No EEG Marker Family Reliably Separates Trauma-Exposed From Control Groups; Aperiodic Measures Have the Best Claim but the Weakest Direct Evidence

Question: Which resting-state EEG markers—spectral power, functional connectivity, or aperiodic/exponent measures—most reliably distinguish trauma-exposed populations from controls in large-scale studies?

Answer

Taken together, the retrieved papers do not support any of the three families as a reliable discriminator of trauma exposure at scale, and they actively undermine the premise that marker family is the decisive variable. The one direct, adequately sized test in a trauma-exposed population found that no spectral power or connectivity contrast survived correction across 78 comparisons [1], while the largest synthesis of adversity–EEG power associations reports a robust but very small effect (r = −0.086) that is not moderated by frequency band [13]. Aperiodic parameters have the strongest claim on trait-likeness—they are the only family that survived subject-level aggregation in a 1,404-recording case-control corpus [10] and show fair-to-excellent five-year reliability [9]—but the trauma-specific aperiodic evidence comes from small group comparisons without reported discrimination metrics [2]. Connectivity is the most variable family: strongly discriminative in chronic mild TBI [7], null in conflict exposure [1], and in trauma the best-supported connectivity findings are fMRI and track symptom dimensions rather than exposure status [6]. Across these papers, effect-size inflation at small N and unstandardised analytic choices explain more of the between-study variance than the choice of marker [16].

Direct case-control tests of spectral power and EEG connectivity in trauma-exposed adults return nulls after correction

In 122 Colombian adults (74 exposed to armed conflict, 48 controls) with 240 s of eyes-closed EEG, occipital and global alpha power was lower in exposed participants but adjusted estimates were close to the null with broad confidence intervals, temporal theta relative power did not survive Holm correction, and no spectral power or regional connectivity contrast remained significant after correction across 78 tests [1]. The authors frame this as evidence that large or widespread resting alterations are not detectable with that sample and analytical framework, and explicitly position candidate signals as priorities for larger studies rather than as findings [1]. This is the only retrieved study that tests both power and connectivity in the same trauma-exposed cohort with a correction burden stated up front.

Spectral power effects on adversity are real but too small and too band-general to serve as discriminator

The meta-analytic association between early adversity and resting-state EEG power across 19 papers (N = 2,669; 228 effect sizes) is r = −0.086, surviving sensitivity analyses and publication-bias correction and comparable for correlational and randomised controlled trial designs [13]. Critically, adversity type, frequency band, resting-state procedure, developmental stage at EEG, and study design did not moderate the association, while absolute versus relative power did [13]—so band selection buys little, whereas the arithmetic of the power measure matters. Even where power effects are meta-analytically confirmed—increased beta in depression across 52 studies—risk of bias is high and peak-frequency and cordance data were insufficient for interpretation [15]. The frequently assumed "reduced alpha in PTSD" is, in the retrieved literature, a treatment premise rather than a case-control result: alpha downregulation was the target in a 38-patient neurofeedback RCT, and greater alpha suppression tracked symptom improvement [5].

Aperiodic parameters are the only family shown to survive trait-level aggregation, but that evidence is not from trauma

In 1,404 EEG recordings from 96 individuals with epilepsy and 85 controls, aperiodic exponent and offset showed the most robust case-control separation and remained significant after subject-level averaging (exponent median difference 0.20, q = 0.010), whereas entropy and alpha power distinguished groups only at the recording and session levels and did not survive averaging, indicating greater state-dependency [10]. Medication burden attenuated group differences in alpha power and detrended fluctuation analysis but left residual effects in the aperiodic exponent [10]. Parameterised measures are demonstrably stable over long intervals: across two sessions roughly five years apart, aperiodic exponent, aperiodic offset, peak alpha power, and individual alpha peak frequency all showed fair-to-excellent test–retest reliability (ICC = 0.51–0.88), with the exponent flattening and peak frequency decreasing with age [9]. In the one retrieved trauma-specific aperiodic study, adults with high versus low childhood interpersonal trauma differed in resting 1/f slope and in model-derived E/I estimates, consistent with a shift toward increased excitation, and these E/I measures were not correlated with transient affective state, which the authors interpret as enduring, trait-like alteration [2].

