Chronic pain and oscillatory neuroplasticity: Review on impaired compensatory network responses in resting-state electroencephalography

Abstract

Chronic pain is a multidimensional condition that involves persistent alterations in sensory, cognitive, and affective processes. Owing to its high temporal resolution and capacity to measure large-scale neural communications, electroencephalography (EEG) has emerged as a promising tool for identifying objective biomarkers of chronic pain. However, the findings of existing studies remain heterogeneous, which limits their clinical translation. In this narrative review, we synthesized recent resting-state EEG functional connectivity studies across a range of chronic pain conditions, highlighting the consistent frequency-specific abnormalities in the theta, alpha, beta, and gamma bands. Across studies, theta connectivity consistently increased prominently within sensory-limbic pathways, but decreased in cognitive-control networks, suggesting a maladaptive reallocation of learning-related neural resources. Alpha neurons showed reduced inhibitory maintenance in regulatory regions and excessive stabilization of neuropathic gating circuits. Beta oscillations demonstrated both overstabilization of affective–salience networks and weakened maintenance of top-down control, consistent with its role as a “status quo” rhythm. Gamma findings reflect disrupted high-frequency plasticity, ranging from excessive sensory precision to global microcircuit fragmentation. To integrate these findings, we propose a theoretical oscillatory neuroplasticity framework that may help to organize current observations across populations with chronic pain. Within this framework, frequency-specific alterations observed across studies can be interpreted as reflecting an imbalance across the learning (theta), maintenance (alpha), stabilization (beta), and fast plasticity (gamma) systems. This review further outlines methodological recommendations for enhancing reproducibility, improving cross-study comparability, and supporting the development of mechanistically informed EEG biomarkers of chronic pain.

1. Introduction

Chronic pain is a complex condition defined as persistent pain lasting more than 3 months.¹ In 2021, the prevalence of chronic pain in adults living in the USA was approximately 65 million, with the most common pain locations being the back, hip, and knee, resulting in economic costs of $722.8 billion, including expenses for medical care and loss of work productivity.² Diagnosis is often challenging, as in many conditions such as fibromyalgia, chronic pain is strongly linked to emotional distress and psychological factors.³

In this context, the development of objective quantitative biomarkers to improve diagnostic precision, inform treatment stratification, and monitor therapeutic responses is gaining increasing attention. Electroencephalography (EEG) is a promising tool for this effort. As a non-invasive method with high temporal resolution, EEG provides quantitative information regarding cortical oscillatory dynamics across frequency bands during the resting-state and task conditions, offering insight into both cognitive and affective processes.⁴ Two complementary approaches are commonly used: Spectral power, which quantifies the amplitude of oscillations within a region; and functional connectivity (FC), which measures the synchrony or statistical dependence between spatially distinct signals.^5,6 While power captures local neural activity, connectivity characterizes large-scale network coordination, an especially relevant feature, given that chronic pain alters sensory, cognitive, and emotional circuits.^7,8

From a clinical perspective, FC metrics may add value beyond spectral power, as they quantify the statistical relationships between spatially distinct neural signals. This is particularly relevant in chronic pain, which is increasingly understood as a disorder involving large-scale sensory, emotional, and cognitive networks, rather than isolated changes in regional brain activity.⁹ While spectral power provides information about local oscillatory activity,¹⁰ FC provides information regarding network-level coordination and may, therefore, offer a complementary perspective on the neurophysiological organization of chronic pain.¹¹ However, despite the growing interest, the findings from EEG studies remain conflicting. Differences in acquisition protocols, preprocessing pipelines, band definitions, analytical frameworks (sensor versus source space), and FC metrics complicate comparisons and hinder reproducibility. To advance beyond heterogeneous findings, we propose a conceptual framework in which frequency-specific changes across the theta, alpha, beta, and gamma oscillations are understood not as isolated abnormalities, but rather as complementary signatures of neuroplastic processes that may contribute to chronic pain.

