Age-Related Differences in EEG Correlates of Performance on Cognitive Tasks Varying in Complexity: From the Stroop Task to Chemical Compound Classification

Abstract. Background and Problem. Traditional assessments of cognitive abilities have focused on performance metrics of speed and accuracy. The concept of neural efficiency—the hypothesis that more capable individuals process information with greater metabolic economy—has been influential but has yielded mixed findings. While some studies support the classical view that higher ability is associated with reduced cortical activation, others report increased activation in experts during complex tasks, a phenomenon termed the “neural efficiency paradox.” Furthermore, the role of developmental stage and its interaction with task complexity in shaping neural efficiency remains poorly understood. Objective. This study investigated age-related differences in neural efficiency by comparing EEG spectral power across multiple frequency bands (δ, θ, β1, β2) during the performance of cognitive tasks of varying complexity, while controlling for psychological characteristics of participants. Methods: Three age groups (13–14 years, n = 93; 15–17 years, n = 123; 18–27 years, n = 87) completed a battery of computerized cognitive tasks: chemical compound classification (simple, complex, and most complex levels), the Matching Familiar Figures Test (MFFT), the Elementary Logical Operations (ELO) test, and the Stroop task. EEG was recorded from 30 electrodes, and spectral power was analyzed for each frequency band. Temperament (STQ-s), cognitive style (CPS-Q), and fluid intelligence (Raven’s SPM) were assessed. Group comparisons were performed using Kruskal–Wallis H tests. Results: A dominant pattern emerged across most tasks and frequency bands: the youngest group (13–14 years) exhibited the highest spectral power, while the oldest group (18–27 years) exhibited the lowest, consistent with the classical neural efficiency hypothesis. However, notable exceptions were observed. In the β2 band, associated with cognitive complexity, the oldest group showed increased power during the most challenging tasks (complex chemical classification, Stroop interference, and ELO test), alongside superior behavioral performance. Topographic analysis revealed task-specific engagement of cortical regions: parietal and central sites (spatial thinking) during chemical classification, frontal sites (cognitive control) during the Stroop task, and fronto-central sites (working memory, information integration) during the ELO test. Conclusions. Neural efficiency is a dynamic, context-dependent phenomenon. While the classical pattern of decreasing cortical activation with age and ability holds for simpler tasks, successful performance on complex tasks in older, more cognitively mature individuals is associated with increased high-frequency oscillatory activity, reflecting the recruitment of specialized neural networks. These findings reconcile the classical neural efficiency hypothesis with the “neural efficiency paradox” and underscore the importance of considering task complexity, frequency band, and individual differences in cognitive development.

Introduction

Traditional assessments of cognitive abilities have predominantly relied on performance metrics of speed and accuracy in intelligence and creativity tests. A robust correlation exists between higher ability levels and superior performance on these measures. However, a crucial dimension of cognitive performance—the subjective and objective “easiness” or cognitive fluency with which a task is executed—remains largely unexamined and objectively unmeasured. A promising theoretical framework for addressing this gap is the concept of neural efficiency, first proposed by Haier and colleagues (1988). This hypothesis posits that individuals with higher cognitive abilities process information more efficiently, utilizing fewer neural resources and exhibiting lower brain metabolic activity during task performance.

Subsequent research on neural efficiency, however, has produced a complex and often contradictory picture. While numerous studies have supported the classical hypothesis, demonstrating decreased cortical activation in high-ability individuals—interpreted as evidence of optimized and more selective neural network engagement—others have reported increased activation in specific brain regions, attributed to the recruitment of additional neural circuits or functional reorganization in response to task demands (Li & Smith, 2021). Furthermore, modern data analysis approaches, particularly the application of machine learning algorithms, open new perspectives for studying the neurophysiological foundations of cognitive abilities. They allow not only for the identification of group differences but also for the classification of individuals based on their unique EEG patterns, including resting-state activity, which may act as a predictor for success in subsequent cognitive performance (Steiner et al., 2023).This inconsistency suggests that neural efficiency is not a static, universal principle but rather a dynamic phenomenon modulated by multiple factors, including task type (verbal versus spatial), task complexity, participant sex, and individual differences in cognitive strategies.

However, recent advances in EEG analysis have revealed that the power spectrum consists of both periodic and aperiodic (1/f-like) components, with the latter reflecting the cortical balance of excitation and inhibition (Donoghue et al., 2020). Critically, variations in aperiodic activity are associated with cognitive performance across adulthood, independent of age-related changes in oscillatory power (McKeown et al., 2025). This suggests that the classical ‘neural efficiency’ pattern may be partially driven by shifts in aperiodic activity—a possibility that remains unexplored in developmental studies of task-related EEG.

The paradox of neural efficiency is particularly evident in studies of expert performance. Contrary to the classical model, research in domains requiring rapid, complex actions (e.g., table tennis, soccer) shows that experts often exhibit higher cortical activation compared to novices, especially during challenging tasks (Mann, Wright, & Janelle, 2016). Li and Smith (2021) explain this by suggesting that under conditions of high cognitive demand or performance pressure, the brain prioritizes the optimization of spatial orientation and sustained attention, leading to increased, rather than decreased, neural activity. Thus, the neural efficiency hypothesis appears to hold primarily for simpler tasks, while complex problem-solving in experts may require enhanced activation of specialized neural networks.

Electroencephalography (EEG) offers a valuable, non-invasive tool for investigating these neurocognitive dynamics. Specifically, the analysis of EEG spectral power across fundamental frequency bands (delta, theta, alpha, beta) provides a robust measure of cortical activation that is less susceptible to artifacts than methods like event-related potentials or wavelet coherence analysis. However, a critical limitation of many existing EEG studies is their failure to account for the psychological characteristics of participants—such as cognitive style, personality traits, or developmental level—which significantly hinders the interpretation and generalizability of neurophysiological findings.

Recent advances have clarified how low-frequency oscillations support distinct cognitive operations. Ericson et al. (2025), using MEG during visuospatial working memory tasks, identified four networks in the theta and alpha bands with specific functional roles: a dorsal alpha network linked to maintenance/stability and a posterior theta network linked to encoding/flexibility. The rate of switching between these states followed a U-shaped relationship with cognitive performance, suggesting optimal cognitive control requires balanced transitioning between stability and flexibility. These findings provide a framework for interpreting age-related changes in low-frequency oscillatory power, motivating our investigation of delta, theta, and beta bands across developmental stages.

The present study addresses this gap by integrating neurophysiological measurements with a well-defined psychological framework. It is grounded in the differentiation-integration theory of abilities (Chuprikova, 2007; Volkova, 2011), which conceptualizes abilities as functional properties of representational-cognitive structures. These structures develop from global, undifferentiated states toward differentiated, hierarchically organized forms, and their degree of correspondence with task demands determines performance in terms of speed, accuracy, and critically, the easiness of execution (Volkova, 2011). While a substantial body of literature describes abilities through speed and accuracy (Chuprikova, 2019; Knorr & Neubauer, 1996; Ratanova, 2011; Volkova, 2011), research operationalizing abilities via the criterion of “easiness” is notably absent, and validated instruments for its objective assessment are lacking.

Luo and Zhou (2020) recently examined whether working memory-related EEG biomarkers could measure fluid intelligence in children aged 9–12 years. Although high-ability children showed larger P3 amplitudes, shorter P2 latencies, and lower theta power, the discriminatory power of these markers was modest (AUC < 0.85), and only P3 amplitude predicted academic achievement. Extending this line of inquiry to later adulthood, Jabes et al. (2021) found that older adults (65–75 years) exhibit lower theta and alpha power, higher beta and gamma power, and poorer working memory performance compared to young adults (20–30 years). Together, these findings demonstrate that age-related changes in oscillatory activity occur across the entire lifespan—from childhood through to old age—yet the critical transitional period from early adolescence to young adulthood remains underexplored. This differentiation-integration process is reflected not only in behavioral performance but also in the reorganization of large-scale cortical networks. For instance, Mayer et al. (2025) recently demonstrated that while adolescents and adults engage a similar core fronto-temporal-insular network during action-effect binding, the directed communication within this network is significantly stronger in adolescents. Furthermore, adolescents recruit additional sensory regions, such as the lingual gyrus, suggesting a compensatory reliance on perceptual processing when higher-order cognitive representations are not yet fully mature. This indicates that the transition to adult-like neural efficiency involves not just a reduction in activation, but a qualitative shift in network dynamics and a ‘pruning’ of ancillary sensory recruitment.

