Elevating understanding: Linking high-altitude hypoxia to brain aging through EEG functional connectivity and spectral analyses

It has been estimated that around 81.6 million people live at >2,500 masl (1.07% of the world’s population) (Tremblay & Ainslie, 2021). In addition, many people work at high altitudes, such as mining workers, who may work at >2,000 masl (Siliverstovs & Herzer, 2007). Working at high altitude is considered a condition of risk for developing health problems (Virués-Ortega et al., 2004), reducing productivity and increasing the probability of accidents (Chiu et al., 2014). Some symptoms associated with hypoxia at high altitudes may transiently persist after returning to sea level. Therefore, the negative outcomes related to acute and chronic hypoxia are of relevance to public health (Hu et al., 2017). The search for biomarkers of health impairments constitutes an avenue for developing treatments, therapies, and devices to treat the health and cognitive problems related to high altitude.

Acute exposure to changes in oxygen—such as when quickly blowing up balloons or climbing high mountains without preparation—can cause a series of neurological symptoms that resemble known hallmarks of aging and neurodegenerative diseases, including changes in the autonomic system, cognitive control, memory, and executive functions (Asmaro et al., 2013; de Aquino Lemos et al., 2012; Malle et al., 2016; Ochi et al., 2018). An example of this state is high-altitude hypoxia, understood as the arterial partial pressure of oxygen (PaO2) triggered by the decline of air oxygen pressure due to altitude, which occurs when subjects are acutely exposed to an altitude above 2,500 masl without preparation (Virués-Ortega et al., 2004). Interestingly, the severity of these symptoms can increase depending on prior acclimatization and the physical activity undertaken by the subject, which suggests that acutely perturbing oxygen supply may generate brain dysfunction due to the brain’s slow adaptation to the increase in energetic demands (Petrassi et al., 2012).

The reversible functional brain alterations triggered by high-altitude hypoxia may be used to understand the functional disturbances associated with healthy aging. Brain function slowly adapts to the availability of energetic supply, which is affected by internal processes such as metabolic aging, as well as by external resources such as oxygen availability. Although brain aging has been thought to affect various functional measures of brain activity, delineating the energetic component from the different sources, for example, neurodegeneration, that occur during aging is essential to the further understanding of healthy and pathological aging.

Hypoxia can disrupt different measures of brain function, observed at different spatial and temporal scales. Evidence from electroencephalography (EEG) has shown an increase in low-frequency oscillations accompanied by a decrease in overall EEG power (Babiloni et al., 2014), which has been confirmed under both artificial hypoxia via the use of a hypobaric chamber as well as altitude-induced hypoxia (Babiloni et al., 2014; Ozaki et al., 1995; Zhao et al., 2016), with recovery of the EEG measures related to chronic acclimatization (Zhao et al., 2016).

The perturbation induced by hypoxia may trigger a cascade of metabolic consequences beyond that which can be characterized by local oscillatory power. Indeed, brain signals change in relation to their state, affecting how information is processed and distributed throughout the brain. Background signals can be addressed by the 1/f slope of the spectrum—a measure of the aperiodicity of the signal (Gao et al., 2017; Lendner et al., 2020; Medel et al., 2023; Trakoshis et al., 2020)—while the regional interactions can be characterized by functional connectivity (FC), which attempts to capture shared functional properties (statistical interdependencies) in the brain. Although the exact mechanism underlying these network-state measures remains unclear, strong evidence has shown that aperiodicity (Martínez-Cañada et al., 2023; Voytek et al., 2015; Vyšata et al., 2014) and FC (Gatica et al., 2021, 2022; King et al., 2018; Song et al., 2014) are strongly affected by aging and neurodegenerative diseases (e.g., Alzheimer’s disease).