Reliability does not imply discriminability, and the exponent's E/I interpretation is contested

Across nine frozen EEG representations, test–retest reliability varied enormously (mean ICC 0.08–0.76; CV 53%) while disease discrimination occupied a far narrower range (AUC CV 5.4%), and alpha-band information contributed disproportionately to reliability while theta ranked first for dementia discrimination [11]. That dissociation undercuts any inference from stability to case-control sensitivity, and it complicates the aperiodic case, whose appeal rests heavily on stability [11] [9]. The E/I reading of the exponent is also bounded: during propofol anaesthesia, spectral band powers continued to track dose at surgery-level depths, but the aperiodic exponent was associated neither with absolute propofol dose (R² = 0.011, p = 0.43) nor with its relative change, implying the exponent–E/I relation is confined to superficial hypnotic depths [8]. Aperiodic activity does outperform oscillatory features as a continuous state index, predicting vigilance more linearly and spatially consistently across three datasets [3]—which is a reason to expect it to index arousal state in trauma cohorts, not necessarily exposure history.

Connectivity findings in trauma index symptom dimensions and injury, not exposure status

In chronic mild TBI (18 cases, 26 controls), spectral power and microstate metrics showed no group differences, but source-reconstructed connectivity was elevated across theta, beta, gamma, and broadband frequencies, with classification AUCs of 0.79–0.89 for injury status and 0.82–0.87 for symptom severity, stable across reduced montages and persisting up to eight years [7]. In trauma-exposed adults, the sharper connectivity result is fMRI-based: shorter DMN–sensory-cortex functional distance was associated with intrusive-memory vividness and intrusiveness, explained unique variance beyond conventional connectivity, and was not associated with clinician-rated symptom severity [6]. Data-driven clustering of resting fMRI connectivity in 77 trauma-exposed individuals separated childhood- from adulthood-onset trauma but produced clusters that did not differ in PTSD diagnosis, trauma type, or cumulative exposure [4]. Connectivity, in other words, is doing symptom-and-mechanism work rather than group-discrimination work in this literature.

Sample size and analytic freedom, not marker family, dominate the apparent reliability ranking

In a multisite resting-state dataset of 2,874 participants aged 5–18, repeated case-control comparisons across subsampled sizes showed that small samples produced inflated and highly variable effect sizes, whereas larger samples converged on uniformly small but robust effects, with false positive rates remaining relatively stable as power rose steeply—leading the authors to challenge conventional case-control designs for psychiatric EEG biomarker discovery [16]. A review of 66 resting-state EEG biomarker studies in depression found large variability in populations, diagnostic procedures, and medication status, with these factors frequently omitted from analyses even though studies that include them tend to find effects [14]. Where phenotypes are stratified biologically rather than diagnostically, effects are far larger: rsEEG frequency bands showed good-to-excellent stability and distinguished psychosis biotypes with large effect sizes across 1,401 patients and 750 healthy participants, while showing no differences between DSM psychosis groups [17].

Conflicts and gaps

The retrieved papers disagree on connectivity: elevated theta/beta/gamma connectivity discriminated chronic mTBI from controls with AUCs near 0.8–0.9 [7], whereas no regional connectivity contrast survived correction in conflict-exposed adults [1]. They also disagree on whether stability is informative: aperiodic parameters are both the most stable over five years [9] and the most robust to subject-level averaging [10], yet reliability and discrimination are dissociable across representations [11]. Most consequentially, the retrieved set contains no large-sample aperiodic case-control study in a trauma-exposed population: the trauma-specific exponent evidence is a group comparison without a reported effect size, AUC, or sample size [2], and the large-N evidence for the exponent comes from epilepsy [10]. Whether the exponent's advantage generalises to trauma is therefore untested. Two further gaps: absolute versus relative power changes the adversity–power association [13], so studies using different power definitions are not comparable; and no retrieved trauma study reports subject-level versus session-level contrasts, which is precisely the distinction that separated aperiodic from oscillatory measures elsewhere [10].

