Cortical Hyperexcitability Shapes Large-Scale Brain Dynamics and Behavioral Outcome in Angelman Syndrome Authors: Gian Marco Duma, Margherita Bagnoli, Giulia Stefanelli, Camille Mazzara, Giovanni Pellegrino, Giovanni Mento, Pierpaolo Sorrentino, Lisa Toffoli, Fiorella Del Popolo Cristaldi, Lisa Antoniazzi, Rebecca Azzolini, Jacopo Dei Tos, Martina Baggio, Paolo Bonanni, Alberto Danie Published: 2026 ABSTRACT BackgroundAngelman syndrome (AS) is a rare neurodevelopmental disorder with characteristic electroencephalographic abnormalities caused by loss of function of the maternally inherited UBE3A gene. Converging evidence suggests a disrupted excitation-inhibition (E/I) balance toward hyperexcitability. However, noninvasive approaches capable of characterizing intrinsic cortical excitability and its relationship with large-scale brain dynamics in AS are still lacking. We used resting-state electroencephalography (EEG) to derive cortical excitability, testing the hypothesis that altered local E/I balance in AS is associated with instability of large-scale functional brain networks.MethodsWe recorded 7 minutes of task-free high-density EEG in 29 individuals with AS and 36 typically developing control participants. Source-reconstructed cortical activity was used to compute the excitability index (EI), based on mean spatial phase synchronization in the gamma band. Dynamic functional connectivity was computed and summarized as the fluidity index, which estimates temporal variability of network configurations. We assessed group differences and associations between EEG features, clinical variables, and caregiver-reported questionnaires.ResultsParticipants with AS showed increased EI in the anterior cingulate, dorsolateral prefrontal, temporoparietal, and occipital regions. Fluidity was larger in AS across frequency bands, indicating greater network instability. EI positively predicted fluidity in widespread regions in the AS group, whereas the opposite pattern was observed in the control group. Higher EI correlated with fewer antiseizure medications and greater sensory-seeking behavior.ConclusionsAS is characterized by cortical hyperexcitability coupled with unstable large-scale network dynamics. EI provides a biologically meaningful marker linking intrinsic E/I imbalance to behavioral features and treatment-related variables. Angelman syndrome (AS) is a rare neurodevelopmental disorder caused by loss of function of the maternally inherited UBE3A gene, most commonly due to chromosomal deletions at 15q11–q13 (1). UBE3A encodes an E3 ubiquitin ligase that is selectively expressed from the maternal allele in neurons, regulating cellular protein homeostasis and influencing the expression of genes involved in neural development and functioning (1,2). Individuals with AS present with severe intellectual disability, speech impairment, motor disorder, epilepsy, sleep disturbances, and distinctive behavior characterized by a happy disposition, hyperactivity, short attention span, strong drive for social interaction, and sensory-seeking behaviors (3,4). Electroencephalography (EEG) is characterized by abnormalities, particularly high-amplitude, rhythmic delta activity with posterior or anterior predominance, intermixed with epileptiform discharges, which are found in almost all individuals (5,6). Converging preclinical evidence suggests that disruption of the excitation-inhibition (E/I) balance is a core mechanism underlying AS. In murine models, maternal loss of UBE3A has been shown to alter multiple pathways involved in synaptic function and plasticity, including GAT1 (GABA [gamma-aminobutyric acid] transporter 1) degradation, activity of calcium- and voltage-dependent large-conductance potassium channels, and CAMK2 (calcium/calmodulin-dependent protein kinase II) signaling (7, 8, 9, 10). These effects vary across brain regions and neuronal populations, producing distinct neurophysiological and behavioral consequences (11). In particular, UBE3A loss in GABAergic neurons has been shown to induce AS-like EEG abnormalities and increased seizure susceptibility, whereas glutamatergic UBE3A loss primarily drives motor and behavioral abnormalities (12,13). Additionally, 15q11–q13 deletions involve the loss of other genes contributing to AS pathophysiology, including the GABRB3-GABRA5-GABRG3 gene cluster encoding GABAA receptor subunits expressed in cortical and thalamocortical circuits. It has been proposed that the resulting reduction in GABAergic neurotransmission underlies the more severe clinical presentation and some specific EEG features observed in deletion cases