Furthermore, while age-related changes in the alpha rhythm are well-documented, their precise characterization has been refined by recent large-scale studies. For instance, Leroy et al. (2025), in a normative study of 532 individuals aged 8 to 92, demonstrated a significant global slowing of the individual alpha peak frequency with age ( Hz/year). This finding underscores the dynamic nature of even fundamental oscillatory activity across the lifespan and highlights the importance of considering not just power, but also peak frequency, when investigating developmental trajectories. The present study addresses this gap by examining neural efficiency across this developmental window.

A fundamental aspect of cognitive flexibility is the ability to switch between task sets, a process central to paradigms like the Stroop task. Recent advances have proposed a dynamical systems account of this process: Ritz et al. (2025) demonstrated that successful task switching relies on convergence toward a ‘neutral’ task state during the inter-trial interval, allowing for rapid reconfiguration upon cue presentation. This aligns with work on cognitive fatigue by Taddeini et al. (2025), who showed that commission errors were associated with fast, automatic responses preceded by reduced pre-stimulus beta power (13–30 Hz) in centroparietal areas, suggesting a failure of proactive response control. Together, these findings provide a quantitative framework linking beta-band dynamics to inhibitory control, motivating our investigation of age-related differences in beta2 power during the Stroop interference task.

Therefore, the aim of this study is to compare EEG power spectra across major frequency bands (delta, theta, alpha, beta1, and beta2) during the performance of cognitive tasks, while systematically controlling for relevant psychological parameters of the participants.

We hypothesize that specific bioelectrical characteristics of cortical activity will be associated with the easiness of cognitive task performance. From the perspective of the classical neural efficiency hypothesis, the most successful participants are expected to exhibit not only higher accuracy and speed but also lower EEG spectral power values and more focal cortical activation. However, in line with the neural efficiency paradox, we anticipate that participants with less developed representational-cognitive structures, or those facing tasks exceeding a certain complexity threshold, may demonstrate increased cortical activation despite successful performance. By examining these patterns across different age groups and cognitive tasks, this study aims to contribute to a more nuanced, psychologically informed understanding of the neural substrates of cognitive abilities and the elusive dimension of “easiness.”

Method

• Participants

A total of 303 volunteers participated in this experimental study. The sample comprised 47% males and included participants from three distinct age groups, residing in Moscow, Ufa, and Grodno, with an overall age range of 13 to 27 years. The breakdown of the age groups was as follows: 93 participants aged 13.73±0.45 years; 123 participants aged 15.46±0.66 years; and 87 participants aged 21.11±2.87 years. All participants reported normal or corrected-to-normal vision and no history of neurological or psychiatric disorders. Written informed consent was obtained from all participants prior to the study. The research protocol was approved by the Local Ethics Committee of the Institute of Psychology of the Russian Academy of Sciences (Protocol No. 7, July 10, 2020).

• Procedure

Data collection took place in a sound-attenuated, electrically shielded room under controlled lighting conditions. Upon arrival, participants were briefed on the study procedure and provided written informed consent. They were then seated comfortably in front of an LCD monitor with a refresh rate of 100 Hz at a viewing distance of approximately 60 cm.

The session began with the application of the EEG cap and the placement of electrodes. Electrode impedance was maintained below 5 kΩ throughout the recording session. A baseline EEG recording was first conducted with eyes open and eyes closed (2 minutes each) to assess resting-state activity and to screen for any epileptiform abnormalities; none were detected in any participant.

Following the baseline recording, participants completed a series of computerized cognitive tasks. The tasks were presented in a randomized order to control for order effects using the InTesting software (Volkova & Nilopets, 2016; State Registration Certificate No. 2016661340). The EEG signal was continuously recorded during task performance. The entire experimental session, including EEG setup, task instructions, and cognitive testing, lasted approximately 60–75 minutes per participant.

• Cognitive Tasks

A computerized battery of four cognitive tasks was administered to assess different aspects of cognitive processing:

Stroop Test (Stroop, 1935): A computerized version of the classic Stroop task was used to assess inhibitory control and attention. Participants were required to name the color of the ink in which color words were presented, with conditions including congruent (e.g., word “RED” in red ink) and incongruent (e.g., word “RED” in blue ink) trials. Reaction times and accuracy were recorded for each trial.

Matching Familiar Figures Test (MFFT) (Kagan, 1966): This task was used to assess the reflection-impulsivity dimension of cognitive style. On each trial, participants were shown a target picture and a set of eight similar variants, only one of which exactly matched the target. Participants were instructed to select the matching variant as quickly and accurately as possible. Latency to first response and total number of errors were recorded.

Elementary Logical Operations Test (ELO; Rusalov & Volkova, 2021; Volkova & Dokuchaev, 2021): This task was designed to assess fluid intelligence and logical reasoning. Participants were presented with a series of logical problems requiring the application of basic operations such as classification, seriation, and analogy. Accuracy and response time were recorded.

Chemical Compound Classification Task (Volkova & Dokuchaev, 2022): A novel task developed to assess domain-specific cognitive abilities in chemistry. Participants were required to classify chemical compounds based on their chemical formulas according to predefined rules (e.g., oxide, acid, base, salt). Accuracy and response time were recorded.

All tasks were implemented in the InTesting software (Volkova & Nilopets, 2015), which ensured precise stimulus presentation and synchronization with the EEG recording system.

• Questionnaires

To assess individual differences in temperament and cognitive style, participants completed the following self-report questionnaires:

Structure of Temperament Questionnaire – Short Form (STQ-s; Rusalov, 2012): This 26-item questionnaire measures three modality-specific aspects of temperament: psychomotor, intellectual, and communicative activity. Items are rated on a 5-point Likert scale.

Cognitive Personality Style Questionnaire (CPS-Q; Volkova, Rusalov, & Dudnikova, 2022): This questionnaire assesses individual differences in cognitive style regulation across multiple dimensions: rigidity/flexibility of cognitive control, impulsivity/reflectivity, etc. Scores on each scale range from 5 to 25 points.

Raven’s Standard Progressive Matrices (SPM; Raven, Court, & Raven, 1987): This widely used test of nonverbal abstract reasoning was administered to assess the level of intelligence.

• EEG Recording and Preprocessing

Continuous EEG data were recorded using a 30-channel Encephalan-EEGR-19/26 electroencephalograph (Medicom MTD, Taganrog, Russia; European CE 538571 certificate). Ag/AgClelectrodes were placed on the scalp according to the international 10–20 system (Jasper, 1958). The ground electrode was placed on the forehead, and linked earlobes (A1 and A2) served as the reference. Vertical and horizontal electrooculograms (EOG) were recorded using bipolar electrodes placed above and below the left eye and at the outer canthi of both eyes to monitor ocular artifacts.

The EEG signal was band-pass filtered online between 0.5 Hz and 50 Hz. Electrode impedance was kept below 5 kΩ for all channels.

Ocular artifacts (blinks and eye movements) were removed using independent component analysis (ICA; Jung et al., 2000). Residual artifacts (e.g., muscle activity, electrode noise) were identified and removed through visual inspection of the data and the ICA components.

The continuous data were then segmented into epochs time-locked to stimulus onset. For each cognitive task, artifact-free epochs of 10 seconds duration were selected for analysis. A total of five epochs were selected per participant per task condition, corresponding to the beginning (two epochs), middle (one epoch), and end (two epochs) of the task block, to capture potential time-on-task effects. Epochs containing any remaining artifacts (voltage fluctuations exceeding ±100 µV) were excluded from further analysis.

