contrast, individuals with ADHD demonstrated a marked decline, characterized by increased errors and greater reaction-time variability18. This pattern is significant, as individuals with ADHD already experience difficulties in recognizing emotional facial expressions20,21, and these difficulties intensify following sleep deprivation19. Such impairments are especially consequential, as facial emotion recognition is a fundamental non-verbal channel for conveying emotional information22,23, and deficits in this domain can substantially hinder social functioning in individuals with ADHD24,25.

Taken together, the behavioral findings suggest that sleep deprivation disrupts the cognitive processes required to attend to and process emotional facial expressions in individuals with ADHD. However, the extent to which these deficits are specific to emotional facial stimuli remains unclear. Furthermore, the specific cognitive and neurophysiological processes associated with these impairments are not yet fully understood. Insight into these processes can be gained through the measurement of event-related potentials (ERPs), which capture neural responses to sensory or cognitive events with millisecond precision by non-invasively monitoring electroencephalographic waves of response26,27,28. ERPs show the evolution of a cognitive response from the pre-stimulus baseline to the stages following a behavioral response. The resulting voltage components, characterized by the timeframe after stimulus onset, the scalp location of the response, and its magnitude, are linked to specific stages of cognitive processing26,27,28.

Accordingly, the present study aimed to clarify the effects of sleep deprivation on the ability of young adults with and without ADHD to respond accurately and efficiently to facial expressions. Furthermore, we sought to characterize the neural correlates of these effects using ERPs recorded during task performance. To this end, we employed a visual-emotional oddball paradigm29 in which participants were presented with target stimuli requiring rapid responses (an angry face or a cross-bearing geometric shape) and non-target stimuli (a neutral face or a shape without a cross).

We focused on four early ERP components closely linked to early visual attention and the processing of facial and emotional information: the P100, N170, P200, and P300. The P100 (or P1) is a positive-going component reflecting an initial perceptual response that is particularly sensitive to unpleasant or negative emotional content30,31,32. Although influenced by facial emotions, P100 does not appear to encode the emotional significance or social meaning of stimuli33,34,35,36. The N170 is a negative-going component29 that responds specifically to facial stimuli37,38,39. The P200 (or P2) is associated with selective attention and the perception of arousing stimuli40; it indexes early attentional processes involved in evaluating image features and prioritizing affectively salient stimuli for further processing41,42. Finally, the P300 (or P3) is a positive-going ERP primarily linked to attention and working memory43. It typically reflects the detection and cognitive evaluation of task-relevant stimuli44, with its amplitude representing the allocation of attentional resources - higher amplitudes indicating greater attentional engagement45.

We targeted these components because previous research has shown that sleep deprivation affects early ERP responses in healthy individuals performing both emotional and neutral (shape-based) Go/No-go tasks46,47. In one of our previous studies, we observed differences in early ERP components between young adults with ADHD and healthy controls in response to emotional facial expressions, but not to a neutral expression, in participants who were not sleep-deprived29. Therefore, we expected that examining the early ERP components within approximately 300 ms of stimulus onset would provide a better understanding of how sleep deprivation influences the processing of emotional and neutral cues in individuals with ADHD.

We hypothesized that sleep deprivation would produce two primary effects. First, in the healthy control group, we expected sleep deprivation to elicit more ADHD-like responses. This expectation was based on findings by Magnuson and colleagues47, who reported that sleep deprivation impaired task performance and reduced neural efficiency in response to an emotional cues more than to neutral (shape) cues, an effect similar to what we previously observed in individuals with ADHD who were not sleep-deprived29. Second, we expected sleep deprivation to have a stronger impact on individuals with ADHD, particularly when processing emotional cues, with comparatively smaller effects in response to neutral stimuli (neutral expressions or shapes). Accordingly, we predicted that:

Behavioral hypothesis

Sleep deprivation would impair behavioral performance and attentional stability, with the ADHD group showing increased commission and omission errors, slower reaction times, and greater reaction time variability (RTSD), particularly for emotional stimuli (target faces). In contrast, the control group was expected to exhibit milder or no changes across these measures.

Neural hypotheses

  1. a.

    After sleep deprivation, early ERP components in the control group would become more similar to those of the ADHD group at baseline (rested) for both emotional and neutral cues.

  2. b.

    In the ADHD group, the effect of sleep deprivation would be more pronounced in the ERP components for the emotional cue (angry facial expression) than for the neutral cues(neutral face or shapes).