compared with other genetic etiologies (14). Together, these mechanisms are thought to shift the E/I balance toward cortical hyperexcitability, linking molecular deficits with the electrophysiological and clinical features of AS. Traditionally, brain excitability in humans has been assessed using perturbational approaches such as transcranial magnetic stimulation, which primarily probe the motor cortex. Recent advances suggest that intrinsic brain activity may also carry signatures of underlying excitability (15,16). In particular, features derived from resting-state EEG have been proposed as noninvasive proxies for neuronal excitability, enabling broader spatial coverage and facilitating investigation in populations for whom stimulus-based paradigms are difficult, such as children and individuals with severe neurodevelopmental disorders (15, 16, 17). Among resting-state markers linked to E/I balance, the aperiodic (1/f) exponent of the EEG power spectrum has received considerable attention, with steeper slopes typically being interpreted as reflecting increased inhibition (18, 19, 20, 21). However, recent studies have reported increased aperiodic exponents in AS, suggesting heightened inhibition, an observation that appears inconsistent with evidence of reduced GABAergic signaling in the disorder (22). This discrepancy raises important questions about the specificity and interpretation of current resting-state excitability markers in AS. In the current study, we leveraged a metric derived from intrinsic neural dynamics, grounded in theoretical and empirical work linking spatial signal properties to E/I balance in epilepsy, to estimate cortical excitability in AS (16,23,24). We recorded resting-state brain activity using high-density EEG (hdEEG) (128 channels) in individuals with AS and typically developing control participants. Through source reconstruction, we estimated this excitability index (EI) at the regional level, enabling spatially resolved characterization of cortical E/I balance. We further examined how local excitability relates to large-scale network dynamics. We hypothesized that this EI would provide a sensitive marker of AS-related neurophysiology and better align with known mechanisms of E/I balance disruption. Finally, we explored associations between regional excitability alterations and behavioral measures assessing sensory processing, aberrant behaviors, and social-communicative functioning, aiming to connect neurophysiological mechanisms and clinical variables. Methods and Materials Participants We enrolled patients with a diagnosis of AS who underwent hdEEG between 2020 and 2025 at the Epilepsy and Clinical Neurophysiology Unit, IRCCS Eugenio Medea in Conegliano, Italy. Thirty-five cases were retrospectively screened, and 5 patients were excluded due to poor data quality or missing behavioral questionnaires (n = 1). The final sample consisted of 29 participants (mean age in years ± SD = 12.82 ± 8.05; 15 female, 14 male) with AS (25 with epilepsy). Amongst the 29 participants with AS, 24 showed deletion of the 15q11–q13 region, while the remaining 5 participants showed nondeletional mutation. We also recorded data from a group of healthy control (HC) participants collected at the Department of General Psychology of Padova University. The HC group consisted of 36 participants (age in years = 10.33 ± 2.56; 16 female, 20 male) with no history of neurological or neuropsychiatric disorders. Patients’ demographic and clinical characteristics are provided in Table 1. Using a 2-tailed independent samples t test, we did not find any significant difference in the age of the 2 groups (t62 = 0.964, p = .331). We ran a χ2 test between the 2 groups and did not find any difference in the sex distribution between the AS and control groups (χ21 = 0.095, p = .758).Table 1Sample Demographic and Clinical InformationPatients With Angelman SyndromeMean ± SDAge, Years12.82 ± 8.05Age of Onset, Years2.5 ± 1.47Duration of Epilepsy, Years5.56 ± 7.14Lifetime Number of ASMs3.04 ± 1.88Current Number of ASMs at the EEG Recording1.52 ± 0.92ASM, antiseizure medication; EEG, electroencephalography. Table 1Sample Demographic and Clinical InformationPatients With Angelman SyndromeMean ± SDAge, Years12.82 ± 8.05Age of Onset, Years2.5 ± 1.47Duration of Epilepsy, Years5.56 ± 7.14Lifetime Number of ASMs3.04 ± 1.88Current Number of ASMs at the EEG Recording1.52 ± 0.92ASM, antiseizure medication; EEG, electroencephalography. Sample Demographic and Clinical Information ASM, antiseizure medication; EEG, electroencephalography. The study protocol was conducted according to the