The present study aims to evaluate the utility of high-altitude hypoxia as a model for investigating disrupted brain function similar to that observed in brain aging and neurodegenerative conditions (Babiloni et al., 2014; King et al., 2018; Martínez-Cañada et al., 2023; Ozaki et al., 1995; Song et al., 2014; Voytek et al., 2015; Vyšata et al., 2014; Zhao et al., 2016). We assessed brain function from measures including oscillatory power, aperiodicity (absence of regular patterns in the signal), and FC in resting-state EEG (RS-EEG) of 16 healthy, low-altitude inhabitants recorded at both sea level and 4,000 masl (acute high-altitude hypoxia). We hypothesize that reduced oxygen levels at high altitudes cause functional changes similar to those found in both aging and Alzheimer’s disease and that this can provide valuable insight with respect to treatment and mitigation of associated deficits.

Sixteen healthy volunteers were recruited for the experiment. This study was a collaboration between the University of Valparaíso (Chile), the Universidad Técnica Federico Santa María (Chile), and CODELCO-Andina, within the framework of the ANID-FONDEF ID16I10322 project (Chile). The age of the participants ranged from 19 to 45 years (34 ± 7.9 years). Exclusion criteria included nicotine consumption (smoking), diabetes, and any cardiovascular, pulmonary, or neurological condition (including epilepsy, depression, schizophrenia, or bipolar disorder). Conditions were counterbalanced. Measurements were first obtained from 10 subjects at Codelco-Andina (4,000 masl), and then at the Engineering Faculty of the University of Valparaíso (sea level). For the remaining six subjects, recordings were acquired first at sea level and then at Cristo Redentor (4,000 masl). The study was approved by The University Research Ethics Committee of The University of Manchester, UK (ref. 14374), and was conducted according to the Declaration of Helsinki.

The measurements were taken between 1–3 hours after arriving at high altitude from sea level. In general, when arriving from sea level, proper adjustment to 4,000 masl takes several days, and requires adequate sleep (Muza et al., 2010). We recorded 10 min (10.3 ± 0.76 min) of eyes-open resting-state EEG (RS-EEG) from 16 healthy participants (Biosemi Active II, 64 channels, sampling rate of 4096 Hz). In addition, four external electrodes were used to record blinking and ocular movements. RS-EEG data was cleaned in MATLAB (MathWorks Natick, MA) using the EEGLab toolbox (Delorme & Makeig, 2004) and custom routines. Visual inspection was performed to identify and remove channels or data blocks containing large artifacts (noisy EEG segments). Recordings were downsampled to 2048 Hz and band-pass filtered between 0.01 and 120 Hz in two stages of FIR filtering. Independent component analysis was used to decompose the data (JADER algorithm, 30 principal components). Components resembling stereotyped artifacts, such as blinks, eye movement, cardiac signal, and muscle activity, were identified and rejected based on their temporal profile, spectral signature, and topography. Then, the signal was reconstructed with the remaining components. Re-reference was performed using the average across electrodes.

Power spectral density (PSD) was estimated for each channel using the Welch method: Hanning windows of 4 seconds in length (allowing a resolution of 0.25 Hz) and 50% overlap (Welch, 1967). The absolute power within each band was calculated by integrating the PSD functions in the frequency range of interest. We reported the relative power with respect to the broadband spectrum (0.5–30 Hz).

The aperiodic 1/f component of the EEG signals was analyzed using the FOOOF algorithm, as described by Donoghue et al. (2020). FOOOF is an open-source Python package that estimates the aperiodic components of neural power spectra using a combination of Gaussian fitting and background frequency scaling. Thus, it separates oscillatory from background aperiodic activity. The FOOOF analysis was performed on the PSD for each participant. We accounted for aperiodic activity in the frequency range of 4–40 Hz. From the model, we here report the spectral exponent.

Finally, for FC analysis, signals were band-pass filtered with a Bessel 3rd order filter in the common EEG frequency bands δ (0.5–4 Hz), θ (4–8 Hz), α (8–13 Hz), and β (13–30 Hz).

Starting from the FC matrices, we first applied a proportional threshold to remove spurious connectivity values (van den Heuvel et al., 2017). FC matrices were subsequently binarized and graph measures were computed. We employed global efficiency and transitivity, which are metrics related to integration and segregation, respectively (Rubinov & Sporns, 2010).