Spectral power was computed for each artifact-free epoch using a fast Fourier transform (FFT) with a Hanning window (10% taper) to minimize spectral leakage. Power spectral density (µV²/Hz) was estimated for the following frequency bands, defined according to standard conventions: Delta (δ): 1–4 Hz, Theta (θ): 4–8 Hz, Alpha (α): 8–13 Hz, Beta1 (β1): 13–20 Hz, Beta2 (β2): 20–30 Hz.

Absolute power (µV²) was calculated for each frequency band by averaging the power spectral density across the corresponding frequency range. Spectral power was computed for all 30 electrode sites and subsequently averaged across regions of interest (frontal, central, parietal, temporal, occipital) for statistical analysis, based on the standard 10–20 system groupings.

Statistical Analysis

Statistical analyses were conducted using IBM SPSS Statistics version 28. Descriptive statistics (means, standard deviations, medians, and interquartile ranges) were computed for all demographic, behavioral, and EEG variables. The normality of the data distribution was assessed using the Shapiro–Wilk test and visual inspection of Q–Q plots.

Given that the assumption of normality was violated for several key variables (as indicated by significant Shapiro–Wilk tests), non-parametric statistical methods were employed for group comparisons. Specifically, the Kruskal–Wallis H test (one-way analysis of variance by ranks) was used to examine differences in EEG spectral power and behavioral performance (accuracy, reaction time) across independent groups: 13–14 years, 15–17 years, and 18–27 years.

Results

3.1. Age-Related Differences in Psychological Characteristics

The psychological characteristics of the three age groups (13–14 years, 15–17 years, and 18–27 years) are presented in Table 1. As shown in Table 1, participants across all groups were characterized by below-average levels of intellectual activity and fluid intelligence (as measured by Raven’s Standard Progressive Matrices), as well as average levels of cognitive style dimensions, including rigidity–flexibility of cognitive control and reflection–impulsivity (assessed by the CPS-Q). However, age-related trends were evident: participants aged 13–14 years exhibited a greater tendency toward reflectiveness, whereas those aged 15–17 years demonstrated higher impulsivity. In contrast, the oldest group (18–27 years) showed more pronounced reflectiveness, along with higher levels of both cognitive flexibility and rigidity. The highest overall levels of psychomotor, intellectual, and communicative activity were observed in the 15–17-year-old group.

Table 2. The significance of age-related differences in the EEG power spectrum

Significant age-related differences were observed across multiple cognitive tasks. Accuracy on the chemical compound classification task, the elementary logical operations (ELO) test, and the Matching Familiar Figures Test (MFFT) increased significantly with age. Concurrently, response times decreased significantly with age for the ELO test, the MFFT, and the first and third subtests of the Stroop task; however, the change in response time for the second Stroop subtest did not reach statistical significance.

For the chemical classification task, a distinct pattern emerged as a function of task complexity and age. The time required to complete simple classifications was highest in the youngest age group (13–14 years). For complex classifications, response times peaked in the middle group (15–17 years), while the most complex classifications elicited the longest response times in the oldest group (18–27 years). Based on accuracy and response time data, the zone of proximal development for the youngest group encompassed simple classifications, specifically the recognition of simple and complex compounds—a finding consistent with the increased time these participants dedicated to the task. For the 15–17-year-old group, the zone of proximal development extended to the recognition of classes of inorganic substances. For the oldest group (18–27 years), it included the most complex level of chemical compound classification.

On the MFFT, a statistically significant increase in first response latency was observed with age, indicating a more reflective approach. Although the number of errors decreased with age, this reduction did not reach statistical significance.

On the ELO test, both speed and accuracy improved significantly across age groups, reflecting enhanced logical reasoning abilities.

On the Stroop task, the time required to read color names (congruent condition) and to read color names printed in incongruent ink colors (interference condition) decreased significantly with age. In contrast, the time required to name the color of a shape (a control condition) did not change significantly across groups. These findings suggest an age-related increase in cognitive flexibility and verbal processing efficiency.

Finally, consistent with the developmental literature, overall intelligence test scores (Raven’s SPM) increased progressively across the three age groups.

3.2. Age-Related Differences in EEG Spectral Power During Cognitive Task Performance

To investigate age-related changes in neural efficiency, EEG spectral power was compared across three age groups (13–14 years, 15–17 years, and 18–27 years) for each cognitive task and frequency band. A Kruskal–Wallis H test was conducted to assess overall group differences for each electrode site. The pattern of results across tasks and frequency bands is presented below. A consistent trend was observed across most tasks and frequency bands: the youngest group (13–14 years) typically exhibited the highest spectral power, while the oldest group (18–27 years) exhibited the lowest. However, notable deviations from this pattern emerged, particularly in the beta frequency bands and for specific tasks.

3.2.1. Chemical Compound Classification Task

Analysis of EEG spectral power during the classification of simple and complex chemical compounds revealed statistically significant age group differences across all 30 electrode sites in the delta (δ) band (1–4 Hz). As shown in Table 2, the lowest δ power values were consistently observed in the 18–27-year-old group, while the highest values were observed in the 13–14-year-old group, with the exception of electrodes Fp2, Pz, Cz, and FC4.

In the theta (θ) band (4–8 Hz), significant group differences were found at 27 electrode sites (all except C4, F4, and Fp2). At all significant sites, θ power was lowest in the oldest group (18–27 years) and highest in the youngest group (13–14 years), with the exception of electrodes C3 and F3, where the 15–17-year-old group showed higher values than the youngest group.

Consistent with the neural efficiency hypothesis, the combination of lower δ and θ power in the oldest group, alongside their higher task accuracy (see Table 2), suggests that neural efficiency in solving simple chemical classification tasks increases with age.

In accordance with previous research indicating that the alpha (α) rhythm is not fully mature until late adolescence or early adulthood (Tröndle & Langer, 2024; Mason et al., 2022), spectral power in the α band (8–13 Hz) was excluded from the present analysis.

In the beta1 (β1) band (13–20 Hz), significant group differences were observed at all 30 electrode sites. As in the δ and θ bands, the highest β1 power was consistently found in the 13–14-year-old group. However, the lowest β1 power was more frequently observed in the 15–17-year-old group (19 sites) than in the 18–27-year-old group (11 sites).

In the beta2 (β2) band (20–30 Hz), which is traditionally associated with cognitive complexity, significant group differences were found at 21 electrode sites. The pattern of results was more heterogeneous: the highest β2 power was observed at 10 sites in the 18–27-year-old group, at 9 sites in the 13–14-year-old group, and at 2 sites in the 15–17-year-old group. The lowest β2 power was predominantly observed in the 15–17-year-old group, except for electrodes O2, C3, Oz, and CP3, where the minimum values were found in the oldest group. The elevated β2 power in the youngest group may reflect the greater subjective difficulty of the task for these participants, who have only recently begun their formal education in chemistry and consequently lack the “chemical experience” of the older groups.

During the recognition of inorganic compound classes, significant group differences in the δ and θ bands were again observed at all 30 electrode sites. Across both bands, the 13–14-year-old group exhibited the highest power values, and the 18–27-year-old group the lowest.

In the β1 band, significant differences were found at 13 electrode sites. At all but one of these sites (Oz), the maximum power was observed in the 13–14-year-old group. The minimum β1 power was observed in the 15–17-year-old group at nine sites, and in the 18–27-year-old group at the remaining four sites (O2, P4, Oz, FC4).

In the β2 band, significant group differences were identified at six electrode sites (O2, Fpz, FC3, FC4, FT8, TP7). The pattern varied by site:

At O2 and FC3, the lowest power was in the 18–27-year-old group.

At Fpz, FT8, and TP7, the lowest power was in the 15–17-year-old group.

At FC4, the lowest power was in the 13–14-year-old group.

The highest β2 power was observed in the youngest group at O2, Fpz, FT8, and TP7, and in the middle group (15–17 years) at FC3 and FC4.

For the most complex level of chemical classification, significant group differences in the δ band were found at all 30 sites. Across all sites, power was lowest in the 18–27-year-old group and highest in the 13–14-year-old group.