This study was designed as a narrative review aiming to synthesize representative resting-state EEG functional connectivity findings in chronic pain and proposing a conceptual framework to integrate recurring oscillatory patterns. To define the scope of the narrative synthesis, we focused on original peer-reviewed English-language studies in adult chronic non-cancer pain populations that used resting-state EEG and reported quantitative functional connectivity or network-level analyses. Studies limited to spectral power without any connectivity measures, task-based EEG without a resting-state condition, acute or experimentally induced pain, pediatric populations, or non-human subjects were not included; more details are provided in the Supplementary Material.

2. Neurophysiology of pain and electroencephalography methods

Pain perception emerges from dynamic interactions between sensory, affective, and cognitive processes within the central nervous system. Nociceptive input ascends through the spinal cord and thalamus to cortical regions, including the somatosensory cortex, insula, anterior cingulate cortex, and prefrontal cortex. These areas integrate the sensory, emotional, and attentional components, making pain a multifactorial experience rather than a simple sensory signal.¹² Persistent nociceptive activity induces maladaptive plasticity, resulting in central sensitization and transition to chronic pain states.¹³

The functions of the top-down pain regulation system depend on descending modulatory pathways that integrate cortical, subcortical, and spinal mechanisms. In this network, the primary somatosensory cortex (S1) and the primary motor cortex (M1) modulate pain via projection- and layer-dependent mechanisms. In chronic pain, reduced inhibitory interneuron networks in the S1 and increased pyramidal neuron hyperactivity enhance nociceptive transmission and affective responses. In particular, layer 6 neurons promote nociceptive effects through projections to the striatum, thalamus, and anterior cingulate cortex, thus integrating sensory processing with the emotional and motivational aspects of pain.¹⁴ Descending pathways from the cortex and limbic system reach the periaqueductal gray and rostroventromedial medulla, where serotonergic, noradrenergic, and dopaminergic fibers modulate the spinal excitability. This balance between facilitation and inhibition may explain how cognitive and emotional states influence pain perception and potentially contribute to the amplification or attenuation of pain experiences.^14,15

Chronic pain can be categorized as nociceptive, neuropathic, or nociplastic. Nociceptive pain follows actual or threatened tissue injury, and may be more closely linked to stimulus-dependent sensorimotor and pain-evoked responses.^16,17 Neuropathic pain results from a lesion or disease of the somatosensory nervous system and may involve abnormal sensory gating, thalamocortical dysrhythmia, and altered alpha–gamma coupling within deafferented or disinhibited circuits.¹⁸ Consequently, the rhythmic abnormalities discussed in this review are hypothesized to vary by pain phenotype. For example, nociceptive pain may show stronger stimulus-linked sensorimotor changes, neuropathic pain may show more prominent sensory-gating and thalamocortical abnormalities, and neuropathic pain may show more distributed alterations in affective salience and top-down regulatory networks.

These mechanistic categories are clinically relevant as EEG oscillatory abnormalities may not be uniformly generalized across all chronic pain phenotypes. Nociceptive or mixed nociceptive–nociplastic conditions, such as osteoarthritis or chronic low back pain, may show stronger involvement of sensorimotor and descending modulatory networks.^17,19,20 Conversely, neuropathic and deafferentation-related pain states, such as diabetic neuropathy, postherpetic neuralgia, and phantom pain, may involve more prominent thalamocortical, sensory gating, and deafferentation-related mechanisms.²¹ Nociplastic conditions, such as fibromyalgia, may show broader alterations in salience, affective, and cognitive-control networks.²² Consequently, the framework proposed in this review should be interpreted as a transdiagnostic organizational model rather than as specific evidence of a single universal EEG signature across chronic pain disorders.

Central sensitization is characterized as a hyperexcitable state of dorsal horn neurons driven by sustained nociceptive input characterized by N-methyl-D-aspartate (NMDA) receptor phosphorylation and reduced gamma-aminobutyric acid (GABA)/glycine inhibition, such that even subthreshold signals can evoke pain. The activity spreads across neighboring synapses, producing hyperalgesia and allodynia. Through ongoing stimulation, these changes are consolidated into long-term plasticity which ultimately maintains the presence of pain over time.^8,13 The relationship between chronic pain and mood disorders is particularly strong; Vadivelu et al²³ reported that 30%–45% of patients with chronic pain also experience depression, which increases pain vulnerability through shared serotonergic, dopaminergic, and noradrenergic dysregulation.¹² Cognitive–emotional factors, such as catastrophizing intensify pain, whereas adaptive traits such as optimism and acceptance mitigate its impact.^23,24 From an integrative perspective, pain perception must be understood as a biopsychosocial phenomenon that relies on neurophysiological processes interacting with emotions, cognition, and individual experiences.