Declaration of Helsinki and approved by the ethical committees (Comitato Etico Area Nord Veneto and Comitato etico della ricerca psicologica [area 17], Dipartimenti/Sezione di Psicologia, Università di Padova; Protocol No. 387-a). All participants provided written informed consent to participate in the study. Resting-State EEG Recording Seven minutes of task-free hdEEG data were recorded with an R-Net EEG cap (128 sponge electrodes). Participants were sitting comfortably on a chair in a silent room, looking at the Inscapes video (25). We purposely used the Inscapes video as it is validated for task-free neuroimaging studies, and it represents a source of engagement, which is fundamental in a population with severe neurodevelopmental disorders. The signal was sampled at 500 Hz (Padova) or 512 Hz (Conegliano) and referenced to the vertex. EEG Preprocessing Signal preprocessing was performed in EEGLAB version 14.1.2 (26) according to a pipeline validated in previous works (27,28). The preprocessing pipeline included the following steps: 1) data resample to 250 Hz; 2) bandpass filter (1 Hz–45 Hz) with a Hamming windowed sinc finite impulse response filter; 3) epoching (1-second-long epochs); 4) manual removal of epochs containing interictal epileptiform discharges and automatic removal of artifacted epochs using TBT EEGLAB plug-in; 5) removal of flat channels; 6) independent component analysis (40 independent components decomposition) (29), using the Infomax algorithm (30) as implemented in EEGLAB; 7) rejection of independent components related to eyes, muscle, channel noise, and heart artifacts; 8) channel interpolation with the spherical spline interpolation method (31); 9) rereference to average; and 10) reconcatenation of the epochs obtaining a continuous signal. For the patient group, we rejected an average ± SD of 143.18 ± 58.82 epochs (average of 34.04% of rejected epochs) and an average of 23.52 ± 4.17 components (average of 58.8% of rejected components). For the control group, we rejected an average of 2.43 ± 2.79 epochs (average of 0.59% of rejected epochs) and an average of 30.64 ± 4.80 (average of 76.6% of rejected components). Cortical Source Modeling We used publicly available age-specific anatomy templates (32, 33, 34). The magnetic resonance images were segmented with Freesurfer (35) using CBRAIN (36) analytical cluster to obtain skin, skull, and gray matter surfaces. Anatomical data were then imported in Brainstorm (37). The coregistration of the EEG electrodes with brain magnetic resonance imaging was performed using Brainstorm, considering anatomical landmarks. Manual coregistration corrections were applied as needed. The cortical mesh was downsampled at 15,002 vertices, and the 3 boundary element model (BEM) surfaces were reconstructed (inner skull, outer skull, and head). We computed the BEM OpenMEEG plug-in (38,39) implemented in Brainstorm. We used the weighted minimum norm as an inverse solution method, with Brainstorm’s default parameter settings (40). Vertexwise activity was then downsampled to the Desikan-Killiany atlas (41). EI Computation The EI, originally proposed and validated in epileptic populations, quantifies mean spatial phase synchronization, particularly within high-gamma frequency bands (23,24). It showed a strong correlation with cortical excitability as assessed using perturbational methods and as modulated by antiseizure medications (ASMs) in patients with epilepsy (16,23,24). This E/I measure is conceptually related to the widely used phase-locking value, with the key distinction that it is computed across spatial dimensions rather than over time. Following prior recommendations, we focused on the gamma band (30 Hz–45 Hz) (16). Functional Connectivity Dynamic functional connectivity (dFC) was estimated at the source level using lagged coherence (lagCoh). Unlike magnitude-squared coherence, lagCoh removes the instantaneous (zero-phase) component, thereby reducing the effects of volume conduction and common referencing and providing a more robust estimate of genuine time-delayed interactions (42, 43, 44). Connectivity was computed using Brainstorm functions. Connectivity was calculated using a sliding-window approach (8-second window, 90% overlap) in the 5 canonical frequency bands: delta (2 Hz–4 Hz), theta (4 Hz–7 Hz), alpha (8 Hz–12 Hz), beta (13 Hz–30 Hz), and gamma (30 Hz–45 Hz), yielding a connectivity tensor of size N (brain regions) × N × T (number of windows) × F (frequencies). The selected window length provided a compromise between reliable estimation of low-frequency oscillatory interactions, particularly