In the θ band, significant differences were found at 25 sites. Again, power was lowest in the oldest group and highest in the youngest group at all significant sites.

In the β1 and β2 bands, significant differences were found at 9 and 10 sites, respectively. In both beta bands, spectral power was consistently higher in the 13–14-year-old group and lower in the 18–27-year-old group.

3.2.2. Matching Familiar Figures Test (MFFT)

During the MFFT (similar images search task), significant group differences in the δ band were observed at all 30 electrode sites. Power was highest in the 13–14-year-old group and lowest in the 18–27-year-old group at every site.

In the θ band, significant differences were found at 26 sites. As in the δ band, power was lowest in the oldest group and highest in the youngest group at all significant sites.

In the β1 band, significant differences were observed at 19 sites. The highest power was again found in the 13–14-year-old group, and the lowest in the 18–27-year-old group.

In the β2 band, significant differences were found at 14 sites. The pattern was more complex:

The highest power was observed in the youngest group at electrodes O2, P3, C3, Fp2, T3, and Cz, and in the oldest group at all other significant sites.

The lowest power was observed in the oldest group at O2, P3, and C3; in the youngest group at T5, FCz, and FC4; and in the middle group (15–17 years) at all remaining sites.

As shown in Table 1, the oldest group (18–27 years) demonstrated high accuracy on the MFFT but exhibited slower response times (longer first response latency). Based on the scalp topography of the EEG effects (involving parietal and central sites), this pattern may suggest a greater reliance on memory-related processes and cognitive functions such as sustained attention and logical–abstract thinking in this age group.

3.2.3. Elementary Logical Operations Test (ELO)

During the ELO test, significant group differences in the δ band were observed at all 30 electrode sites. Power was lowest in the 18–27-year-old group and highest in the 13–14-year-old group across all sites.

In the θ band, significant differences were found at 24 sites. The same pattern was observed: the oldest group exhibited the lowest power, and the youngest group the highest.

In the β1 band, significant differences were found at nine electrode sites. Again, power was generally lowest in the oldest group and highest in the youngest group, with the sole exception of electrode F7, where the maximum power was observed in the 15–17-year-old group.

In the β2 band, a more contradictory pattern emerged:

At electrode C3, power was lowest in the oldest group and highest in the youngest group.

However, at electrodes Cz, FC4, and FT8, power was lowest in the youngest group (13–14 years) and highest in the oldest group (18–27 years).

At Cz, Fpz, FC4, and FT8, the maximum values were observed in the oldest group.

3.2.4. Stroop Task

Subtest 1: Reading Color Names Printed in Black Ink

In the δ band, significant group differences were found at all 30 sites. Power was highest in the 13–14-year-old group and lowest in the 18–27-year-old group.

In the θ band, significant differences were again found at all 30 sites. Power was lowest in the oldest group at all sites. The highest power was observed in the youngest group at most sites, with the exception of O1, P3, C3, F3, T3, Fpz, FC3, and Cpz, where the 15–17-year-old group exhibited the highest values.

In the β1 band, significant differences were found at 18 sites. At all of these, power was highest in the youngest group and lowest in the oldest group.

In the β2 band, significant differences were found at 11 sites, following the same pattern: highest power in the 13–14-year-old group, lowest in the 18–27-year-old group.

Subtest 2: Naming the Color of Shapes

In the δ band, significant group differences were found at all 30 sites. The highest power was observed in the 15–17-year-old group at 20 sites, and in the 13–14-year-old group at the remaining 10 sites. The lowest power was consistently observed in the 18–27-year-old group.

In the θ band, significant differences were found at 27 sites. The highest power was split between the 13–14-year-old group (11 sites) and the 15–17-year-old group (16 sites). The lowest power was again observed in the oldest group.

In the β1 band, significant differences were found at 22 sites. The distribution of maximum power was: three sites in the 13–14-year-old group, 12 sites in the 15–17-year-old group, and seven sites in the 18–27-year-old group. The distribution of minimum power was 11, 3, and 7 sites for the youngest, middle, and oldest groups, respectively.

In the β2 band, significant differences were found at 14 sites. At all of these, power was lowest in the 15–17-year-old group and highest in the 13–14-year-old group.

Subtest 3: Naming the Color of Incongruent Color Words (Stroop Interference)

In the δ band, significant group differences were found at all 30 sites. Power was lowest in the 18–27-year-old group at all sites. The highest power was observed in the 13–14-year-old group at seven sites and in the 15–17-year-old group at the remaining 23 sites.

In the θ band, significant differences were found at all 30 sites. Again, power was lowest in the oldest group. The highest power was observed more frequently in the 15–17-year-old group (17 sites) than in the 13–14-year-old group (13 sites).

In the β1 band, significant differences were found at 16 sites. Power was lowest in the 18–27-year-old group at all significant sites. The highest power was observed in the 15–17-year-old group at 10 sites and in the 13–14-year-old group at six sites.

In the β2 band, significant differences were found at 10 sites. The highest power was observed in the 18–27-year-old group at seven sites and in the 13–14-year-old group at three sites. The lowest power was observed in the 13–14-year-old group (two sites) and the 15–17-year-old group (eight sites).

Summary of Age-Related Trends: Across the majority of cognitive tasks and frequency bands, a consistent pattern emerged: the youngest participants (13–14 years) exhibited the highest EEG spectral power, while the oldest participants (18–27 years) exhibited the lowest. This pattern is consistent with the neural efficiency hypothesis, suggesting that brain activity becomes more focal and energy-efficient with maturation and cognitive development. However, several notable exceptions to this trend were observed:

In the β1 band during simple and complex chemical classification, the lowest power was more frequently observed in the 15–17-year-old group than in the oldest group, suggesting that the peak of neural efficiency for certain tasks may occur in mid-adolescence.

In the β2 band, associated with cognitive complexity, a more heterogeneous pattern emerged. During the most challenging tasks (e.g., complex chemical classification, Stroop interference), the oldest group sometimes exhibited increased beta power, potentially reflecting the recruitment of additional neural resources to meet high task demands—a pattern consistent with the neural efficiency paradox.

Task-specific variations were observed, particularly for the MFFT and the ELO test, where the topography of effects suggested differential involvement of memory, attention, and logical reasoning processes across age groups.

These findings underscore the dynamic nature of neural efficiency, which varies not only with age but also with task type, task complexity, and frequency band.

Discussion

The primary aim of the present study was to investigate age-related differences in neural efficiency by comparing EEG spectral power across multiple frequency bands during the performance of cognitive tasks of varying complexity, while controlling for psychological characteristics of participants. The results revealed a nuanced pattern that both supports and extends the classical neural efficiency hypothesis (Haier et al., 1988), while also providing empirical evidence for the recently proposed “neural efficiency paradox” (Mann, Wright, & Janelle, 2016; Li & Smith, 2021). The overarching finding is that neural efficiency is not a monolithic phenomenon but rather a dynamic, context-dependent property of the developing brain, modulated by task complexity, frequency band, and the specific cognitive processes engaged.

• Evidence Supporting Classical Neural Efficiency

Across the majority of cognitive tasks and frequency bands, a consistent pattern was observed: the youngest participants (13–14 years) exhibited the highest spectral power, while the oldest participants (18–27 years) exhibited the lowest. This pattern was most pronounced in the delta (δ) and theta (θ) bands, which are associated with sustained attention, memory encoding, and cognitive control (Cavanagh & Frank, 2014). For example, during simple and complex chemical compound classification, as well as during the Matching Familiar Figures Test (MFFT) and the Elementary Logical Operations (ELO) test, δ and θ power were uniformly lowest in the 18–27-year-old group and highest in the 13–14-year-old group (see Table 2). This inverse relationship between age and low-frequency oscillatory activity is consistent with the classical neural efficiency hypothesis, which posits that more capable individuals (in this case, older, more cognitively mature participants) perform cognitive tasks with greater neural economy, consuming fewer metabolic resources (Haier et al., 1988; Neubauer & Fink, 2009).