3. Power and functional connectivity: Conceptual and methodological distinctions

EEG analysis has been used to quantify the statistical dependencies between neural signals recorded from different brain regions, reflecting the coordination of oscillatory activity across cortical networks.^6,8 The power spectrum quantifies local oscillatory activity at each sensor or source across the different frequency bands (delta, theta, alpha, beta, gamma), and can be represented in terms of the absolute power (μV²) and relative power (%).²⁵ FC represents the temporal synchronization or phase relationships between spatially distinct signals, thereby providing insights into large-scale network coordination.^6,26–28 Resting-state EEG is particularly valuable in chronic pain research as it captures spontaneous brain activity across multiple frequency bands (delta, theta, alpha, beta, and gamma), providing insights into large-scale network organization and potential disruptions in communication between sensory, cognitive, and affective regions.²⁹ Through these analyses, EEG-based FC allows for the examination of both local and global patterns of cortical coordination that may underlie chronic pain processing.

These measures capture the distinct aspects of brain function, and should not be interpreted interchangeably. Changes in spectral power may reflect local alterations in neuronal excitability or oscillatory dynamics, without necessarily indicating any changes in interregional communication. Functional connectivity is designed to characterize the coordination of activity across various distributed brain regions, but does not establish the direction of information flow.³⁰ Several analytical approaches have been developed to assess FC, each of which targets different aspects of neural synchronization. Phase-based metrics, such as the phase-locking value (PLV), weighted phase lag index (wPLI), and imaginary part of coherence, estimate the stability of phase differences between signals while minimizing the effects of any common sources and volume conduction.³¹ Amplitude-based measures, including the amplitude envelope correlation (AEC), assess cofluctuations in signal power over time.³² Source-space methods (e.g., standardized low-resolution brain electromagnetic tomography [sLORETA]) estimate the cortical current density and improve anatomical precision relative to the sensor space. Connectivity between these source signals can then be computed using leakage-resistant metrics to obtain source-space connectivity maps.³³ In addition, graph-theoretical approaches quantify network topology through measures of efficiency, clustering, and hub structure, offering insights into how pain may alter brain network organization.³⁴ Collectively, these methods provide complementary insights into the functional integration and segregation of neural activity during chronic pain.

Chronic pain is considered a multifactorial disorder caused by large-scale network dysfunction, rather than isolated regional abnormalities.^35,36 The FC provides a more direct framework for investigating its neurophysiological basis, particularly when aiming to identify trait-level neural signatures and neuroplastic reorganization between patients with chronic pain and controls.^{30,31,37–39} Previous studies have demonstrated alterations in communication among the sensory, limbic, salience, and cognitive control systems,^37,38 which are inherently interregional and are therefore better captured by FC than by power alone. Connectivity-based approaches also align with network theories, such as thalamocortical dysrhythmia, which frames chronic pain as a sustained imbalance in coordinated oscillatory signaling across distributed circuits.^25,40 Although spectral power remains informative, particularly for characterizing global shifts in dominant rhythms or excitation–inhibition balance, FC metrics may be more sensitive to stable, circuit-level reorganization, and therefore better suited for trait-oriented biomarker development in chronic pain. Accordingly, spectral power and functional connectivity should be considered as complementary, but distinct, analytical approaches, each capturing different aspects of brain function and network organization.