To avoid the arbitrariness of choosing a single proportional threshold, we used a range of thresholds from 5% to 40% with linear increments of 1%. Proportional thresholding was used to ensure that all networks have the same number of connections across different thresholds, considering that graph metrics are very sensitive to network density (van den Heuvel et al., 2017). FC matrices were then binarized after thresholding, to divorce possible alterations of EEG network topology from overall FC (the average strength of FC) triggered by altitude (van den Heuvel et al., 2017). We reported the area under the curve (AUC) of each graph metric as a function of the threshold (Ginestet et al., 2011).

All group comparisons, such as relative power, overall FC, and network metrics, were performed with nonparametric permutation tests. These tests are suitable when using small sample sizes, and do not require any assumptions regarding normality (Nichols & Holmes, 2002). The real difference between groups (computed as the mean difference) was compared with the distribution obtained from 10,000 random surrogates; they were acquired by randomly reassigning the measures between groups (sea level and altitude). Results were considered statistically significant for p values < 0.05. In addition to the permutation tests, we reported the Cohen’s D effect size. Correlations between age and EEG measurements were computed with Pearson’s correlation. To analyze the possible confounding effects of age on the computed metrics, we repeated the principal analysis of each subsection, but first partialling out the effect of age from measurements. We compared both conditions, that is, sea level and altitude, using the residuals from the regression with age. All p values were corrected for multiple comparisons with the Benjamini-Hochberg method (Benjamini & Hochberg, 1995), to decrease the probability of making a type I error (false positives).

To test if high-altitude hypoxia affected brain function, we first analyzed the effect of high-altitude conditions on the power spectrum of the EEG. We calculated the relative power in each frequency band. Results are presented in Figure 1, where we observed an increment of the δ band relative power in high altitude (p = 0.049, D = 0.277), and a reduction of the α band relative power (p = 0.003, D = −0.285), considering the average power across electrodes (Figure 1A). At a more local scale, we found an increase, by high altitude, of the δ band relative power in occipitoparietal scalp regions (p = 0.011, D = 0.385), and a decrease of α band relative power in occipitoparietal (p = 0.001, D = −0.336) and frontocentral (p = 0.006, D = −0.357) sensors (Figure 1B). No differences were found for α and β bands. Thus, results suggest that altitude-induced hypoxia concentrates the relative power in the slower bands of the EEG spectrum. Furthermore, changes in the α and δ bands are localized at both the posterior and anterior and only posterior scalp sensors, respectively.

Next, we calculated the aperiodic component of the PSD. Spectral parametrization showed a global decrease of 1/f slope driven by high-altitude hypoxia (Figure 1C, p = 0.0001, D = −0.42). When grouping channels across topographic regions, we found that the effect of high altitude on 1/f slope was significant for frontal (p = 0.012, D = −0.483), frontocentral (p = 0.0001, D = −0.700), centroparietal (p = 0.0001, D = −0.665), and temporal scalp regions (p = 0.0001, D = −0.587), but not for occipitoparietal sensors (p = 0.222, D = −0.222), similar to the topographic distribution of the effect of aging in aperiodic activity recently reported (Merkin et al., 2023).

To explore if high-altitude hypoxia produced altered brain patterns on functional networks, we computed the FC matrices in each frequency band using wPLI, firstly characterizing FC overall strength and then network topology. The results, presented in Figure 2, showed an increase in the global connectivity, named overall FC (Figure 2A) in the δ band (p = 0.048, D = 0.653), and a trend to be higher in the β band (p = 0.075, D = 0.644). At the local level, we computed the difference—altitude minus sea level—in nodal strength (defined as the weighted degree), using the weighted and nonthresholded FC matrices. We found that δ nodal strength was greater at altitude in temporal (p = 0.039, D = 0.805) and frontocentral scalp regions (p = 0.039, D = 0.946). No differences were found for the remaining frequency bands.

Qualitatively, the global increase in overall FC can be observed by visual inspection of the FC matrices at sea level and 4,000 masl (high altitude) (Figure 2C), where the contrast between the δ band FCs can be observed. This overall increase of δ band FC strength has been observed in several conditions, such as deep sleep (Imperatori et al., 2019) and anesthesia (Lee et al., 2017), where arousal levels abruptly decreased.