The reduction in low-frequency power with age may reflect not merely decreased metabolic cost, but more efficient organization of functional brain states. Ericson et al. (2025) showed that optimal cognitive performance depends on switching between distinct networks – a posterior theta network for encoding/flexibility and a dorsal alpha network for maintenance/stability. Lower delta and theta power in our oldest group may therefore indicate more rapid, efficient transitions between these states, requiring less sustained oscillatory engagement.

Our finding aligns with Azamin et al. (2019), who showed that higher IQ is associated with increased alpha ratios and decreased theta/beta ratios in resting EEG – a pattern reflecting efficient attentional regulation. Together with Jabes et al.’s (2021) observations of similar low-frequency reductions in older adults, these findings support a lifespan trajectory of neural efficiency from adolescence through old age. Kit et al. (2023) further demonstrated that combinations of theta, alpha, and beta bands achieve up to 95% accuracy in stress classification, with beta features critical for high-stress conditions—mirroring our finding of increased β2 power during complex tasks. This convergence confirms that these bands carry complementary information that shifts with task demands.

The pattern of reduced low-frequency power alongside improved performance also aligns with conceptualizations of ‘system intelligence’—the capacity to integrate cognitive, emotional, and social processes in goal-directed behavior (Heilala & Karwowski, 2023). This framework links neural efficiency to flow experiences and autotelic personality traits (Tse et al., 2021). The focused high-frequency activation in young adults during complex tasks may thus reflect not merely reduced metabolic cost, but the emergence of integrated, system-level cognitive processing supporting flow states—consistent with the differentiation-integration theory (Volkova, 2011).

The behavioral data confirm this interpretation. As shown in Table 1, accuracy on the chemical classification tasks, the MFFT, and the ELO test increased significantly with age, while response times decreased. The combination of superior behavioral performance and reduced low-frequency EEG power in the oldest group strongly suggests that neural networks become more selective and efficient with maturation, requiring less widespread cortical activation to achieve optimal task outcomes (Pfurtscheller & Lopes da Silva, 1999; Klimesch, 2012).

• The Neural Efficiency Paradox: Increased High-Frequency Activation During Complex Tasks

However, a strikingly different pattern emerged in the beta2 (β2) band (20–30 Hz), which is traditionally associated with cognitive complexity, active information processing, and the maintenance of cognitive sets (Engel & Fries, 2010). During the most demanding cognitive tasks—specifically, the most complex level of chemical compound classification, the Stroop interference condition (naming the color of incongruent color words), and the ELO test—the oldest group (18–27 years) exhibited increased β2 power relative to the younger groups. This was most evident at fronto-central electrode sites (Cz, FC4, Fpz) and parietal sites (Pz, Oz).

This finding directly illustrates the neural efficiency paradox (Mann, Wright, & Janelle, 2016). In contrast to the classical hypothesis, successful performance on highly complex tasks in experts or high-ability individuals is associated with greater, not lesser, cortical activation, particularly in high-frequency bands. Li and Smith (2021) have argued that this reflects the recruitment of additional, specialized neural resources to meet the demands of tasks that exceed a certain threshold of complexity. In the present study, the oldest participants, who had the highest levels of intellectual development (Raven’s SPM) and cognitive flexibility (CPS-Q), appear to have engaged distinct neural strategies for these challenging tasks.

The interpretation of increased β2 power as a marker of focused cognitive control is strongly supported by Taddeini et al. (2025), who showed that successful inhibitory control was characterized by higher pre-stimulus beta power, while fast, erroneous responses followed periods of low beta activity—indicating a lapse in proactive control. Viewed through this lens, increased β2 power in our oldest group during Stroop interference reflects successful engagement of proactive inhibitory mechanisms. Younger groups, with lower β2 power, may fail to engage this critical control process.

The increased β2 power may also be understood within the stability-flexibility framework (Ericson et al., 2025), where optimal performance requires transitioning between states supporting stability (dorsal alpha) and flexibility (posterior theta). The focused β2 increases during our most demanding tasks—particularly at frontal sites during Stroop interference—may reflect recruitment of a high-frequency ‘control state’ facilitating rapid reconfiguration when task demands exceed low-frequency network capacity. Thus, the neural efficiency paradox resolves when considering that low frequencies support baseline state maintenance, while high frequencies enable active reconfiguration under high demand.

The topographic specificity of these effects provides insight into the underlying cognitive processes. During the most complex chemical classification, increased β2 power in the oldest group was observed at parietal and occipital sites (Pz, Oz), suggesting engagement of visuospatial analysis and mental imagery—a plausible strategy for discriminating between complex chemical formulas (Solis-Ortiz & Corsi-Cabrera, 2008). During the Stroop interference task, increased β2 power was maximal at frontal sites (Fpz, FCz, FC4), consistent with the well-established role of prefrontal cortex in inhibitory control, conflict monitoring, and cognitive flexibility (Cavanagh & Frank, 2014). During the ELO test, increased β2 power at Cz and FC4 may reflect the integration of spatial thinking (Cz) and working memory/attentional control (FC4) required for logical reasoning (Gevins et al., 1997).

This pattern parallels Luo and Zhou’s (2020) finding of larger P3 amplitudes in high-ability children—both reflecting enhanced attentional engagement under cognitive load. Our findings are also consistent with Steiner et al. (2023), who showed that resting-state theta and beta power in frontal areas distinguishes high-performing individuals.

Thus, the paradox is resolved by considering task complexity: for simple tasks, efficiency manifests as reduced activation; for complex tasks, it manifests as precise, focused activation of task-relevant networks, reflected in increased high-frequency power localized to specific cortical regions.

• Task-Specific Cognitive Mechanisms Across Development

The pattern of results also revealed distinct developmental trajectories and cognitive mechanisms for each task, which can be interpreted within the framework of the differentiation-integration theory of abilities (Volkova, 2011; Ratanova, 2011; Chuprikova, 2007).

Chemical Compound Classification. The three levels of this task (simple, complex, most complex) appear to lie in different “zones of proximal development” (Vygotsky, 1983) for the three age groups. For the youngest group (13–14 years), simple classifications fell within their zone of actual development: they achieved moderate accuracy but required maximal time and exhibited the highest δ and θ power, indicating effortful, resource-intensive processing. For the middle group (15–17 years), complex classifications (recognition of inorganic compound classes) fell within their zone of proximal development: they achieved high speed with low β1/β2 power (except at Oz, FC3, FC4), suggesting available cognitive resources but insufficient domain-specific “chemical experience.” For the oldest group (18–27 years), the most complex classifications fell within their zone of proximal development: they possessed the necessary cognitive resources (as indexed by high intelligence and cognitive flexibility) but lacked the specialized chemical knowledge to achieve high accuracy, a finding consistent with Volkova’s (2011) emphasis on the interplay between general intelligence and domain-specific aptitude. The age-related differences in the chemical classification task may also involve the maturation of conceptual integration networks. Work by Mayer et al. (2025) highlights the anterior temporal lobe’s (ATL) role in integrating multimodal information into coherent concepts, noting that in adolescents this region acts as a demanding central hub. The reduced low-frequency power and task-specific beta2 increases we observe in young adults during complex classifications may therefore reflect a mature ATL network that can efficiently manipulate conceptual chemical knowledge without the widespread, high-power engagement seen in younger participants.

The load-dependent modulation in our study parallels Ericson et al. (2025), who found dorsal alpha synchronization peaked at moderate load (3 items) and decreased at higher loads, suggesting capacity limits for stability networks. In our chemical classification task, the oldest group’s increased β2 power at the most complex level may represent engagement of a complementary high-frequency network when low-frequency stability networks reach capacity. Additionally, their reported U-shaped relationship between state-switching rate and performance may explain our age differences: the youngest group may exhibit suboptimal switching dynamics (excessive flexibility or stability), while the oldest achieves more balanced state transitions.