4. Electroencephalography functional connectivity in chronic pain

In this narrative review, we identified 11 studies that assessed resting-state EEG-FC across a range of chronic pain conditions, compared with healthy controls or non-painful clinical controls. The study selection criteria are described in the Supplementary Material. The acquisition protocols were heterogeneous, with 19–64 channels, recordings from 3 to 24 min, and both sensor- and source-space pipelines. Connectivity was quantified using phase-based metrics (PLV, wPLI, lagged phase/linear coherence, and imaginary coherence), amplitude-based measures (orthogonalized amplitude envelope correlation), and source-based lagged connectivity (sLORETA/exact low-resolution brain electromagnetic tomography [eLORETA]), which are commonly combined with graph-theoretical analyses or machine learning classifiers. Most studies included low and high-frequency bands between delta and gamma ranges (0.5–100 Hz) in their analysis, although the specific bands differed by paper. Most of the results focused on theta, beta, and gamma oscillations. To ensure consistency in terminology across the included studies, when reported in the original studies, sub-frequency bands were defined according to conventional EEG ranges: alpha-1 (8–10 Hz), alpha-2 (10–12 Hz), beta-1 (13–18 Hz), beta-2 (18.5–21 Hz), and beta-3 (21.5–30 Hz); detailed frequency definitions across studies are summarized in Table 1. The primary focus of this review was resting-state EEG-FC. However, spectral power analyses and other findings, such as alpha peak frequency and theta/alpha ratios, were considered when they provided a relevant context for interpreting FC alterations and broader patterns of neurophysiological dysregulation in chronic pain. Throughout this review, spectral power findings are discussed separately from FC findings whenever applicable, recognizing that power reflects a local oscillatory activity, while FC captures the synchronization between spatially distributed neural populations.

Table 1.

Summary of studies’ analysis in electroencephalography connectivity.