Method

Participants

The participants included 19 male adults with ADHD (Mean age = 24.3; SD = 5.3) and 14 male adults without ADHD (Mean age = 26.2; SD = 3.1). The participants of both groups were either college students, recruited through ads on campus, or from the general population, recruited by snowball sampling. To reduce variability, only men were included in the study because the menstrual cycle may affect various psychological functions, including sleep48. The general characteristics of the participants are described in Table 1.

Participants in the ADHD study group met the following inclusion criteria: (a) display a minimum of six symptoms on both the inattention and hyperactivity-impulsivity scales, as assessed by the Rating Scale-IV49; (b) present a prior clinical diagnosis of adult ADHD, either confirmed by a neurologist or a psychiatrist affiliated with an established clinic specializing in psychoeducational assessment; (c) satisfy the diagnostic criteria for ADHD, as delineated in the modified young adult iteration of the ADHD module of the Diagnostic Interview Schedule for Children (DISC)50.

To be included in the control group, individuals met the following criteria: (a) manifest fewer than four symptoms pertaining to inattention or hyperactivity-impulsivity on the ADHD Rating Scale-IV; (b) affirm the absence of any prior diagnosis of ADHD; (c) Not meet the diagnostic criteria for ADHD as stipulated in the modified young adult version of the ADHD module of the Diagnostic Interview Schedule for Children (DISC).

Exclusion criteria for both study groups included: (a) the presence of any psychopathological condition, as determined by evaluation with the Symptom Checklist-90 under the supervision of a clinical psychologist; (b) diagnosis of obstructive sleep apnea, periodic limb movements during sleep, or restless legs syndrome, as assessed through interviews and the Mini Sleep Questionnaire51; (c) in employment involving night shifts; (d) the use of centrally acting agents, excluding those prescribed for the treatment of ADHD.

In the ADHD group, eight participants were regular users of stimulant ADHD medications and the others used such medications sporadically. All participants in this subgroup willingly refrained from using these medications during the entire day of the experiment and in the 24 h preceding it. This practice aligns with the established “washout” period previously used for such medications in similar studies52.

Eight of the participants were habitual cigarette smokers, predominantly light to moderate smokers, averaging ten cigarettes per day or fewer. All study participants had either normal vision or vision that had been properly corrected.

The study conformed to the principles of the Declaration of Helsinki. The Max Stern Yezreel Valley College Ethics Review Board approved the study protocol (approval number: EMEK YVC 2016-40). After participants received a detailed description of the study, they provided written informed consent. They received $125 compensation for participating in the study.

Sample size considerations

We based our power analysis on an effect size of the interaction between group and stimulus on omission errors from a previous study examining differences between participants with and without ADHD, using the same emotional visual oddball task as in the current study29. The effect size was η2 = 0.10, α = 0.05, and 80% power to detect an effect. The minimal sample size required was N = 32.

Measures

ADHD questionnaire. The questionnaire comprises 18 items, aligning with the symptoms listed in the DSM-IV for diagnosing ADHD, including 9 items appraising attentiveness and 9 items assessing hyperactivity and impulsivity. In the version adopted for this study, participants were tasked with indicating whether each described situation did or did not apply to them. The questionnaire has previously served as a screening tool for ADHD in various studies (e.g29,53.,.

The Symptom Checklist-90 (SCL-90)54 was used to assess the frequency of distress symptoms. Patients were prompted to rate the extent to which they had experienced each of the symptoms covered by 90 items over the preceding 7 days on a 5-point Likert scale. The checklist serves as an international standard and has been translated into many languages.

Structured clinical interview is a modified version of the ADHD module from the DISC-IV50. The interview generates clinician-assessed symptom counts for inattentive and hyperactive-impulsive ADHD symptoms. The interviews were conducted by graduate students in psychology who received comprehensive training on the administration and scoring of the DISC-IV. The scoring process was overseen by a senior clinical psychologist. The presence or absence of ADHD was determined based on the DISC-IV scoring algorithm (for details, see Bart et al., 201455.

Actigraph sleep recording. The actigraph (Mini Motionlogger, Ambulatory Monitoring Inc., New York) is a wrist-worn ambulatory device that measures wrist movements using a piezoelectric element and translates them into 1-minute long epochs of sleep and wake. Wrist activity levels were sampled at 10-second intervals and summed across 1-minute intervals. Actigraphic raw data were converted to sleep measures using the Actigraphic Scoring Analysis program (W2 scoring algorithm) for a personal computer provided by the manufacturer.