Elementary Logical Operations (ELO). The ELO test, designed to assess fluid intelligence and logical reasoning (Dokuchaev & Volkova, 2021; Rusalov & Volkova, 2021), revealed the most striking example of the neural efficiency paradox. The oldest group achieved the highest speed and accuracy (Table 1) but exhibited the highest β2 power at Cz, Fpz, FC4, and FT8. As noted above, these sites have been linked to spatial thinking (Cz), information synthesis (Fpz), working memory (FC4), and memory retrieval (FT8) (Tran, Craig, & McIsaac, 2001; Volf & Razumnikova, 1999). This pattern suggests that successful logical reasoning in young adults engages a distributed network of higher-order cognitive processes, each contributing to efficient problem-solving. In contrast, the middle group (15–17 years) achieved intermediate performance with generally lower β2 power, suggesting a less integrated, perhaps more effortful but less specialized, neural strategy.

Stroop Task. The Stroop task, a classic measure of inhibitory control and cognitive flexibility (Stroop, 1935), provided further insight into the neural efficiency paradox. For the simple condition (reading color names), performance improved with age, and spectral power decreased across all bands, consistent with classical efficiency. For the complex interference condition (naming the color of incongruent words), however, the oldest group exhibited the fastest response times but the highest β2 power at multiple sites, particularly frontal electrodes. This pattern aligns with the work of Lansbergen and colleagues (2007), who demonstrated that successful conflict resolution in the Stroop task is associated with enhanced activation of prefrontal inhibitory networks. The fact that this effect was most pronounced in the oldest group, who also had the highest levels of cognitive flexibility (CPS-Q), suggests that mature cognitive control requires not just inhibition but the active, energy-consuming engagement of prefrontal circuits to override prepotent responses (Baddeley & Della Sala, 1996; Cohen et al., 1990).

The dynamical systems framework of Ritz et al. (2025) further illuminate these findings. In their model, cognitive flexibility involves convergence to a ‘neutral’ task state during preparation, followed by a cue-driven ‘task energy’ pulse. Reduced low-frequency power in the oldest group during simple conditions may reflect efficient convergence to this neutral baseline. Increased β2 power during interference may index focused ‘task energy’ to actively reconfigure the system. This suggests mature cognitive control is characterized by strategic, temporally precise deployment of high-frequency resources.

The age-related dynamics may also reflect a shift between two distinct response modes (Taddeini et al., 2025). Cognitive load can trigger a ‘fast trial’ mode with automatic responses lacking frontal preparation. The diffuse, high-power EEG in our youngest group during complex tasks may represent inability to consistently engage the controlled ‘standard’ mode, resulting in poorly coordinated attempts. Conversely, the focused β2 increase in young adults’ indexes stable engagement of the controlled mode, enabling rapid, accurate reconfiguration. This aligns with differentiation-integration theory, suggesting maturation from a unitary, inefficient mode to a flexible, dual-mode system.

The observed increase in β2 power during the Stroop interference task in the oldest group, despite their faster performance, directly supports the neural efficiency paradox. It challenges the simplistic notion that “less activation is always better” and instead highlights that for complex executive function tasks, optimal performance may require the recruitment of specialized, energy-consuming neural circuits.

• The Role of Individual Differences: Temperament, Cognitive Style, and Intelligence

A critical feature of this study was the inclusion of psychological parameters (temperament, cognitive style, intelligence), which allows for a more nuanced interpretation of the EEG findings. As shown in Table 1, the sample was characterized by below-average levels of intellectual activity (STQ-s) and fluid intelligence (Raven’s SPM), as well as average levels of cognitive style dimensions (rigidity–flexibility, reflection–impulsivity). These characteristics likely modulated the observed patterns of neural efficiency.

For example, the unexpected finding during the color-naming control task (Stroop subtest 2), where spectral power increased with age despite minimal changes in speed, may be attributable to the sample’s low intelligence and unpronounced cognitive styles. In a more intellectually able sample, one might expect a clearer pattern of reduced activation with age even for this simple task (Neubauer & Fink, 2009). Similarly, the pronounced neural efficiency paradox observed during the Stroop interference task may be amplified in this sample due to their reduced baseline levels of cognitive control. Participants with lower cognitive control may need to expend greater neural effort to achieve the same level of performance as individuals with higher baseline control, a phenomenon consistent with the “compensation” hypothesis (Cabeza et al., 2002).

The dynamical systems perspective also offers a neurophysiological interpretation for the individual differences in cognitive style we observed. Specifically, the greater reflectivity and cognitive flexibility in the oldest group (CPS-Q) may reflect faster convergence to a ‘neutral’ task state or a more optimally scaled ‘task energy’ pulse (Ritz et al., 2025). Future studies should test whether state-space EEG parameters—such as convergence rate—mediate the relationship between cognitive style and behavioral performance on tasks like the Stroop test.

These findings underscore the importance of measuring and controlling for psychological characteristics in EEG studies of cognitive development. Failure to do so may obscure meaningful individual differences and lead to oversimplified conclusions about neural efficiency (Wacker & Stemmler, 2006).

Taken together, the results of this study support a dynamic, multi-faceted model of neural efficiency that integrates classical and paradoxical findings. We propose that neural efficiency is best conceptualized along two dimensions:

Task Complexity: For simple, well-learned, or low-demand tasks, neural efficiency manifests as reduced cortical activation, particularly in low-frequency bands (δ, θ), reflecting optimized, automatic processing (classical efficiency).

Network Specialization: For complex, novel, or high-demand tasks, neural efficiency manifests as the precise, focused, and energy-consuming recruitment of specialized cortical networks, reflected in increased high-frequency (β2) activation localized to task-relevant regions (the paradox).

This two-dimensional model aligns with the differentiation-integration theory of abilities (Volkova, 2011), which posits that cognitive development proceeds from global, undifferentiated neural representations toward differentiated, hierarchically integrated networks. The youngest group in this study (13–14 years) exhibited a global, undifferentiated pattern of high power across all bands and tasks, reflecting immature neural specialization. The middle group (15–17 years) showed evidence of emerging specialization, with reduced power in some bands/tasks but not others. The oldest group (18–27 years) demonstrated the most mature pattern: differentiated, task-specific engagement of neural resources, with reduced low-frequency power for routine processing and enhanced high-frequency power in specialized networks for complex challenges.

• Limitations and Future Directions

Several limitations of this study should be acknowledged. First, the cross-sectional design limits inferences about developmental change; longitudinal studies are needed to confirm the observed age-related trends. Second, the sample was characterized by below-average intelligence and specific temperamental profiles, which may limit the generalizability of the findings to other populations. Third, the analysis focused on spectral power; future studies should complement this with measures of functional connectivity (e.g., coherence, phase synchronization) to better characterize network dynamics. Fourth, the absence of a true “expert” group in the chemical domain limited our ability to fully dissociate age effects from domain-specific expertise. Finally, the use of a 10-second epoch length, while appropriate for capturing sustained cognitive processing, may have obscured finer-grained temporal dynamics.

Future research should aim to replicate these findings in larger, more diverse samples, incorporate longitudinal designs, and employ multimodal neuroimaging (e.g., EEG-fMRI) to better localize the sources of age-related changes in oscillatory activity. Additionally, studies manipulating task complexity parametrically within individuals would provide a more direct test of the proposed two-dimensional model of neural efficiency.