in the delta band, and sensitivity to temporal fluctuations in connectivity patterns. The high overlap was adopted to improve temporal continuity between consecutive connectivity estimates and reduce boundary-related fluctuations, enabling a more stable characterization of dynamic network configurations (28). Fluidity The dFC matrix was obtained by correlating the upper triangular elements of lagCoh matrices from consecutive windows, capturing the temporal evolution of interregional synchronization:(1)dFC(t1,t2)=corr[UpperTri(FC(t1)),UpperTri(FC(t2))](2)dFC(t1,t2)∈RNxN,forMtimewindows Fluidity was defined as the variance of the upper triangular values of the dFC matrix after removing overlapping windows.(3)fluidity=Var[UpperTri(dFC−B)](4)Bij=1if[ti−twin2;ti+twin2]∩[tj−twin2;tj+twin2]≠∅;0otherwisewith the diagonal offset to compute fluidity given by the following formula(5)offset=noverlaptwin−noverlap+1 This metric reflects the temporal variability of FC and was interpreted as an index of the brain’s dynamic adaptability (45,46) (see Figure 1).Figure 1Analysis pipeline for excitability and fluidity. The figure shows the analysis pipeline from source-reconstructed resting-state electroencephalography (EEG) to obtain a local measure of cortical excitability (excitability index) and global measure of network dynamics (fluidity). dFC, dynamic functional connectivity; ROI, region of interest. Analysis pipeline for excitability and fluidity. The figure shows the analysis pipeline from source-reconstructed resting-state electroencephalography (EEG) to obtain a local measure of cortical excitability (excitability index) and global measure of network dynamics (fluidity). dFC, dynamic functional connectivity; ROI, region of interest. Statistics Differences in regional excitability were evaluated using a permutation t test approach (5000 permutations) with false discovery rate (FDR) correction for multiple comparisons across the number of regions. Differences in fluidity were assessed using a generalized linear model (GLM) in R, using gamma family distribution, which best approximated fluidity distribution with the following model: Fluidity ∼ Group × Frequency_band. The relationship between fluidity and EI was investigated iteratively for each brain parcel with a mass-univariate GLM separately for each frequency. Relevantly, we also included covariates in interaction with the local excitability to understand whether the interplay between local brain properties (excitability) and global network organization (fluidity) was mediated by age or sex, using the following model: Fluidity ∼ Region_EI × Group + Region_EI × Age + Region_EI × Sex. p Values of the β of the model for each factor and the interaction effect were corrected with FDR correction. Spearman’s correlations with questionnaires and clinical variables, as well as with age and sex, were computed with FDR correction. Although epilepsy status and genotype may also influence cortical excitability, the limited variability and highly unbalanced distribution of these features (86% with epilepsy, 83% with deletion) in the current cohort prevented reliable covariate or subgroup analyses. Caregiver-Reported Questionnaires To investigate the behavioral profile in AS, we used the following caregiver-reported questionnaires: the Aberrant Behavior Checklist (47), Social Communication Questionnaire (48), and Child Sensory Profile 2 (SP-2) (49) (see Table 2). All details are reported in the Supplement. Results Alterations in Local Excitability and Network Configuration We identified increased values of EI in patients with AS compared with control participants. This effect was expressed bilaterally including fundamental hubs of the default mode network such as the anterior cingulate, prefrontal dorsolateral cortices, and cuneus (pFDR < .05). EI increases were also detected in the temporoparietal junction and temporal gyri (see Figure 2A). When looking at the fluidity index, the GLM analysis showed a significant effect of the group factor (F = 7.710, p = .005) but no effect on the frequency band (F = 0.082, p = .987) or the group × frequency interaction (F = 0.368, p = .937).Figure 2Group differences in excitability index (EI) and fluidity. Panel (A) shows the spatial differences in the comparison of the EI between patients with Angelman syndrome (ANG) and control (CTR) participants. Panel (B) shows the group differences in fluidity across the 5 canonical frequency bands between patients with ANG (pale red) and CTR participants (teal). Group differences in excitability index (EI) and fluidity. Panel (A) shows the