The actigraph provides four sleep measures: sleep onset latency, sleep efficiency (percentage of total sleep time between sleep onset and final awakening), wake time after sleep onset, and total sleep time. Daily actigraphy data of each participant were averaged over the five days of actigraph use to obtain aggregated measures. The participants were instructed to press a button on the actigraph when they began trying to fall asleep and when they woke up the following morning. The first button press was used to determine bedtime and the second wake time. For the purpose of precise analysis of the actigraph data, participants completed sleep diaries to document their bedtime times and wake-up times over the course of actigraphic recording.

The Pittsburgh Sleep Quality Index (PSQI) is an established instrument for assessing sleep difficulties56. It consists of 18 items, generating scores that reflect various facets of sleep quality. The scores are totaled to yield a global PSQI score. Buysse et al.56 demonstrated that a global PSQI score exceeding 5 produces a diagnostic sensitivity of 89.6% and specificity of 86.5% in distinguishing between good and poor sleepers. Therefore, a PSQI score of 5 can be considered the cut-off for clinically relevant sleep impairments. In the present study, the internal consistency (Cronbach’s α) of the PSQI was 0.66.

Visual Oddball Task. Stimuli for the task included photographs of the faces of three male individuals and three geometric shapes (circle, triangle, and square). All faces were taken from a standard set of pictures of facial affect, the NimStim face stimulus set57. The facial expressions were either neutral (non-target) or angry (target). Each actor presented both emotional (angry) and neutral expressions. Geometric shapes were “empty” (non-target) or contained a black cross in the middle of the shape (target). A four-stimulus visual oddball paradigm was used, and participants were exposed to two regularly repeated standard stimuli (neutral faces and empty shapes), presented with a probability of 0.75, and two “target” deviant stimuli (angry faces and shapes with a cross), presented with a probability of 0.25. Each target stimulus was presented 10 times and each non-target stimulus 30 times, totaling 240 stimuli (90 neutral faces, 90 empty shapes, 30 angry faces, and 30 shapes with a cross). Stimulus presentation order was randomized between trials. All trials consisted of a 500-ms stimulus display followed by a 1,000-ms blank screen intertrial interval. Before the experimental block, participants completed a brief practice block (32 trials) featuring stimuli identical to those in the experimental phase.

Procedure

For the five days leading up to the experimental trial, participants adhered to a set regimen: bedtime between 23:00–24:00, wake-up time between 7:00–9:00, a minimum of 7 h of nightly sleep, refraining from napping, and consumption of no more than three caffeinated beverages per day. Throughout this period, participants wore actigraphs to monitor their sleep patterns and adherence to the prescribed schedule.

The experimental session started at 8 am on day 6. The participants (4–8 participants in each session) completed the PSQI and performed the visual oddball task while EEG/ERP recordings were conducted (9:00–10:30 AM). Subsequently, they remained awake in the laboratory for the next 25 h before repeating the visual oddball task. During the visual oddball sessions, participants sat 80 cm from a 19-in computer screen and were instructed to fixate their gaze on the stimuli to be presented at the center of the screen and to point out as quickly as possible (without compromising accuracy) the occurrence of a “target” (deviant) stimulus by pressing the spacebar on the keyboard with their right index finger and withholding response to the “non-target” (standard) stimuli. Response time and error rate were recorded, distinguishing between omission errors (failure to respond to a deviant stimulus) and commission errors (responding to a standard stimulus). Responses could be made during stimulus presentation as well as during inter-trial intervals. Participants were instructed to minimize eye movements during the task and were assured that they could complete it without the need for eye movements.

Throughout the session, participants wore wrist-worn actigraphs to ensure they did not fall asleep. The laboratory maintained controlled conditions, with constant artificial lighting (~ 500 lx), an undetectable amount of sunlight, and a temperature set at 25 °C. Participants abstained from consuming caffeine, coco-containing products, or tobacco. They were served organized and supervised meals at scheduled time points. Strenuous physical activity was prohibited throughout the session. Participants engaged in activities such as board games and social interactions during periods of rest. Participants of both study groups were similarly involved in each of these activities. Members of the experiment staff were present throughout the experiment to ensure participant wakefulness.