Conclusions

This study provides several novel contributions to the understanding of neural efficiency across development:

1. Neural efficiency is age-dependent and task-dependent. The classical pattern of reduced cortical activation with increasing age and ability was confirmed for simple tasks and low-frequency bands (δ, θ). However, this pattern reversed for complex tasks in the high-frequency β2 band, supporting the neural efficiency paradox.
2. The neural efficiency paradox is empirically validated. During the most demanding cognitive tasks—complex chemical classification, Stroop interference, and logical reasoning—the oldest, most cognitively mature group exhibited increased β2 power alongside superior behavioral performance. This challenges the simplistic view that “less is always more” and highlights the importance of task complexity in moderating neural efficiency effects.
3. Task-specific neural strategies emerge with development. Topographic analysis revealed that increased β2 power in the oldest group was localized to cortical regions appropriate for each task: parietal sites for visuospatial-chemical processing, frontal sites for inhibitory control, and fronto-central sites for logical reasoning and working memory integration.
4. Psychological characteristics matter. The inclusion of temperament, cognitive style, and intelligence measures allowed for a more nuanced interpretation of the EEG findings. The observed patterns of neural efficiency were modulated by individual differences in these psychological constructs, underscoring the need for multi-level approaches in cognitive neuroscience.
5. The differentiation-integration theory provides a useful framework. The developmental trajectory observed—from global, undifferentiated high power in the youngest group to differentiated, task-specific engagement in the oldest group—is consistent with the differentiation-integration theory of abilities (Volkova, 2011) and suggests that neural efficiency is an emergent property of maturing representational-cognitive structures.
6. The obtained data, together with the results of modern research (Steiner et al., 2023), suggest that neural efficiency has not only a situational (task-specific) but also a stable, tonic component, reflected in resting-state EEG patterns. This opens prospects for using machine learning methods to predict the success of cognitive activity based on individual neurophysiological profiles.

In sum, this study demonstrates that the developing brain navigates a complex trade-off between energy conservation and the recruitment of specialized neural resources. The “efficient” brain is not simply one that does more with less; it is one that flexibly adapts its neural strategy to the demands of the task, conserving energy when possible but investing it strategically when necessary. This dynamic, context-sensitive view of neural efficiency has important implications for theories of cognitive development, educational practice, and our understanding of individual differences in human abilities.

Ethics Statement: All procedures performed in this study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The study protocol was approved by the Local Ethics Committee of the Institute of Psychology of the Russian Academy of Sciences (Protocol No. 7, July 10, 2020). Written informed consent was obtained from all individual participants included in the study. For participants under 18 years of age, written informed consent was obtained from a parent or legal guardian, and written assent was obtained from the minor participant. All data were anonymized prior to analysis. Personal identifiers were removed and replaced with unique participant codes to ensure privacy and confidentiality throughout the research process.

CRediT author statement: The author made a substantial contribution to the conception and design of the work, the acquisition and analysis of data, and the drafting and revision of the manuscript. The authors have approved the final version for publication and agree to be accountable for all aspects of the work.

D.A. Dokuchaev: Investigation, Formal analysis, Visualization, Writing – original draft.

N.E. Volkova: Conceptualization, Formal analysis, Writing – review & editing.

Conflict of interest: The authors declare no conflict of interest.

Funding: The study was performed as part of the state assignment “Intelligent Systems and Human Abilities” – 0138-2025-0016. The registration number in the Unified State Information System for Research. Development and Technological Work (EGISU NIOTKR) is 1023033000550-6-5.1.1