We studied how high-altitude hypoxia affects the functional network topology of the thresholded and binarized FC matrices. Specifically, we computed the global efficiency (integration) and transitivity (segregation) of the thresholded networks. We computed the AUC values for the range of thresholds employed (from 5% to 40%).

With respect to integration, we observed an increase of global efficiency in the θ band only at altitude versus sea level (p = 0.013, D = 0.594) (Figure 3A). Although we did not find differences in the δ, α, and β bands, it is possible that the changes in functional network topology are more localized rather than global. Consistent with this, we found regional changes in the δ band, where nodal efficiency decreased by altitude in the occipitoparietal sensors (p = 0.045, D = −0.870). In the θ band, nodal efficiency was higher in the centroparietal (p = 0.048, D = 0.618) regions. We did not find any local changes in the α and β bands. The local difference (altitude minus seal level) in nodal efficiency is presented in the head maps of Figure 3B.

Regarding segregation, we did not find any change in transitivity by altitude in all frequency bands (Figure 4A). However, we repeated the local network analysis to calculate the nodal clustering coefficient. In the θ band, high-altitude conditions produced an increase in the nodal clustering coefficient in the temporal scalp regions (p = 0.007, D = 0.529). In the β band, an increase in clustering triggered by altitude was found in occipital regions (p = 0.0001, D = 1.156). No local differences were observed in the δ and α bands. The local difference (altitude minus seal level) in nodal clustering is presented in the topoplots of Figure 4B.

In the previous sections, we reported how high-altitude hypoxia impacts normal brain activity and connectivity. We proposed that these changes might be analogous with aging and neurodegeneration. To connect hypoxia-induced alterations more directly with brain aging, we correlated subjects’ chronological age with our EEG data measurements. To minimize the risk of multiple comparisons and consequently reduce potential false positives, we concentrated on the primary findings within each subsection, specifically: α band relative power, 1/f slope, δ band overall FC, and θ band integration (Figure 5A). Pooling data from both sea level and altitude conditions, we first identified a negative correlation between the 1/f slope and age (r = −0.40, p = 0.046), aligning with previously reported findings in the field (Merkin et al., 2023). Conversely, θ band global efficiency (integration) showed a positive correlation with age (r = 0.41, p = 0.046). These results imply that both the 1/f slope and θ band integration effectively capture changes characteristic of healthy aging, potentially indicating brain activity marked by increased electrophysiological noise and enhanced global functional integration.

It is possible that the changes ascribed to hypoxia can be influenced by participants’ age. To separate the effects of age from the EEG measurements, we regressed out the effect of age. (Figure 5B). We found that the results remained robust when controlling for age, observing a decrease in the α band relative power (p = 0.001, D = −0.289), a reduction of the 1/f slope (p = 0.0001, D = −0.638), an increase in δ band overall FC (p = 0.038, D = 0.658) and in θ band global efficiency (p = 0.017, D = 0.658) at high altitude compared to sea level. As such, the alterations triggered by hypoxia not only are observed when controlling for chronological age, but also followed the same direction as the effects observed in aging.

The brain requires an ongoing and stable oxygen supply to support its underlying functions. The role of oxygen in the functional activity of the brain and its relationship with aging and its pathologies has been recently highlighted (Li et al., 2016). Metabolism is known to modulate brain function and neural activity (Medel et al., 2022), especially during aging (Deery et al., 2023), and metabolic biomarkers are widely used to diagnose age-related brain pathologies (Chételat et al., 2020). We extended this knowledge by isolating the effect of oxygen supply on brain function, thus controlling for other processes involved in brain development and neurodegeneration. Our results suggest that canonical age-related effects on functional EEG activity can be simulated by inducing hypoxia (in this case using high altitude; 4,000 masl).