EEG/ERP recording: data acquisition

The electroencephalogram (EEG) data were acquired using specialized equipment, including a 128-channel HydroCel Geodesic Sensor Net, Net Amps 400 amplifier, and Net Station, Version 5.2 software (Electrical Geodesics Inc., Eugene, OR, USA). The EEG signals were recorded at a sampling rate of 1000 Hz with a 0.1 Hz high-pass filter. Throughout the recording, all channels were referenced to Cz. In the subsequent offline processing stage, the continuous EEG data were referenced to an average reference, and a bandpass filter was applied with a frequency range of 0.1–30 Hz. The data were then segmented into 900 ms epochs locked to the stimulus, spanning from 100 ms before the stimulus onset to 800 ms after. To ensure data quality, epochs containing artifacts such as vertical eye movements (eye blinks; ±140 µV) and horizontal eye movements (± 55 µV) were automatically detected using a computerized algorithm and further confirmed through visual inspection, and these epochs were removed. Additionally, segments of recording were flagged as “bad” if they exhibited more than 10 bad channels, with “bad channel” defined as having voltage amplitudes exceeding ± 200 µV for the entire segment. Individual bad channels were corrected on a segment-by-segment basis using spherical spline interpolation. Subsequently, the averaged ERPs were baseline-corrected. All aspects of stimulus presentation and collection of behavioral responses were controlled by a computer running E-prime professional 2.0 software (Psychology SoftwareTools Inc., Sharpsburg, PA, USA).

Target-evoked ERP components

Following inspection of the grand average ERP waveforms and the topographic maps of the voltage-distribution, and drawing on previous studies employing oddball (target-nontarget) tasks with emotional-faces stimuli29,58,59, we decided to quantify the peak amplitudes of the following ERP components within specified latency windows (centered on the component’s peak): P1 (70–130 ms post-stimulus onset), N170 (140–200 ms post-stimulus onset), P2 (200–260 ms post-stimulus onset), and P3 (320–500 ms post-stimulus onset). Based on the described inspection of the grand average ERP waveforms and the topographic maps, and drawing on previous studies employing oddball tasks with emotional-faces stimuli29,58,59, mean amplitudes of these ERP components were quantified at the frontal scalp location (average of channels 4, 10, 11, 16, 18, and 19) and at the posterior-parietal scalp location (average of channels 59, 66, 71, 76, 84, and 91). For the electrode array, see Fig. 1.

Data analysis

Analyses were conducted using Jamovi (v. 2.7.6). The normality assumption was examined by examining the Q-Q plots and Kolmogorov-Smirnov tests. Differences between the groups in the pre-experimental sleep measures and demographic variables were examined using t-tests or chi-square tests. For the behavioral and ERP data analysis, when the normality assumption was not violated, we used linear mixed models. The linear mixed models used the Satterthwaite approximation for degrees of freedom, which resulted in non-integer degrees of freedom for the F-tests, making inference more accurate for small, unbalanced samples. When the normality assumption was violated, we used generalized mixed models. For all statistical tests, the significance threshold was set at 0.05.

Behavioral data

Behavioral measures included omission errors, commission errors, reaction time (RT), and variability of the reaction time (RTSD). To examine the effects of sleep deprivation on participants with and without ADHD, 2 × 2 general or generalized mixed models (see below for specific models for each of the 4 measures) were conducted separately for the responses to target/non-target faces and the responses to target/non-target shapes, with the between-subjects variable being group (ADHD/control) and the within-subject variable being time (baseline/0 + 25). As count data that did not normally distributed, commission and omission errors were analyzed by negative binominal regression model. The omission error in response to target shapes was analyzed only at time 2 (after sleep deprivation) as only one participant had such errors at baseline. RT was positively skewed and were thus analyzed by gamma regression. As it was normally distributed, RTSD data was analyzed by linear mixed model.

ERP data

All ERP data were normally distributed. For each component (P1, N170, P2, P3), a 2 × 2 × 2 linear mixed model analysis was conducted, with the between-subjects variable being group (ADHD/control) and the within-subject variables being time (baseline/0 + 25) and condition (target/non-target). In line with the behavioral data analyses, these ERP analyses were conducted separately for faces stimuli and for shapes stimuli. Guided by our theoretical framework, after a significant 3-way interaction, we tested for the presence of two-way interactions (group × time) separately for the ERPs to target/non-target stimuli.

Fig. 1
Fig. 1
Full size image

The ERP configuration used in our experiment: electrode locations.

Results

Demographic data and sleep measures

Demographic comparisons between young adults with and without ADHD are presented in Table 1. The groups did not differ in age, caffeine and alcohol consumption, proportion of cigarette smokers, or objectively measured sleep patterns. The average global PSQI score (a subjective measure of sleep quality) of the ADHD group was above 5, which is the cut-off for clinically relevant sleep difficulties. The groups did not differ significantly in their PSQI scores or the percentage of participants receiving a global PSQI score above 5.