References

1. Azamin, N. H. R., Taib, M. N., Jahidin, A. H., Awang, D. S., & Megat Ali, M. S. A. (2019). IQ level prediction and cross-relational analysis with perceptual ability using EEG-based SVM classification model. IAES International Journal of Artificial Intelligence (IJ-AI), 8(4), 436– https://doi.org/10.11591/ijai.v8.i4.pp436-442
2. Baddeley, A. D., & Della Sala, S. (1996). Working memory and executive control. Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences, 351(1346), 1397–1404. https://doi.org/10.1098/rstb.1996.0123
3. Cabeza, R., Anderson, N. D., Locantore, J. K., & McIntosh, A. R. (2002). Aging gracefully: Compensatory brain activity in high-performing older adults. NeuroImage, 17(3), 1394– https://doi.org/10.1006/nimg.2002.1280
4. Cavanagh, J. F., & Frank, M. J. (2014). Frontal theta as a mechanism for cognitive control. Trends in Cognitive Sciences, 18(8), 414–421. https://doi.org/10.1016/j.tics.2014.04.012
5. Chuprikova, N. I. (2007). Umstvennoe razvitie. Printsip differentsiatsii [Mental development: The principle of differentiation]. Piter.
6. Chuprikova, N. I. (2019). Vremya reaktsiy cheloveka: Fiziologicheskie mekhanizmy, verbal’no-smyslovaya regulyatsiya, svyaz’ s intellektom i svoystvami nervnoy sistemy [Human reaction time: Physiological mechanisms, verbal-semantic regulation, relationship with intelligence and properties of the nervous system]. LRC Publishing House.
7. Cohen, J. D., Dunbar, K., & McClelland, J. L. (1990). On the control of automatic processes: A parallel distributed processing account of the Stroop effect. Psychological Review, 97(3), 332–361.
   https://doi.org/10.1037/0033-295X.97.3.332
8. Dokuchaev, D. A., & Volkova, N. E. (2021). Inductive reasoning and neural efficiency. Natural Systems of Mind, 1(2), 53–78. https://doi.org/10.38098/nsom_2021_01_04_05
9. Donoghue, T., Haller, M., Peterson, E. J., Varma, P., Sebastian, P., Gao, R., Noto, T., Lara, A. H., Wallis, J. D., Knight, R. T., Shestyuk, A., & Voytek, B. (2020). Parameterizing neural power spectra into periodic and aperiodic components. Nature Neuroscience, 23(12), 1655–1665. https://doi.org/10.1038/s41593-020-00744-x
10. Engel, A. K., & Fries, P. (2010). Beta-band oscillations – Signalling the status quo? Current Opinion in Neurobiology, 20(2), 156–165. https://doi.org/10.1016/j.conb.2010.02.015
11. Ericson, J., Ruiz Ibanez, N., Lundqvist, M., & Klingberg, T. (2025). Low frequency oscillations – Neural correlates of stability and flexibility in cognition. Nature Communications, 16, Article 60821. https://doi.org/10.1038/s41467-025-60821-2
12. Gevins, A., Smith, M. E., McEvoy, L., & Yu, D. (1997). High-resolution EEG mapping of cortical activation related to working memory: Effects of task difficulty, type of processing, and practice. Cerebral Cortex, 7(4), 374–385. https://doi.org/10.1093/cercor/7.4.374
13. Haier, R. J., Siegel, B. V., Nuechterlein, K. H., Hazlett, E., Wu, J. C., Paek, J., Browning, H. L., & Buchsbaum, M. S. (1988). Cortical glucose metabolic rate correlates of abstract reasoning and attention studied with positron emission tomography. Intelligence, 12(2), 199–217. https://doi.org/10.1016/0160-2896(88)90016-5
14. Heilala, J., & Karwowski, W. (2023). Unlocking human potential: The power of neural-interface technology measuring cognitive ability and traits. In AHFE International (Vol. 100, pp. 248–257). AHFE International. https://doi.org/10.54941/ahfe1003025
15. Jabès, A., Klencklen, G., Ruggeri, P., Antonietti, J.-P., Banta Lavenex, P., & Lavenex, P. (2021). Age-related differences in resting-state EEG and allocentric spatial working memory performance. Frontiers in Aging Neuroscience, 13, Article 704362. https://doi.org/10.3389/fnagi.2021.704362
16. Jasper, H. H. (1958). The ten-twenty electrode system of the International Federation. Electroencephalography and Clinical Neurophysiology, 10, 371–375.
17. Jung, T.-P., Makeig, S., Humphries, C., Lee, T.-W., McKeown, M. J., Iragui, V., & Sejnowski, T. J. (2000). Removing electroencephalographic artifacts by blind source separation. Psychophysiology, 37(2), 163–178.
18. Kagan, J. (1966). Reflection-impulsivity: The generality and dynamics of conceptual tempo. Journal of Abnormal Psychology, 71(1), 17–24. https://doi.org/10.1037/h0022886
19. Klimesch, W. (2012). Alpha-band oscillations, attention, and controlled access to stored information. Trends in Cognitive Sciences, 16(12), 606–617. https://doi.org/10.1016/j.tics.2012.10.007
20. Kit, N. K., Amin, H. U., Ng, K. H., Price, J., & Subhani, A. R. (2023). EEG feature extraction based on fast Fourier transform and wavelet analysis for classification of mental stress levels using machine learning. Advances in Science, Technology and Engineering Systems Journal, 8(6), 49–59. https://doi.org/10.25046/aj080606
21. Knorr, E., & Neubauer, A. C. (1996). Speed of information-processing in an inductive reasoning task and its relationship to psychometric intelligence. Personality and Individual Differences, 20(5), 653–660. https://doi.org/10.1016/0191-8869(95)00214-6
22. Lansbergen, M. M., Kenemans, J. L., & van Engeland, H. (2007). Stroop interference and attention-deficit/hyperactivity disorder: A review and meta-analysis. Neuropsychology, 21(2), 251–262. https://doi.org/10.1037/0894-4105.21.2.251
23. Leroy, S., Bublitz, V., von Dincklage, F., Antonenko, D., & Fleischmann, R. (2025). Normative characterization of age-related periodic and aperiodic activity in resting-state real-world clinical EEG recordings. Frontiers in Aging Neuroscience, 17, Article 1540040 https://doi.org/10.3389/fnagi.2025.1540040
24. Li, L., & Smith, D. M. (2021). Neural efficiency in athletes: A systematic review. Frontiers in Behavioral Neuroscience, 15, Article 698555. https://doi.org/10.3389/fnbeh.2021.698555
25. Luo, W., & Zhou, R. (2020). Can working memory task-related EEG biomarkers measure fluid intelligence and predict academic achievement in healthy children? Frontiers in Behavioral Neuroscience, 14, Article 2. https://doi.org/10.3389/fnbeh.2020.00002
26. McKeown, D. J., Roberts, E., Finley, A. J., Kelley, N. J., Keage, H. A. D., Schinazi, V. R., Baumann, O., Moustafa, A. A., & Angus, D. J. (2025). Lower aperiodic EEG activity is associated with reduced verbal fluency performance across adulthood. Neurobiology of Aging, 15129–41. https://doi.org/10.1016/j.neurobiolaging.2025.03.005
27. Mayer, J., Mückschel, M., Hommel, B., & Beste, C. (2025). Maturational changes in action-effect integration processes are reflected by changes in the directed cortical network communication. Human Brain Mapping, 46(13), Article e70339. https://doi.org/10.1002/hbm.70339
28. Mann, D. T. Y., Wright, A., & Janelle, C. M. (2016). Quiet eye: The efficiency paradox comment on Vickers. Current Issues in Sport Science, 1, Article 111. https://doi.org/10.15203/CISS_2016.111
29. Mason, L. M., Barry, R. J., & Clarke, A. R. (2022). Age-related changes in the EEG in an eyes-open condition: I. Normal development. International Journal of Psychophysiology, 172, 40–45. https://doi.org/10.1016/j.ijpsycho.2021.12.006
30. Neubauer, A. C., & Fink, A. (2009). Intelligence and neural efficiency. Neuroscience & Biobehavioral Reviews, 33(7), 1004–1023. https://doi.org/10.1016/j.neubiorev.2009.04.001
31. Pfurtscheller, G., & Lopes da Silva, F. H. (1999). Event-related EEG/MEG synchronization and desynchronization: Basic principles. Clinical Neurophysiology, 110(11), 1842–1857. https://doi.org/10.1016/S1388-2457(99)00141-8
32. Ratanova, T. A. (2011). Vremya reaktsii v sisteme izucheniya prirody intellekta i spetsial’nykh sposobnostey [Reaction time in the system of studying the nature of intelligence and special abilities]. Eksperimental’naya psikhologiya, 4(3), 86–96.
33. Raven, J. C., Court, J. H., & Raven, J. (1987). Manual for Raven’s Progressive Matrices and Vocabulary Scales. H. K. Lewis.
34. Ritz, H., Jha, A., Daw, N. D., & Cohen, J. D. (2025). Convergent neural dynamical systems for task control in artificial networks and human brains. bioRxiv. https://doi.org/10.1101/2024.09.29.615736
35. Rusalov, V. M. (2012). Temperament v strukture individual’nosti cheloveka. Differentsial’no-psikhofiziologicheskie i psikhologicheskie issledovaniya [Temperament in the structure of human individuality: Differential psychophysiological and psychological studies]. Institute of Psychology, Russian Academy of Sciences.
36. Rusalov, V. M., & Volkova, N. E. (2021). A test “Elementary Logical Operations”: Psychometric characteristics on the Russian sample. Natural Systems of Mind, 1(1), 48–58. https://doi.org/10.38098/nsom_2021_01_03_05
37. Solis-Ortiz, S., & Corsi-Cabrera, M. (2008). Sustained attention is favored by progesterone during early luteal phase and visuo-spatial memory by estrogens during ovulatory phase in young women. Psychoneuroendocrinology, 33(7), 989–998. https://doi.org/10.1016/j.psyneuen.2008.04.003
38. Steiner, H., Mikheev, I., & Martynova, O. (2023). Cross-subject classification of effectiveness in performing cognitive tasks using resting-state EEG. Applied Sciences, 13(11), Article 6606. https://doi.org/10.3390/app13116606
39. Stroop, J. R. (1935). Studies of interference in serial verbal reactions. Journal of Experimental Psychology, 18(6), 643–662. https://doi.org/10.1037/h0054651
40. Taddeini, F., Arrighi, G., Viglione, A. A., Cecchetti, J., Santi, F., Bertelli, D., Fiori, L., Pietrini, P., Ricciardi, E., Bernardi, G., & Mazzoni, A. (2025). Extended cognitive load induces fast neural responses leading to commission errors. eNeuro, 12(2), Article ENEURO.0354-24.2024. https://doi.org/10.1523/ENEURO.0354-24.2024
41. Tran, Y., Craig, A., & McIsaac, P. (2001). Extraversion–introversion and 8–13 Hz waves in frontal cortical regions. Personality and Individual Differences, 30(2), 205–215. https://doi.org/10.1016/S0191-8869(00)00027-1
42. Tse, D. C. K., Nakamura, J., & Csikszentmihalyi, M. (2021). Living well by “flowing” well: The indirect effect of autotelic personality on well-being through flow experience. The Journal of Positive Psychology, 16(3), 310–321. https://doi.org/10.1080/17439760.2020.1716055
43. Volf, N. V., & Razumnikova, O. M. (1999). Sex differences in EEG coherence during a verbal memory task in normal adults. International Journal of Psychophysiology, 34(2), 113–122. https://doi.org/10.1016/S0167-8760(99)00068-9
44. Volkova, E. V. (2011). Psikhologiya spetsial’nykh sposobnostey: Differentsionno-integratsionnyy podkhod [Psychology of special abilities: A differentiation-integration approach]. Institute of Psychology, Russian Academy of Sciences.
45. Volkova, E. V., & Nilopets, M. N. (2015). Konstruktor testov “INT”[Test constructor “INT”] [Computer software]. Russian Federation Patent No. RU 2015618029.
46. Volkova, E. V., & Nilopets, M. N. (2016). InTesting [Computer software]. Russian Federation Patent No. RU 2016661340.
47. Volkova, E. V., Rusalov, V. M., & Dudnikova, T. A. (2022). Cognitive and personal styles questionnaire: Validity and reliability of the CPS-Q based on a Russian sample. Natural Systems of Mind, 2(4), Article 03. https://doi.org/10.38098/nsom_2022_02_04_03
48. Volkova, N. E., & Dokuchaev, D. A. (2022). Neuroefficiency and the ability to classify stimulus-objects of chemistry. Natural Systems of Mind, 2(2), Article 04. https://doi.org/10.38098/nsom_2022_02_02_04
49. Vygotsky, L. S. (1983). Istoriya razvitiya vysshikh psikhicheskikh funktsiy [The history of the development of higher mental functions]. In A. M. Matyushkin (Ed.), Sobranie sochineniy: Vol. 3. Problemy razvitiya psikhiki (pp. 5–328). Pedagogika.
50. Wacker, J., & Stemmler, G. (2006). Sex differences in ERP and EEG correlates of decision-making and their relationship to personality.