We found that at high altitude the relative power of the slower oscillations were increased, while α band relative power decreased compared to sea level. Convergently, we observed a widespread reduction of 1/f slope (flattened spectrum) at altitude, which suggests that hypoxia increases neural (electrophysiological) noise (Voytek et al., 2015). Thus, our spectral analysis suggests a shift to slower and more noisy brain activity at high altitude.

With respect to functional connectivity, we found an increase in overall FC at high altitude (global frequency synchronization) in δ band. Regarding topology, global integration was higher in high altitude in the θ band compared to sea level. The results suggest a shift to a more integrated functional network topology at high altitude. At the local level, changes in nodal properties were region-specific: occipital sensors showed a decrease in their integration, with the opposite observed for central sensors. Finally, we found a correlation between chronological age, 1/f slope and θ band functional integration, demonstrating the potential parallels between hypoxia and brain aging. Together these results indicate that high-altitude hypoxia affects specific EEG bands and network metrics, suggesting potential biomarkers to identify and study the effects of high-altitude hypoxia on the brain. Furthermore, changes seen in brain activity due to hypoxia were reminiscent of those observed across age, which suggests that hypoxia may serve as a model for aging or neurodegeneration.

Our results are consistent with previous reports (Babiloni et al., 2014) in which an increase in the δ and θ relative power was triggered by hypoxia, disrupting the normal α rhythms observed in RS-EEG, especially in the occipitoparietal scalp regions (Lozano-Soldevilla, 2018; Mandal et al., 2018). High-altitude hypoxia might produce deficits in vasomotor reactivity and neurovascular coupling, decoupling thalamocortical networks, and ultimately generating the slow rhythms observed by high altitude (Babiloni et al., 2014). Similar effects are observed as a result of GABAergic anesthetics (Ching et al., 2010; Purdon et al., 2013) and dissociative drugs (Akeju et al., 2016; Vlisides et al., 2017); whereas pro-cholinergic drugs, usually associated with increased attention and arousal (Coronel-Oliveros et al., 2023; Giessing et al., 2013; Hasselmo & Sarter, 2011; Honey et al., 2017), produce an opposing profile of effects (see Lozano-Soldevilla, 2018). Furthermore, the spectral analysis revealed an increase of the 1/f slope at altitude compared to sea level. In line with previous works in aging and Alzheimer’s disease (Martínez-Cañada et al., 2023; Voytek et al., 2015; Vyšata et al., 2014), this suggests a shift toward higher electrophysiological noise by high-altitude hypoxia. This increase in neural noise is linked to desynchronized spiking activity, disrupting network communication, and ultimately producing cognitive deficits (Voytek & Knight, 2015). Furthermore, the higher 1/f slope generated by high altitude was stronger in fronto-parietal scalp regions. These results are consistent both with the cognitive deficits triggered by high altitude (Asmaro et al., 2013; de Aquino Lemos et al., 2012; Malle et al., 2016) and the role of the fronto-parietal network in supporting attention and executive functions (Marek & Dosenbach, 2018).

Network analysis at the global level revealed FC alterations triggered by hypoxia. Firstly, using wPLI, similar results were reported for different states of arousal and consciousness increases in δ median wPLI (overall FC), and a decrease in α median wPLI in N3 sleep (Imperatori et al., 2019) and propofol anesthesia (Lee et al., 2017). High-altitude hypoxia also produced a global reconfiguration of brain networks, reflected in the shift to functional integration produced within the θ band. Although the brain “integrates” functions during some types of behavioral tasks, usually in tasks with a high cognitive load (Shine & Poldrack, 2018), this may be undesirable in resting-state conditions (Wig, 2017). Some examples where a reduction of RS segregation (or an increase in integration) have been linked to deficits or impairments in brain functions include childhood-onset schizophrenia (Alexander-Bloch et al., 2010), cognitive decline in healthy aging (Pedersen et al., 2021; Song et al., 2014), Alzheimer’s disease, mild cognitive impairment (Brier et al., 2014; Franciotti et al., 2019; Sanz-Arigita et al., 2010; Seo et al., 2013), and obstructive sleep apnea (Luo et al., 2015). Alzheimer’s disease and obstructive sleep apnea are of particular interest, as the slowing of RS-EEG rhythms observed during hypoxia were also observed in these cases (Babiloni et al., 2014). Furthermore, they are associated with impairments in vasomotor reactivity and neurovascular coupling (Li et al., 2020; Smoliński & Członkowska, 2016), as observed in acute mountain sickness produced by high-altitude conditions (Jansen et al., 1999).