spatial differences in the comparison of the EI between patients with Angelman syndrome (ANG) and control (CTR) participants. Panel (B) shows the group differences in fluidity across the 5 canonical frequency bands between patients with ANG (pale red) and CTR participants (teal). When investigating the relationship between local EI values and the global organization of the network dynamics (fluidity) with a mass-univariate GLM approach, we observed a significant interaction between fluidity and excitability. Specifically, after FDR correction, we identified positive significant βs (pFDR < .05) engaging a wide range of brain regions across all the frequency bands excluding gamma. In Figure 3A, we report the results at delta, which is the brain rhythm which alterations characterize AS, and alpha, which represent the dominant rhythm of physiological brain activity during a task-free condition. Results in other frequencies are reported in the Supplement (see Figure S1A). In contrast, we observed a negative relationship between EI values and fluidity at the level of the parahippocampal cortex.Figure 3Mass-univariate generalized linear model effects. (A) Spatial distribution of the main effect of excitability index (EI) in the generalized linear model (GLM). The colored areas represent the brain regions with significant beta values after false discovery rate correction. This panel shows the effect only in delta and alpha, which represent the oscillatory rhythms with specific alterations in patients with Angelman syndrome (ANG) and the dominant rhythm of the brain during resting condition, respectively. (B) Interaction effect of the GLM. The colored brain areas represent the regions with significant interaction effects. The blue regions (negative beta) indicate that the relationship between excitability and fluidity is more positive in the ANG group compared with the control (CTR) group. On the right, the scatter plot and the regression lines show the interaction effect from the left rostral middle frontal area of the Desikan-Killiany atlas (scatter plot: orange = ANG, blue = CTR). The red regions (positive beta values) display an opposite direction of effect, where the EI-fluidity relationship is more positive in the control group. The scatter plot and regression lines in the right lower corner display the interaction effect in the left parahippocampal cortex. Mass-univariate generalized linear model effects. (A) Spatial distribution of the main effect of excitability index (EI) in the generalized linear model (GLM). The colored areas represent the brain regions with significant beta values after false discovery rate correction. This panel shows the effect only in delta and alpha, which represent the oscillatory rhythms with specific alterations in patients with Angelman syndrome (ANG) and the dominant rhythm of the brain during resting condition, respectively. (B) Interaction effect of the GLM. The colored brain areas represent the regions with significant interaction effects. The blue regions (negative beta) indicate that the relationship between excitability and fluidity is more positive in the ANG group compared with the control (CTR) group. On the right, the scatter plot and the regression lines show the interaction effect from the left rostral middle frontal area of the Desikan-Killiany atlas (scatter plot: orange = ANG, blue = CTR). The red regions (positive beta values) display an opposite direction of effect, where the EI-fluidity relationship is more positive in the control group. The scatter plot and regression lines in the right lower corner display the interaction effect in the left parahippocampal cortex. When characterizing the difference between groups by looking at the excitability × group interaction, we observed negative significant βs (pFDR < .05) at the level of the dorsolateral prefrontal and anterior cingulate cortices, precuneus, and temporoparietal junction (see Figure 3B). When looking at the slopes, we observed that the 2 groups were characterized by an opposite effect. In patients with AS, larger EI values are associated with larger fluidity, while the opposite pattern is present in the control participants (see Figure 3B). These effects are present across multiple frequencies. In Figure 3B, we report results in alpha bands. The remaining frequencies are reported in the Supplement (see Figure S1B). Finally, the parahippocampal cortex showed an opposite effect, with larger fluidity associated with lower EI values in AS (see Figure 3B). No significant effect of age or sex was detected in modulating the EI-fluidity interplay after FDR