Author	Country	Study design	Sample	Acquisition	Connectivity analyzed conducted	Frequency bands range	Results	Conclusions
Alves et al41	Brazil	Cross-sectional study	49 FM, 15 HC	18-channel EEG (ENOBIO 20, Neuroelectrics), 8-min resting (eyes open + eyes closed)	Lagged coherence connectivity using sLORETA algorithms; analysis of 8 regions of interest (bilateral S1, INS, ACC, DLPFC); statistical non-parametric mapping with 5000 permutations.	Delta (1–3.5 Hz), Theta (4–7.5 Hz), Alpha-1 (8–10 Hz), Alpha-2 (10–12 Hz), Beta-1 (13–18 Hz), Beta-2 (18.5–21 Hz), Beta-3 (21.5–30 Hz), Gamma (30.5–44 Hz).	↑Lagged coherence connectivity between the left DLPFC and right ACC in β3 band (EO); ↑inter-insular and left INS–right DLPFC connectivity (β3, EO–EC). Negative correlations: pain disability versus β3 right ACC– right S1 connectivity (EC); central sensitization versus α2 right ACC– left S1 (EO); serum BDNF versus γ left DLPFC – right INS (EO–EC).	FM patients showed hyperconnectivity in pain-processing circuits, particularly in β3 band at rest, suggesting a neural signature of chronic pain and neuroplastic changes linked to symptom severity.
De Ridder et al42	Belgium	Cross-sectional study	50 chronic neuropathic pain, 50 tinnitus, 50 HC	19-channel EEG (Mitsar-201, NovaTech); ~7-min recording (≥5-min artifact-free), eyes closed	Lagged phase coherence through Pascual-Marqui’s approach Log-transformed electric current densities within the gamma frequency bands (30.5–44 Hz) for specific ROIs. LORETA-Key program	Delta (2–3.5 Hz), Theta (4–7.5 Hz), Alpha (8–12 Hz), Beta (13–30 Hz), Gamma (30.5–44 Hz)	↑ Theta connectivity between bilateral ACs, SCs, and PHCs, and between ipsilateral AC/SC/VEC/VIC and PHC in pain vs HCs. ↑ Common theta connectivity across the same bilateral and ipsilateral networks in both tinnitus and pain vs HCs.  ↑ Gamma-band current density across multiple nonchemical sensory cortices.	Pain and tinnitus are associated with prediction errors in all sensory cortices, except the olfactory and gustatory cortex.
Ding et al43	United States	Cross-sectional case study	8 chronic pain, 8 HC	64-channel EEG (WaveguardTM cap, ANT Neuro, or SynAmps2, Compumedics Neuroscan); 2-min eyes open and 2-min eyes closed	Graph theory analysis, PLV network connectivity strength EEGLAB toolbox, MATLAB	Alpha (8–12 Hz), Beta (13–30 Hz), Theta (4–8 Hz), Gamma (30–80 Hz)	↓ theta frontoparietal connectivity in chronic pain versus HCs during mechanical QST, and reduced peak alpha frequency. Less pain-modulated theta connectivity.	Central sensitization may be associated with impaired pain theta connectivity.
Erlenwein et al44	Germany	Cross-sectional study	87 hip osteoarthritis	34-channel cap (Ag/AgCl, MEQNordic); signals amplified with NuAmps (Neuroscan); four 2.5-min sequences (eyes open/closed alternating)	PLI SW: Neurophysiological Biomarker Toolbox	Delta (0.5–4 Hz), Theta (4–8 Hz), Alpha (8–12 Hz), Beta (12–32 Hz)	↑delta in whole-brain FC Women ↑comparable during the coldpressor test and evoked clinical pain.	Increasing delta ban whole-brain functional connectivity is observed in response to tonic and clinical pain stimulation across the advancing pain chronification stages.
Gao et al45	China	Cross-sectional, study	21 PHN, 17 HC	32-channel EEG (NuAmps, NeuroScan labs), 10–20 system; 10-min resting, eyes closed	wPLI, analyzed with the Network-Based Statistic (NBS) method MATLAB 2013b	Delta (1–4 Hz), Theta (4–7 Hz), Alpha (7–13 Hz), Beta (13–30 Hz), Gamma (30–70 Hz)	↓gamma wPLI in CP4, Cz, C4 (CP4 most affected); ↓connectivity across frontal, central, parietal, and occipital regions. Also, widespread increases in resting-state alpha power.	Gamma-band disconnection indicates impaired sensorimotor–pain network coordination and weakened inhibitory control, independent of the symptom severity.
Makowka et al46	Switzerland	Cross-sectional, case-control pilot study	16 FM, 11 HC	64-channel EEG (BioSemi ActiveTwo), 24-min resting, eyes closed,	Absolute imaginary component of coherence as index of FC and calculated the weighted node degree (WND) for each solution point as the sum of its coherence with all the other cortical solution points. SW.: MATLAB, using the NUTMEG toolbox.	Delta (0.5–3.5 Hz), Theta (3.5–7.5 Hz), Alpha (7.5–12.5 Hz), Low beta (13–20 Hz), High beta (20–30 Hz)	↓ high beta in left mesiotemporal area in CP; ↓ high beta in basolateral amygdala in CP.	EEG measures of GABA-related signaling revealed two key findings: a decrease in high-beta functional connectivity in the basolateral amygdala, linked to the emotional aspect of pain, and an increase in left prefrontal cortex low-beta power, associated with ongoing pain intensity.
Martin-Brufau et al47	Spain	Observational, cross-sectional, case–control study	23 FM, 23 HC	21-channel EEG (Neuron-Spectrum-AM), 15-min eyes-closed resting	Coherence method (Brainstorm software, MATLAB R2015b); FFT for amplitude; functional connectivity indices between frontal and temporal electrodes (Fz-T3/T4); sLORETA for source localization; ROC curve analysis for discrimination accuracy.	Delta (1–4 Hz), Theta (4–7 Hz), Alpha (7–14 Hz), Beta (15–32 Hz).	↓ connectivity mainly in Theta and Alpha in bilateral frontotemporal regions; ↓ Amplitude in all bands except Delta in FM versus HC.	FM patients show distinct EEG connectivity and reduced cortical synchrony, indicating a specific neurophysiological signature useful as a complementary diagnostic marker.
Ta Dinh et al48	Germany	Cross-sectional study	101 mixed chronic pain conditions, 84 HC	64-channel EEG (10–20 + additional), BrainAmp MR plus; 5-min eyes closed + 5-min eyes open	PLV, dwPLI, and orthogonalized AEC SW: BrainVision Analyzer; MATLAB	Theta (4–8 Hz), Alpha (8–13 Hz), Beta (14–30 Hz), Gamma (60–100 Hz), Low gamma (45–55 Hz excluded due to noise)	↑theta and gamma FC in frontal regions, but no significant local group differences by permutation testing; ↓global efficiency and ↓hub disruption index in gamma networks, consistent across PLV and dwPLI; no correlations with clinical outcomes.	Chronic pain is associated with widespread reorganization of rs networks (integration/segregation imbalance) in the theta and gamma frequencies, consistent with thalamocortical dysrhythmia in pain.
Topaz et al49	Israel	Cross-sectional study	133 painful DPN, 47 non-painful DPN	64-channel EEG (ActiCHamp, Brain Products), 3-min resting, eyes closed	Magnitude-squared coherence (MSC) to compute inter-electrode functional connectivity; feature selection using ReliefF algorithm; classification with support vector machine (SVM)	Theta (3.5–7.5 Hz), Alpha (7.5–12.5 Hz), Beta (12.5–30 Hz), Gamma (30–40 Hz)	↑ Connectivity in painful versus non-painful DPN in theta and alpha bands (P = 0.008, P = 0.001). ROC AUC = 0.93 (theta), 0.89 (beta). Significant pairs: AF3–AFz (theta), FC2–C1, C1–F2 (alpha), F1–F2 (gamma). Connectivity correlated with reported pain.	Resting-state EEG FC discriminates painful from non-painful DPN with high accuracy; increased cortical connectivity may reflect central contribution to pain generation.
Ueno et al50	Germany	Cross-sectional study	34 chronic low back pain, 34 HC	64-channel EEG (BrainAmp MR plus), 5-min resting, eyes closed	Source-space (eLORETA) and lagged linear connectivity between 24 cortical regions of interest. EEGLAB toolbox, MATLAB	Delta (1.5–6.0 Hz), Theta (6.5–8.0 Hz), Alpha-1 (8.5–10.0 Hz), Alpha-2 (10.5–12.0 Hz), Beta-1 (12.5–18.0 Hz), Beta-2 (18.5–21.0 Hz), and Beta-3 (21.5–30.0 Hz)	↓ in the b3 in the left middle temporal gyrus–posterior cingulate cortex and ventral medial prefrontal cortex left inferior parietal lobule connection in CP, indicating weakened coupling within the default-mode and fronto-parietal networks. Also, prefrontal theta and delta activity correlated positively with pain symptoms.	Altered FC in specific brain regions and frequencies, particularly increased beta-1 FC between the right DLPFC and right auditory cortex was associated with greater pain intensity.
Vanneste et al51	Ireland	Cross-sectional study	50 neuropathic pain, 50 HC	64-channel EEG (Neuroscan), 5 minutes recorded with eyes closed	Lagged phase coherence sLoreta SW: LORETA-Key software	Delta (2–3.4 Hz), Theta (4–7.5 Hz), Alpha (8–12 Hz), Beta (13–30 Hz) and Gamma (30.5–44 Hz)	↑ theta between pgACC and both SSC in CP patients ↓ Alpha between pgACC, dACC and both SSC in CP patients.	Chronic pain is a brain imbalance disorder resulting from a pathological current density ratio and decreased functional connectivity between the pain input areas and the pain suppression area.