At the local level, differences in functional network changes were observed between high altitude and sea level. The local increase of segregation (or the reduced integration) in temporal and occipital scalp regions may reflect a form of “sensory disconnection” within the brain, considering the role of highly integrated nodes, as the “rich club” brain regions, in supporting long-range network connectivity (Faber et al., 2019). On the other hand, both the increment of δ overall FC in frontal and central scalp regions, and the increase of integration (θ band) in centro-parietal regions suggest a strengthening of frontocentral connectivity. It is hypothesized that the faster oscillations are involved in local connectivity, and the slower ones in long-range connectivity (Harmony, 2013; Niknazar et al., 2022). Interestingly, some of the changes in functional integration (increased global efficiency and reduced occipitoparietal nodal efficiecy) were observed in the slowest frequency bands (δ and θ), while the increase in occipitoparietal segregation occurred in the faster β band, suggesting that the band-specific changes caused by altitude in network topology and FC respond to different spatial scales of signal processing.

The comparable EEG profiles observed in relation to age and high-altitude conditions underscore the importance of oxygen metabolism in understanding the trajectory of brain health across life-span. Moreover, if the patterns in EEG profiles observed across ages are found to be robust, they may offer potential avenues for early detection of age-related brain alterations. Identifying these shifts may facilitate timely interventions aimed at maintaining or even enhancing cognitive health. Further investigations may focus on understanding the mechanisms behind these parallels between high-altitude hypoxia and aging. Potential questions of interest include whether other factors, such as genetic predispositions or environmental variables modulate the observed effects.

Our study has some limitations that constrain the impact of these findings: Firstly, the influence of high altitude on the EEG functional connectivity could comprise not only the reduction of the oxygen pressure by altitude, but also changes in environmental variables. For example, external temperature may affect EEG spectral properties (de Labra et al., 2021; Gaenshirt et al., 1954; Pearcy & Virtue, 1959). Our results must be contrasted by conditions in which environmental variables can be controlled, as usually done when simulating hypobaric hypoxia.

Secondly, it is worth noting that the findings presented here are based on a limited sample size. Future research with larger cohorts is crucial to validate and extend these preliminary observations. Despite this limitation, the study lays the groundwork for more extensive investigations into the intricate relationships between brain activity, aging, and environmental stressors such as high-altitude hypoxia.

Since the computation of EEG spectral features and FC was limited to the sensor space, our analyses precluded direct associations between EEG features and brain regions. However, it has been suggested that networks built from the sensors space can capture the rich structure of brain interactions, considering the effects of volume conduction and even when controlled for them (Schaworonkow & Nikulin, 2022). Furthermore, the decision of estimating EEG connectivity at either the sensors or source level remains arbitrary unless specific hypotheses on regions-of-interest are proposed (Lai et al., 2018). Inverse methods for source reconstruction can introduce variability in network and connectivity estimates (Mahjoory et al., 2017), and sensor-level topological properties have been reported as highly consistent, robust, and stable in the long term (Chu et al., 2012). Finally, it is essential to note that EEG connectivity highly relies on the connectivity metric selected for the analysis (Perera et al., 2020; Prado et al., 2022; Wang et al., 2014). In this study, the reason for using wPLI was to discharge spurious short-range functional connections that can be obtained in high-density electrode layouts due to head volume conduction. Future studies can implement EEG connectivity metrics that can detect physiological close-to-zero lag interactions while reducing the impact of head volume conduction (Sanchez Bornot et al., 2018). Alternatively, future work can employ composite metrics of connectivity, that is, machine-learning approaches that integrate information provided by different connectivity metrics, which has been successful in estimating brain aging (Mohanty et al., 2020) and classifying neurodegenerative diseases (Prado et al., 2023a, 2023b).