What would change the answer

A preregistered multisite trauma case-control study with harmonised spectra (site effects are mitigable, as shown in a 639-recording, 11-site neurodegenerative cohort [12]), thousands of participants, and a single fixed feature set that reports aperiodic exponent, periodic power, and connectivity side by side under one correction burden would settle the ranking. Embedding a within-study subsampling analysis would reveal whether any null reflects true absence or insufficient N [16]. Finally, stratifying trauma-exposed participants by data-driven biotype rather than by exposure or diagnosis is the design most likely to convert small, band-agnostic effects into large ones, given how much larger biotype contrasts are than DSM-group contrasts in an independent 2,151-participant cohort [17].

References

[1] Nitola-Pulido, L., Bonilla, D. S., Herrera Murcia, E., Reyes Gavilan, P. A., Santamaria-Garcia, H.. "Spectral power and functional connectivity of resting EEG in people exposed to armed conflict and controls". medRxiv. 2026-10-19T00:00:00.000Z. URL: https://doi.org/10.64898/2026.10.15.26365676

[2] Orozco Valero, A., Kopis-Posiej, N., Rodriguez-Gonzalez, V., Gutierrez-de Pablo, V., Morillas, C., Poza, J., Gomez, C., Martinez-Canada, P., Krukow, P.. "Neurophysiological excitation/inhibition imbalance in young adults burdened with childhood interpersonal trauma". bioRxiv. 2026-01-14T00:00:00.000Z. URL: https://doi.org/10.64898/2026.01.14.699432

[3] Hebron, H., Stephan, A. M.-, Cataldi, J., Siclari, F.. "Aperiodic EEG Activity Provides a Linear, Bidirectional, and Spatially Uniform Marker of Subjective and Objective Vigilance in Humans, Both Within and Across States". bioRxiv. 2025-09-02T00:00:00.000Z. URL: https://doi.org/10.1101/2025.08.30.673229

[4] Haim-Nachum, S., Zhang, C., Peng, K., Neria, Y., Zilcha-Mano, S., Zhu, X.. "Transdiagnostic Neurobiological Biotypes of Trauma Timing: Data-driven approach of Childhood and Adulthood-Onset Trauma". bioRxiv. 2026-04-27T00:00:00.000Z. URL: https://doi.org/10.64898/2026.04.22.720234

[5] Jonathan Lieberman, Brendan Brady, Ruth Lanius, Tomas Ros, Jean Theberge, Benicio Frey, Venkat Bhat, Frank Scharnowski, Andrew Nicholson. "Longitudinal Mapping of a Randomized Controlled Trial of EEG Neurofeedback in PTSD: Early Symptom Stabilization and Implications for Adjunctive Therapeutic Potential". psyarxiv. 2026-02-04T20:54:02.055Z. URL: https://osf.io/34tn2_v1

[6] Clancy, K. J., Ren, B., Rosso, I. M.. "Weaker functional segregation of the sensory and default mode networks in trauma-related intrusive memories". medRxiv. 2026-09-23T00:00:00.000Z. URL: https://doi.org/10.64898/2026.09.18.26363166

[7] Marshall, W. J., Conner, A. N., Key, A. P., Rex, T. S.. "Resting-state EEG and machine learning to investigate cortical connectivity as a biomarker in chronic mTBI". medRxiv. 2025-11-02T00:00:00.000Z. URL: https://doi.org/10.1101/2025.10.28.25338644

[8] Boncompte, G., Pedemonte, J. C., Cortinez, L. I., Ibacache, M., Calderon, N., Biggs, D.. "Aperiodic exponent ceases to track changes in propofol dose at deep surgery-level hypnotic depths". bioRxiv. 2026-06-22T00:00:00.000Z. URL: https://doi.org/10.64898/2026.06.17.732951

[9] Politanskaia, P., Bywater, J., Finley, A. J., Keage, H. A. D., Kelley, N. J., McKeown, D. J., Schinazi, V. R., Angus, D. J.. "Long-term reliability and stability of parameterized resting state EEG: Evidence from a five-year follow-up". bioRxiv. 2026-06-11T00:00:00.000Z. URL: https://doi.org/10.64898/2026.03.03.709208