Table 1 Demographic characteristics and subjective measures of the participants’ sleep before the experimental session.
Full size table

SD Standard deviation, ADHD-RS ADHD Rating Scale, Alcohol number of alcoholic beverages per week, Caffeine number of caffeinated beverages per day, WASO wake after sleep onset, PSQI Pittsburgh Sleep Quality Index. Differences between the groups in the demographic variables and the pre-experimental sleep measures were examined using two-tailed independent samples t-tests or chi-squared tests. As multiple t-test comparisons were performed, significance (*p < 0.05) was determined following the Bonferroni correction.

Behavioral oddball data

Descriptive statistics of the omission errors, RTs, and RTSDs in response to face and shape targets and commission errors to face and shape non-targets in the ADHD and control groups before and after sleep deprivation are presented in Table 2.

Table 2 Measures of the oddball task: descriptive statistics.
Full size table

Performance of the ADHD and control group participants in response to target faces/shapes (omission errors, reaction time [RT], and reaction time variability [RT standard deviation, RTSD]) and non-target faces/shapes (commission errors) before (time 0, baseline) and after (0 + 25) sleep deprivation.

Commission errors

For non-target faces, the analysis revealed no significant main effect for time (X² (1) = 1.54, p = 0.132) or group (X²(1) = 2.27, p = 0.215) but there was a significant time × group interaction (X²(1) = 4.81, p = 0.028). To analyze this interaction, we conducted a simple effects analysis, demonstrating a significant increase in commission errors following sleep deprivation in the ADHD group (Z (1) = 2.92, p = 0.003), but not in the control group (Z = 0.55, p = 0.586). The rate of commission errors at baseline was similar in both groups (Z = 0.29, p = 0.774) but significantly higher in the ADHD group following sleep deprivation (Z = 2.38, p = 0.017).

For non-target shapes, the analysis revealed no significant main effect for group (X²(1) = 0.13, p = 0.077) and no time × group interaction (X²(1) = 0.002, p = 0.963) but there was a main effect for time (X²(1) = 11.43, p < 0.001), with more commission errors following sleep deprivation.

Omission errors

For target faces, the analysis revealed no significant main effect for time (X²(1) = 3.26, p = 0.071) or group (X²(1) = 1.37, p = 0.243) but there was a significant time × group interaction (X²(1) = 4.59, p = 0.032). To analyze this interaction, we conducted a simple effects analysis, demonstrating a significant increase in omission errors following sleep deprivation in the ADHD group (Z = 3.89, p < 0.001) but not in the control group (Z = 1.96, p = 0.884). The rate of omission errors at baseline was similar in both groups (Z = 0.07, p = 0.941) but significantly higher in the ADHD group following sleep deprivation (Z = 2.04, p = 0.041).

Only one participant had omission errors in response to target shapes at baseline. Therefore, study groups were compared only in relation to the rate of post-deprivation omission errors in response to the target shapes, using a negative binomial regression analysis. This analysis revealed no significant difference between the groups (X²(1) = 0.37, p = 0.543).

Reaction time (RT)

For target faces, the analysis revealed no significant main effect for group (X²(1) = 0.05, p = 0.826) and no time × group interaction (X²(1) = 0.13, p = 0.723) but there was a main effect for time (X²(1) = 7.42, p < 0.01), with an increase in RT following sleep deprivation. For target shapes, the analysis revealed no significant main effect for group (X²(1 = 0.30, p = 0.581) or time (X²(1) = 1.93, p = 0.164) and no time × group interaction (X²(1 = 0.05, p = 0.945).

Reaction time variability (RTSD)

For target faces, the analysis revealed no significant main effect for group (X²(1) = 1.88, p = 0.171) but a significant effect for time (X²(1) = 8.41, p = 0.004), with higher RTSD following sleep deprivation. There was also a significant time × group interaction (X²(1) = 4.37, p = 0.037). Simple effects analysis demonstrated a significant increase in RTSD following sleep deprivation in the ADHD group (Z = 4.08, p < 0.001) but not in the control group (Z = 0.51, p = 0.608). RTSD at baseline was similar in both groups (Z = 0.38, p = 0.702) but significantly higher in the ADHD group following sleep deprivation (Z = 2.09, p = 0.036).