In summary, this work suggests that alterations in brain metabolism via acutely manipulating oxygen supply through high-altitude hypoxia systematically affect brain function and its organization. The alterations triggered by high altitude may be utilized as biomarkers related to hypoxia. Therefore, such findings may help to advise and evidence treatments and therapies that aim to recover normal brain function. Furthermore, our results suggest that altered oxygen metabolism may be an underlying mechanism of the functional changes observed in developmental and pathological aging. By comparing conditions in which oxygen availability differs, we show that hypoxia alone is sufficient to alter EEG spectral and network properties of the brain, replicating previous results seen in aging and dementia. Further research should explore whether the reported relation has an impact on cognitive decline and behavior. The functional and cognitive relevance of oxygen supply underscores the need for studies analyzing the role of metabolic decline in brain physiological and pathological processes.

We want to thank all subjects who kindly participated in the experiments.

Carlos Coronel-Oliveros: Formal analysis; Investigation; Methodology; Validation; Visualization; Writing – original draft; Writing – review & editing. Vicente Medel: Formal analysis; Investigation; Methodology; Validation; Writing – original draft; Writing – review & editing. Grace Whitaker: Formal analysis; Methodology; Validation; Visualization; Writing – original draft; Writing – review & editing. Aland Astudillo: Data curation; Investigation; Methodology; Validation; Writing – original draft; Writing – review & editing. David Gallagher: Conceptualization; Investigation; Writing – original draft; Writing – review & editing. Lucía Z-Rivera: Data curation; Methodology; Writing – original draft; Writing – review & editing. Pavel Prado: Conceptualization; Writing – review & editing. Wael El-Deredy: Conceptualization; Funding acquisition; Investigation; Project administration; Resources; Supervision; Writing – review & editing. Patricio Orio: Investigation; Methodology; Resources; Supervision; Writing – review & editing. Alejandro Weinstein: Conceptualization; Funding acquisition; Investigation; Project administration; Resources; Supervision; Writing – review & editing.

Vicente Medel, Agencia Nacional de Investigación y Desarrollo (https://dx.doi.org/10.13039/501100020884), Award ID: ANID/Fondecyt 3230557. Alejandro Weinstein, Agencia Nacional de Investigación y Desarrollo (https://dx.doi.org/10.13039/501100020884), Award ID: ANID/Fondef Id16i10322 and ANID/Fondecyt 1231132. Patricio Orio, Agencia Nacional de Investigación y Desarrollo (https://dx.doi.org/10.13039/501100020884), Award ID: ANID/Fondecyt 1211750.

This work did not produce original codes for data analysis and simulations. Functional connectivity was assessed through MNE for Python (https://github.com/mne-tools; Gramfort et al., 2013). Functional network analysis was performed using the brain connectivity toolbox (https://github.com/fiuneuro/brainconn; Rubinov & Sporns, 2010). For the 1/f analysis, we used the FOOOF library in Python (https://github.com/fooof-tools/fooof; Donoghue et al., 2020). Data is available on request from Dr. Alejandro Weinstein, University of Valparaíso (email: [email protected]). A formal data sharing agreement is required.

One over F (1/f):

Describes the distribution of power (or energy) across different frequencies in the EEG signal. The power at a given frequency is inversely proportional to that frequency.

Functional connectivity:

Statistical relationships between pairs of neurons, brain regions or networks. Usually measured using the time series of brain activity (spikes, EEG, fMRI BOLD) and a measure of brain connectivity (e.g., Pearson’s correlation).

Network:

A mathematical representation of a system of elements. In the case of the brain, neurons and brain areas consist of nodes connected by edges, represented by anatomical or functional links.

Functional connectivity matrix:

A matrix that captures the functional connectivity values between all the possible pairs of regions in a network.

Integration:

Property of a network that allows the communication and information transfer between specialized, distant, and locally connected nodes.

Segregation:

The degree of local connectivity within a network.

Overall FC:

Mean values (the average) of the functional connectivity matrices, generally named global correlations.