correction. Correlations Between EI and Clinical Variables To investigate the clinical relevance of EI, we examined its association with age, sex, and behavioral and sensory phenotypes assessed through caregiver-reported questionnaires. At first, we did not find any significant correlations of EI or fluidity with age (all psFDR > .05) or sex (all psFDR > .05). Based on the known sensory-seeking profile of individuals with AS, we specifically hypothesized that increased cortical excitability would be associated with altered sensory processing. Consistent with this hypothesis, we observed a significant positive correlation between EI values and sensory-seeking scores of the SP-2. This relationship was localized to the right anterior cingulate and dorsolateral prefrontal cortices (pFDR < .05), indicating that higher cortical excitability in these regions was associated with increased sensory-seeking behavior in patients with AS (Figure 4C). EI values were also related to clinical epilepsy variables. Specifically, we found a negative correlation between the EI and the number of ASMs taken both across the lifespan and at the time of EEG recording (Figure 4A, B), such that patients treated with a higher number of ASMs exhibited lower EI values.Figure 4Correlations between excitability index (EI) and clinical and behavioral variables. Panels (A) and (B) show the correlation between EI and the total number of antiseizure medications (ASMs) taken during the lifetime and at the moment of the electroencephalography recording. Panel (C) shows the correlation with the seeking scale of Sensory Profile II (SPII). Colored regions indicate the Spearman’s ρ of significant brain areas after false discovery rate correction. Correlations between excitability index (EI) and clinical and behavioral variables. Panels (A) and (B) show the correlation between EI and the total number of antiseizure medications (ASMs) taken during the lifetime and at the moment of the electroencephalography recording. Panel (C) shows the correlation with the seeking scale of Sensory Profile II (SPII). Colored regions indicate the Spearman’s ρ of significant brain areas after false discovery rate correction. Discussion The current study provides novel evidence that AS is characterized by widespread alterations in intrinsic cortical excitability and its coupling with large-scale brain dynamics. By combining a spatially resolved EI with measures of dFC, we were able to link local E/I balance with global network organization and clinically relevant behavioral features. First, we observed a robust bilateral increase in EI in individuals with AS compared with control participants, involving core regions of the default mode network, including the anterior cingulate cortex, dorsolateral prefrontal cortex, cuneus, temporoparietal junction, and temporal gyri. This pattern is consistent with the known pathophysiology of AS, in which synaptic dysfunction and reduced GABAergic signaling are thought to bias neural circuits toward hyperexcitability (7, 8, 9, 10, 11, 12, 13). Notably, EI increases were prominent in higher-order associative cortices, suggesting that E/I imbalance in AS operates at a system level and affects integrative networks. These regions are key hubs for internally oriented cognition, multisensory integration, and cognitive control (50, 51, 52). Therefore, altered excitability within this distributed network may influence processes related to sensory integration, behavioral regulation, and information coordination, with potential widespread effects on large-scale brain functioning (53). At the global level, individuals with AS showed consistently higher brain fluidity across frequency bands, indicating a more variable and less stable configuration of large-scale network dynamics. Rather than reflecting enhanced flexibility per se, this pattern may indicate a system that shifts more frequently between transient states, potentially as a consequence of altered E/I balance. Increased excitability may render neural activity less constrained, promoting more frequent reconfiguration of FC. Such instability is consistent with the clinical phenotype of AS, characterized by fluctuations in attention, behavior, and sensory responsivity (54). Importantly, our results revealed a clear relationship between local excitability and global fluidity: Across widespread cortical regions, higher EI values were associated with greater network variability, suggesting that local E/I properties play a key role in shaping the stability of large-scale brain organization (55,56). A key finding emerged from the excitability × group interaction. In associative hubs, including the dorsolateral prefrontal cortex, anterior cingulate cortex, precuneus, and temporoparietal junction (57), we observed opposite coupling patterns in patients with AS and control participants. In patients with AS, higher excitability was associated with increased fluidity, whereas in control participants, it tended to relate to more stable network configurations. These regions support cognitive control, salience processing, multisensory integration, and internally oriented cognition and act as convergence points for long-range communication (57,58). The divergent coupling suggests that in the typical brain, local excitability within these hubs may contribute to stabilizing coordinated activity (59,60), whereas in AS—where baseline excitability is already elevated—further increases may push the system toward a more unstable regime, contributing to widespread variability in network dynamics. Relevantly, we did not observe effects of age or sex in mediating the interplay between local excitability and global network organization (fluidity), suggesting that this interaction is a stable mechanism of brain functioning, which is modulated however by the clinical condition. Nonetheless, given the wide age range in the AS group and the limited power of the current sample to detect small or nonmonotonic age effects, a residual developmental influence on cortical excitability/fluidity in interaction with the group cannot be fully excluded. An additional noteworthy finding concerns the inverse relationship observed between excitability and fluidity in the parahippocampal cortex, where higher EI was associated with lower network variability in AS. In the typical brain, this region may facilitate transitions between large-scale network states through its connections with hippocampal and retrosplenial structures and its role in coordinating slow oscillatory rhythms involved in memory-related processing and large-scale temporal organization of neural activity (61,62). In AS, where the broader cortical system already operates in a hyperexcitable and dynamically unstable regime, the functional role of the parahippocampal cortex may shift. Therefore, the reversal of the EI-fluidity relationship in this region may reflect a compensatory or stabilizing mechanism within medial temporal structures. However, this interpretation should be confirmed by future studies specifically targeting structural and functional alterations of the medial temporal lobe in AS. The clinical relevance of the EI metric is further supported by its relationship with pharmacological and behavioral variables. We found a negative correlation between EI and the number of ASMs, both lifetime and current. This finding is consistent with the pharmacodynamic effects of ASMs, many of which enhance inhibitory neurotransmission or reduce neuronal excitability (63). The association suggests that EI may be sensitive to treatment-related modulation of cortical excitability, reinforcing its validity as a physiologically meaningful index. In addition, we observed a positive correlation between EI values in the right dorsolateral prefrontal and anterior cingulate cortices and the seeking behavior dimension of the SP-2. Individuals with higher excitability in these regions displayed stronger sensory-seeking tendencies. These areas are involved in motivational regulation, salience attribution, and goal-directed exploration (64,65), and increased excitability within these circuits may heighten responsiveness to sensory input or drive increased engagement with environmental stimuli. In AS, sensory-seeking behaviors may represent an adaptive strategy for interacting with the environment in the context of severe communication and cognitive limitations, providing alternative channels for regulating arousal and expressing needs. Our results should be interpreted with caution due to the relatively small sample size and the acquisition of the control and AS datasets at different sites. Because recording site and diagnostic group are fully confounded in the current design, the observed group differences may, to some extent, reflect site-related variability in addition to true diagnostic effects. However, the cohorts were recorded using the same hdEEG system and electrode configuration and were processed using an identical preprocessing and analysis pipeline, including harmonization of sampling frequencies prior to analysis. These methodological consistencies likely reduced potential site-related variability, although they cannot completely eliminate it. Relatedly, the 