For target shapes, the analysis revealed no significant main effect for group (X²(1) = 1.71, p = 0.191) but a significant effect for time (X²(1) = 8.71, p = 0.003), with higher RTSD following sleep deprivation. There was a significant time × group interaction (X²(1) = 6.05, p = 0.014). Simple effects analysis demonstrated a significant increase in RTSD following sleep deprivation in the ADHD group (Z = 3.37, p < 0.001) but not in the control group (Z = 0.42, p = 0.675). RTSD at baseline was similar in both groups (Z = 0.07, p = 0.944) but significantly higher in the ADHD group following sleep deprivation (Z = 2.17, p = 0.030).

Electrophysiological and neural (ERP) results

Figure 2 shows the grand averaged ERPs for facial targets and non-targets before and after sleep deprivation for the ADHD and control groups. As almost no significant effect emerged in the statistical analysis of the shape stimuli, at either the frontal or posterior-parietal scalp sites, these results are reported in the supplementary material.

Fig. 2
Fig. 2
Full size image

Grand averaged ERPs for facial targets and non-targets before sleep deprivation (black line) and after sleep deprivation (red line) by the ADHD and the control groups.

P1 (70–130 Ms post-stimulus)

Frontal cluster

A significant time (before/after sleep deprivation) × group (ADHD/control) × condition (target/non-target) interaction was found for P1 (F(1,64.58) = 13.54, p < 0.001; see Fig. 3), with no significant main effects of time, group, or condition. Follow-up analyses of the significant three-way interaction indicated that the time × group interaction was significant for angry faces (F(1,69.33) = 12.36, p < 0.001), but not for neutral faces (F(1,65.74) = 0.63, p = 0.429).

At baseline, the control group showed a more pronounced (more negative) P1 than the ADHD group (t(55.72)=−2.55, p = 0.013). However, no significant difference was observed after sleep deprivation (t(66.88) = 1.90, p = 0.061). Within-group analyses revealed that in the ADHD group, P1 became more pronounced (more negative) following sleep deprivation (t(67.48)=−2.29, p = 0.024), whereas in the control group, it became less pronounced (less negative) (t(70.65) = 2.66, p = 0.009).

Posterior-parietal cluster

A significant time × group × condition interaction was found for P1 (F(1,64.78) = 5.02, p = 0.028), with no significant main effects of time, group, or condition. Follow-up analyses of the significant three-way interaction indicated that the time × group interaction was significant for angry faces (F(1,57.81) = 6.70, p = 0.012), but not for neutral faces (F(1,56.32) = 0.09, p = 0.761).

At baseline, the control group showed a more pronounced (more positive) P1 than the ADHD group (t(43.36) = 3.06, p = 0.003). However, no significant difference was observed after sleep deprivation (t(44.71) = 0.43, p = 0.666). Within-group analyses revealed that in the control group, P1 became less pronounced (less positive) following sleep deprivation (t(56.32)=−2.39, p = 0.019), whereas in the ADHD group, it did not change significantly (t(59.78) = 1.19, p = 0.236).

Fig. 3
Fig. 3
Full size image

P1 (70–130 ms post-stimulus) for the ADHD and control groups before (time 0, baseline) and after (0 + 25) sleep deprivation. (A) P1for emotional faces (anger) at frontal electrodes. (B) P1 for neutral faces at frontal electrodes. (C) P1 for emotional faces (anger) at posterior-parietal electrodes. (D) P1 for neutral faces at posterior-parietal electrodes. (E) Topographic maps of the voltage distribution of P1 for images of angry facial expressions. Red represents positive values and blue negative ones (µv). *p < 0.05.

N170 (140–200 Ms post-stimulus)

Frontal cluster

A significant time × group × condition interaction was found for N170 (F(1,65.23) = 10.80, p = 0.001), along with a main effect of condition (F(1,65.23) = 5.08, p = 0.027; see Fig. 4) but no main effects of group or time. Follow-up analyses of the significant three-way interaction indicated that the time × group interaction was significant for angry faces (F(1,56.52) = 6.69, p = 0.012), but not for neutral faces (F(1,54.58) = 0.44, p = 0.508).

At baseline, no significant group difference was observed in N170 amplitude (t(49.94)=−1.06, p = 0.293). However, after sleep deprivation, the control group showed a more pronounced (more positive) N170 than the ADHD group (t(48.73) = 2.04, p = 0.046). Within-group analyses revealed that in the control group N170 became more pronounced (more positive) following sleep deprivation (t(54.63) = 2.62, p = 0.011), whereas in the ADHD group N170 did not change significantly after sleep deprivation (t(56.23)=−0.69, p = 0.489).