AC: Auditory cortex; ACC: Anterior cingulate cortex; AEC: Amplitude envelope correlation; AUC: Area under the curve; BDNF: Brain-derived neurotrophic factor; CLBP: Chronic low back pain; CP: Chronic pain; CBT: Cognitive behavioral therapy; dwPLI: Debiased weighted phase-lag index; dACC: Dorsal anterior cingulate cortex; DLPFC: Dorsolateral prefrontal cortex; DMN: Default mode network; DPN: Diabetic polyneuropathy; EC: Eyes closed; EEG: Electroencephalography; eLORETA: Exact low-resolution brain electromagnetic tomography; EO: Eyes open; E/I: Excitation/inhibition; FFT: Fast Fourier transform; FC: Functional connectivity; FM: Fibromyalgia; GABA: Gamma-aminobutyric acid; HC: Healthy control; INS: Insular cortex; M1: Primary motor cortex; MEP: Motor evoked potential; MSC: Magnitude-squared coherence; MTG-PCC-vmPFC: Middle temporal gyrus-posterior cingulate cortex-ventromedial prefrontal cortex; NBS: Network-Based Statistic; NMDA: N-methyl-D-aspartate; OA: Osteoarthritis; PHC: Parahippocampus; PHN: Postherpetic neuralgia; pgACC: Pregenual anterior cingulate cortex; PLI: Phase lag index; PLV: Phase-locking value; qEEG: Quantitative electroencephalography; QST: Quantitative sensory testing; ROI: Region of interest; ROC: Receiver operating characteristic; rs: Resting-state; S1: Primary somatosensory cortex; SC: Somatosensory cortex; SICI: Short-interval intracortical inhibition; sLORETA: Standardized low-resolution brain electromagnetic tomography; SSC: Somatosensory cortex; SVM: Support vector machine; SW: Software; TCD: Thalamocortical dysrhythmia; VEC: Visual extrastriate cortex; VIC: Visual inferior cortex; WND: Weighted node degree; wPLI: Weighted phase-lag index.