[10] Duma, G. M., Valencia, N., Rasero, J., Bonanni, P., Pellegrino, G.. "EEG-Derived Proxies of Cortical Excitability in Epilepsy: Group Discrimination, Temporal Stability and Medication Sensitivity". medRxiv. 2026-06-04T00:00:00.000Z. URL: https://doi.org/10.64898/2026.06.02.26354693

[11] Gebregergis, B. T., Yhdego, H. G., Teklu, T.. "Reliability and disease sensitivity are dissociable properties of EEG foundation-model representations". bioRxiv. 2026-08-24T00:00:00.000Z. URL: https://doi.org/10.64898/2026.08.19.745052

[12] Jaramillo-Jimenez, A., Mantilla-Ramos, Y.-J., Tovar-Rios, D. A., Lopera, F., Aguillon, D., Ochoa-Gomez, J. F., Paquet, C., Gaubert, S., Pardini, M., Arnaldi, D., Taylor, J.-P., Fladby, T., Bronnick, K., Aarsland, D., Bonanni, L., on behalf of the European Dementia with Lewy Bodies Consortium (E-DLB),. "Characterizing resting-state EEG oscillatory and aperiodic activity in neurodegenerative diseases: A multicentric study". medRxiv. 2025-02-18T00:00:00.000Z. URL: https://doi.org/10.1101/2025.02.14.25322283

[13] Haleigh Brown, Emma Hart, Josué Rico-Picó, Kathryn Gray, Man Chen, Sonya Troller-Renfree. "A systematic review and meta-analysis on early adversity and resting-state EEG from infancy to adolescence". psyarxiv. 2026-09-17T21:05:30.147Z. URL: https://osf.io/s6vhm_v1

[14] Maehler, R., Reichenbach, A.. "Beyond the Label 'Major Depressive Disorder': Detailed Characterization of Study Population Matters for EEG-Biomarker Research". medRxiv. 2025-03-17T00:00:00.000Z. URL: https://doi.org/10.1101/2025.03.17.25324119

[15] Heitmann, H., Siani, J.-F., Zebhauser, P. T., Henningsen, P., Leucht, S., Priller, J., Ploner, M.. "Resting-state EEG activity as a Biomarker and Treatment Target in Depression: A Systematic Review and Meta-analysis". medRxiv. 2025-10-24T00:00:00.000Z. URL: https://doi.org/10.1101/2025.10.22.25338525

[16] Ebadi, A., Kaderi, S., Rodriguez-Herreros, B., Chabane, N., Allouch, S., Hassan, M.. "Sample size critically shapes the reliability of EEG case-control findings in psychiatry". bioRxiv. 2026-06-25T00:00:00.000Z. URL: https://doi.org/10.1101/2025.11.10.687610

[17] Trotti, R. L., Doss, I., Parker, D. A., Raymond, N., Sauer, K., Pearlson, G., Keedy, S., Gershon, E., Hill, S. K., Tamminga, C., McDowell, J., Lizano, P., Keshavan, M., Clementz, B.. "Evaluating Resting State EEG Biomarkers Across Psychosis Biotypes: Stability and HD-tDCS Modulation". medRxiv. 2026-02-25T00:00:00.000Z. URL: https://doi.org/10.64898/2026.02.23.26346924

How this survey was made

This survey answers one question: “Which resting-state EEG markers—spectral power, functional connectivity, or aperiodic/exponent measures—most reliably distinguish trauma-exposed populations from controls in large-scale studies?”

We keep our own collection of recent preprints and refresh it every day from bioRxiv, medRxiv, arXiv, PsyArXiv, SocArXiv and ChemRxiv. For each survey, we search that collection for recent work and keep the strongest matches: papers that use the same terms as the question, and papers on the same topic in different words.

The text was written from the titles and abstracts of those papers; the full texts were not read. Statements that draw on a particular paper are marked with a number in brackets. The reference list just above is the 17 papers whose work this text draws on, chosen from the 120 papers those searches returned; the rest of the shortlist is not printed. It is not a complete review of everything published on the question.

Treat this as a starting point and a reading list, not a peer-reviewed review. The searches behind this survey ran on October 22, 2026 and covered work from the previous two years.