Posterior-parietal cluster

A significant time × group × condition interaction was found for N170 (F(1,64.72) = 7.23, p = 0.009), along with a main effect of condition (F(1,64.72) = 6.27, p = 0.014) but no main effects of group or time. Follow-up analyses of the significant three-way interaction indicated that the time × group interaction was significant for angry faces (F(1,46.66) = 5.17, p = 0.027), but not for neutral faces (F(1,45.31) = 0.02, p = 0.888).

At baseline, no significant group difference was observed in N170 amplitude (t(41.61) = 0.75, p = 0.454). However, after sleep deprivation, the control group showed a more pronounced (more negative) N170 than the ADHD group (t(44.22)=−2.07, p = 0.043). Within-group analyses revealed that in the control group, N170 became more pronounced (more negative) following sleep deprivation (t(46.83)=−3.37, p = 0.001), whereas in the ADHD group, N170 did not change significantly (t(46.41)=−0.44, p = 0.660).

P2 (200–260 Ms post-stimulus)

Frontal cluster

The time × group × condition interaction did not reach statistical significance (F(1,62.16) = 3.43, p = 0.068). There was a main effect of condition (F(1,62.16) = 19.16, p < 0.001), whereas the main effects of group and time were not significant.

Posterior-parietal cluster

The time × group × condition interaction was not statistically significant (F(1,65.62) = 0.34, p = 0.557), and no significant main effects emerged for condition, group, or time.

Fig. 4
Fig. 4
Full size image

N170 (140–200 ms post-stimulus) for the ADHD and control groups before (time 0, baseline) and after (0 + 25) sleep deprivation. (A) N170 for emotional faces (anger) at frontal electrodes. (B) N170 for neutral faces at frontal electrodes. (C) N170 for emotional faces (anger) at posterior-parietal electrodes. (D) N170 for neutral faces at posterior-parietal electrodes. (E) Topographic maps of the voltage distribution of N170 for images of angry facial expressions. Red represents positive values and blue negative ones (µv). *p < 0.05.

P3 (320–500 Ms post-stimulus)

Frontal cluster

The time × group × condition interaction was not statistically significant (F(1,63.05) = 1.98, p = 0.163), and no significant main effects emerged for condition, group, or time.

Posterior-parietal cluster

The time × group × condition interaction was not statistically significant (F(1,65.87) = 0.16, p = 0.689), and no significant main effects emerged for condition, group, or time.

Discussion

This study examined the effects of sleep deprivation on emotional face processing in young adults with ADHD by integrating behavioral and neural measures. Participants with and without ADHD completed a visual oddball task before and after 25 h of sustained wakefulness. This task was designed to assess sustained attention, response inhibition, and aspects of executive function at the behavioral level. ERP recordings provided insights into the temporal dynamics of cognitive processing from early visual attention to facial recognition, through analysis of the P1, N170, P2, and P3 components.

Previous research has shown that sleep deprivation impairs cognitive and attentional performance in individuals with and without ADHD60. Our prior study demonstrated that when well rested, young adults with ADHD performed comparably to neurotypical controls on a visual oddball task; however, following sleep deprivation, performance declined in the ADHD group while the control group remained unaffected17. The present study replicated these findings: at baseline, no significant behavioral differences were observed between groups, but after sleep deprivation, performance was modulated based on both group membership and stimulus type (faces vs. shapes).

For facial stimuli, omission errors, commission errors, and reaction time variability (RTSD) increased only in the ADHD group, whereas reaction time (RT) increased in both groups. For geometric shapes, RTSD increased only in the ADHD group but omission and commission errors increased similarly in both groups, and RT remained unchanged. These findings suggest that, when well rested, young adults with ADHD perform comparably to their neurotypical peers. Sleep deprivation, however, impairs cognitive processing in both groups, with the ADHD group showing heightened vulnerability specifically when processing emotional facial expressions.

Neural findings

The primary novelty of this study lies in its examination of the neural processes associated with the observed. We analyzed early ERP components (P1, N170, P2, P3) due to their well-established associations with ADHD-related attentional and perceptual deficits61, as well as their sensitivity to the effects of sleep deprivation on emotional face processing47,62.

The P1 component indexes early visual processing and is particularly responsive to negative emotional stimuli, potentially reflecting rapid amygdala reactivity30,31,32. At baseline, no significant group differences were observed in P1 amplitude in response to neutral stimuli, and sleep deprivation did not significantly affect the processing of neutral stimuli in either group. In contrast, for angry facial expressions, baseline P1 amplitude was significantly more preannounced in the control group. Following sleep deprivation, the control group exhibited a marked reduction in P1 amplitude, consistent with prior evidence indicating that sleep deprivation attenuates P1 amplitude in response to emotional expressions in healthy individuals62. In the ADHD group, P1 amplitude increased following sleep deprivation; however, this effect reached statistical significance only over frontal scalp sites and did not emerge over posterior-parietal regions. This dissociation suggests that sleep deprivation may differentially modulate early visual–emotional processing in ADHD, with region-specific alterations that diverge from those observed in neurotypical controls.