These methodological differences are important and therefore should be considered when interpreting the consistency of findings across studies. The included studies varied widely in sample sizes, from small exploratory cohorts to larger case-control studies. This variation may have influenced the statistical power, result precision, and generalizability of the findings. Therefore, the results from smaller studies should be interpreted with caution, particularly when not replicated in independent samples or across different pain conditions. In addition, the connectivity metrics used across studies were not equivalent. Some metrics focus on phase synchronization, whereas others are designed to reduce the influence of volume conduction or common-source effects. Other approaches, such as amplitude envelope correlation, capture changes in signal amplitude, rather than phase relationships. As these methods measure different aspects of neural communication, they may produce different connectivity patterns. Therefore, the proposed framework should be understood as a synthesis of broad frequency- and network-level trends, rather than as evidence that all connectivity metrics reflect the same underlying mechanisms.

Chronic pain was consistently associated with a frequency-specific reorganization of brain networks across studies, with theta, alpha, beta, and gamma bands each showing distinct alterations across the sensory, affective, and cognitive systems. Despite variations in pain phenotypes and analytical methods, a coherent multiband pattern emerged [Fig. 1].

Fig. 1.

Unified oscillatory neuroplasticity framework of chronic pain across the theta, alpha, beta, and gamma oscillations. This figure summarizes the conceptual framework linking frequency-specific EEG alterations to neuroplastic processes in chronic pain. Theta, alpha, beta, and gamma alterations are interpreted as reflecting changes in learning, salience encoding, and network updating; inhibitory control and sensory gating; stabilization of pain-related states and top-down regulation; and fast-timescale integration, sensory precision, and excitation–inhibition balance. Arrows indicate the direction of reported alterations relative to comparison groups or non-painful conditions. The studies listed in each panel are representative examples discussed in the review, and are intended to illustrate recurring patterns, rather than provide a quantitative meta-analysis. The framework should be interpreted as hypothesis-generating and not as evidence of a single universal EEG signature of chronic pain. ACC: Anterior cingulate cortex; DLPFC: Dorsolateral prefrontal cortex; EEG: Electroencephalography; E/I: Excitation/inhibition.