2 groups differed in the proportion of epochs rejected during preprocessing, reflecting the greater challenges of acquiring artifact-free EEG recordings in this clinical population. Although the excitability and fluidity indices were computed exclusively from retained, artifact-free data, the higher rejection rate might have resulted in differences in the quantity and characteristics of the data available for source modeling (e.g., signal-to-noise ratio or residual physiological artifacts). While we cannot completely exclude a contribution of these factors to the observed group differences, the use of a standardized preprocessing pipeline and strict artifact rejection criteria was intended to minimize their impact. In addition, the clinical composition of the cohort closely reflected the known epidemiology of AS, with most participants presenting with epilepsy and the 15q11–q13 deletion genotype. While this supports the representativeness of the sample, the small number of participants without epilepsy and participants without the deletion prevented reliable subgroup analyses. Future multicenter studies with larger and more balanced cohorts will therefore be important to further validate these findings and explore potential genotype- and epilepsy-related differences in cortical excitability and network dynamics. At the same time, an important strength of the current study lies in the feasibility of acquiring high-density EEG recordings in this population. The use of sponge-based electrode systems allowed rapid and well-tolerated cap application, minimizing discomfort and preparation time. Moreover, the combination of high-density EEG with the Inscapes naturalistic video (25) paradigm proved highly manageable for individuals with AS, who were able to tolerate the recording session and remain engaged with the visual stimulus. This setup represents a particularly promising approach for neurophysiological investigations in populations with severe neurodevelopmental disorders, where compliance with traditional experimental paradigms is often challenging. Beyond research applications, this methodology may therefore offer valuable opportunities for longitudinal monitoring, biomarker development, and the evaluation of treatment effects in clinical settings. Conclusions These findings highlight the translational potential of excitability and network dynamics metrics derived from resting-state EEG. The sensitivity of EI to pharmacological modulation and its association with behavioral features suggest that it may represent a promising biomarker for tracking disease mechanisms and monitoring therapeutic interventions. More broadly, the coupling between local excitability and global network dynamics provides insight into a key dimension of brain organization altered in AS, linking physiological alterations with large-scale brain functioning and behavior. Longitudinal studies will be important to determine whether these measures can inform clinical trajectories and support future therapeutic trials.Table 2Descriptive Statistics for All Subscales of the Parent-Reported QuestionnairesTestScores, Mean ± SDABC Irritability6.72 ± 6.81 Letargy/withdrawal3.69 ± 4.58 Stereotypy5.38 ± 5.08 Hyperactivity15.41 ± 10.48SCQ11.52 ± 5.34SP-2 Auditory processing14.86 ± 6.83 Visual processing11 ± 5.63 Touch processing19.93 ± 9.74 Movement processing14.97 ± 7.95 Body position processing17.17 ± 10.07 Oral sensory processing20.34 ± 7.83 Conduct15.93 ± 8.26 Socioemotional responses19 ± 12.13 Attentional responses27.41 ± 9.07 Seeking42.43 ± 15.41 Avoiding30.36 ± 12.76 Sensitivity36.36 ± 12.01 Registration47.89 ± 17.58ABC, Aberrant Behavior Checklist; SCQ, Social Communication Questionnaire; SP-2, Sensory Profile 2. Table 2Descriptive Statistics for All Subscales of the Parent-Reported QuestionnairesTestScores, Mean ± SDABC Irritability6.72 ± 6.81 Letargy/withdrawal3.69 ± 4.58 Stereotypy5.38 ± 5.08 Hyperactivity15.41 ± 10.48SCQ11.52 ± 5.34SP-2 Auditory processing14.86 ± 6.83 Visual processing11 ± 5.63 Touch processing19.93 ± 9.74 Movement processing14.97 ± 7.95 Body position processing17.17 ± 10.07 Oral sensory processing20.34 ± 7.83 Conduct15.93 ± 8.26 Socioemotional responses19 ± 12.13 Attentional responses27.41 ± 9.07 Seeking42.43 ± 15.41 Avoiding30.36 ± 12.76 Sensitivity36.36 ± 12.01 Registration47.89 ± 17.58ABC, Aberrant Behavior Checklist; SCQ, Social Communication Questionnaire; SP-2, Sensory Profile 2. Descriptive Statistics for All Subscales of the Parent-Reported Questionnaires ABC, Aberrant Behavior Checklist; SCQ, Social Communication Questionnaire; SP-2, Sensory Profile 2.