The N170 component is specifically sensitive to facial stimuli37,38,39 and can reliably distinguish between neutral and emotional facial expressions37,38,39. At baseline, N170 amplitudes did not differ significantly between the ADHD and control groups for any stimulus category, and this pattern remained unchanged for neutral faces following sleep deprivation. In contrast, for angry facial expressions, group differences emerged after sleep deprivation: the control group showed a significant increase in N170 amplitude, whereas the ADHD group displayed no meaningful change. These findings align with previous evidence indicating that sleep deprivation enhances N170 responses to emotional faces in healthy individuals62.

Our ERP findings suggest that sleep deprivation likely depletes neurocognitive resources in both groups when processing emotional stimuli, but the effects differ. In the control group, sleep deprivation was associated with a reduced P1 response and an enhanced N170 response, a pattern consistent with compensatory recruitment of additional neural resources to maintain performance. In contrast, among participants with ADHD, P1 amplitude either increased (in frontal regions) or remained unchanged (in posterior–parietal regions), and N170 amplitude did not change, indicating a lack of comparable compensatory modulation. These patterns correspond closely to the behavioral outcomes observed in the emotional oddball task: performance impairments emerged only in the ADHD group following sleep deprivation. One interpretation is that, despite the depletion of neurocognitive resources, the control group was able to engage compensatory mechanisms that supported intact emotional processing. Conversely, in the ADHD group, the increased frontal P1 activity may reflect an unsuccessful attempt to compensate for a greater depletion of neurocognitive resources.

This interpretation is supported by neuroimaging research showing that sleep deprivation alters activation in frontal regions, including the prefrontal cortex, as well as in posterior cortical areas during cognitive tasks63,64,65,66, with the magnitude of these changes often correlating with subjective sleepiness66. Consistent with this literature, our previous work demonstrated that young adults with ADHD reported significantly higher levels of sleepiness after 25 h of sleep deprivation compared with controls67.

Clinical implications

These findings may have important clinical implications. Emotional facial expressions play a central role in social interaction by enabling accurate perception and interpretation of others’ emotional states25,68,69. Individuals with ADHD have been shown to experience difficulties in emotion recognition, particularly for negative emotions such as anger and fear20,70,71,72, and these impairments likely contribute to the social and emotional challenges frequently reported in this population73,74. Given that ADHD is also associated with sleep disturbances, including insomnia and altered sleep duration8,75, our findings suggest that sleep problems may further exacerbate difficulties in emotional face processing, especially for angry facial expressions. Consequently, addressing sleep-related difficulties may represent an important avenue for improving emotion-processing abilities and, more broadly, social functioning in individuals with ADHD.

Study limitations and future directions

Several limitations should be considered. First, our ERP measurements were limited to the frontal cortex and the posterior-parietal cortex, and to a single emotional facial expression (anger). Prior studies suggest that both ADHD and sleep deprivation may differentially affect the processing of different emotions76,77. Future studies should therefor include a broader range of emotional expressions as well as additional cortical regions. Second, the 25-hour sleep deprivation protocol used in this study represents an extreme condition that does not reflect typical sleep patterns. Chronic insufficient sleep is a more common problem for young adults78, and future research should examine whether similar effects emerge under real-world sleep restrictions. Third, the sample included only young men in order to reduce variability and maintain statistical power. Future studies should assess whether these findings generalize to women and other age groups. Fourth, although potential confounding effects related to repeated task exposure were considered, previous research indicates that oddball paradigms demonstrate high test–retest reliability and are not susceptible to substantial practice effects21,55,79. Finally, participants with ADHD discontinued stimulant medication 24 h before the experiment. Although prior studies suggest minimal withdrawal effects within this timeframe80,81, additional is needed to clarify the potential influence of medication withdrawal on sleep and cognitive performance.

Conclusion

This study provides new evidence that sleep deprivation in ADHD disrupts neural processes associated with emotional face processing, which may lead to behavioral impairments. These findings reinforce underscore the critical role of sleep in managing ADHD symptoms and suggest that interventions targeting sleep quantity and quality may help improve both emotional processing and social functioning in individuals with ADHD.