Theta oscillations are central to learning and plasticity, and support network updating, salience encoding, and adaptive control.^{42,43,47–50} Within our framework, theta hyperconnectivity in sensory–limbic and salience-linked pathways is interpreted as the biased learning of pain-related representations, consistent with the role of theta bands in coordinating control and learning signals across networks. De Ridder et al⁴² reported enhanced theta coherence between the bilateral auditory, somatosensory, and parahippocampal cortices, whereas Ta Dinh et al⁴⁸ found increased frontal theta using PLV and dwPLI. Vanneste et al⁵¹ further demonstrated heightened pregenual anterior cingulate cortex (pgACC) and somatosensory cortex (SSC) theta coupling in deafferentation pain, whereas Topaz et al⁴⁹ showed greater theta coherence (AF3–AFz) in painful versus non-painful diabetic neuropathy. In contrast, theta reductions were identified in control networks. Martín-Brufau et al⁴⁷ described decreased frontotemporal theta coherence in fibromyalgia, while Ding et al⁴³ found identified frontoparietal theta connectivity during stimulation. Ueno et al⁵⁰ further linked elevated prefrontal theta power to higher pain intensity. Taken together, these findings reflect enhanced limbic–sensory coupling and weakened top-down control. However, the exact direction of the relationship remains unclear. One study indicated that persistent nociceptive inputs may reinforce theta-related sensory–limbic learning, whereas reduced frontoparietal and prefrontal theta connectivity may weaken cognitive control and descending modulation.⁹ Overall, existing studies cannot determine whether changes in theta connectivity occur before chronic pain, develop as a result of persistent pain, or reflect a compensatory response. Therefore, theta changes should be interpreted as possible signs of maladaptive or compensatory learning-related network activity rather than as direct evidence that chronic pain is caused by overlearning.

Alpha oscillations mediate inhibitory maintenance and top-down gating, thereby stabilizing recently encoded representations.^{41,43,45,47,49,51} Alpha reductions in regulatory regions, such as the anterior cingulate cortex (ACC), dorsolateral prefrontal cortex (DLPFC), and insula pathways, reflect weakened inhibitory maintenance, consistent with the role of alpha bands in suppressing irrelevant inputs and maintaining stable control states.^41,51 Decreases in alpha bands dominate pain-processing and regulatory circuits. Vanneste et al⁵¹ showed reduced alpha between the pgACC, dorsal anterior cingulate cortex (dACC), and SSC; Martín-Brufau et al⁴⁷ identified diminished long-range frontotemporal alpha coherence; and Alves et al⁴¹ reported alpha-2 reductions across the insula, ACC, DLPFC, and S1, correlating with pain disability. Ding et al⁴³ also observed reduced alpha peak frequencies. Conversely, alpha increases appeared in neuropathic sensory-gating phenotypes. Topaz et al⁴⁹ observed elevated alpha coherence in the frontocentral connections (FC2–C1, C1–F2) in painful neuropathy, and Gao et al⁴⁵ reported global alpha power increases in postherpetic neuralgia. Importantly, alpha oscillations are involved in several processes, including inhibition, attention control, sensory filtering, and maintaining task-relevant information.⁵² Therefore, alpha changes in chronic pain should not be interpreted as a single mechanism. Instead, they should be viewed as context-dependent signs of changes in the inhibitory and regulatory brain networks.

Beta oscillations stabilize the current neural state and resist change, reflecting maintenance of the “status quo”.^41,46,50,53 Beta alterations varied by phenotype, but consistently involved the prefrontal, cingulate, insular, and somatosensory networks. In chronic low back pain, Ueno et al⁵⁰ identified reduced beta-3 connectivity among the temporal, posterior cingulate, and ventromedial prefrontal regions, whereas beta-1 DLPFC auditory connectivity correlated with pain intensity. In fibromyalgia, Alves et al⁴¹ identified beta-3 hyperconnectivity between the DLPFC and ACC and between the bilateral insulae, whereas Makowka et al⁴⁶ observed high-beta hypoconnectivity in the mesiotemporal/amygdala region and increased low-beta power in the prefrontal cortex, both of which track pain symptoms. Topaz et al⁴⁹ further identified that beta connectivity strongly distinguishes painful from non-painful neuropathy. Overall, beta rhythms reflect the overstabilization of maladaptive pain states or weakened maintenance of regulatory configurations. Although beta activity is often linked to the maintenance of the current sensorimotor or cognitive state, beta-band changes in chronic pain may be explained. Other authors have suggested that beta activity is also related to attention, emotions, movement, or medication use.⁵⁴ Therefore, beta findings should be interpreted based on the specific pain type, brain networks involved, and the analysis method used.

The gamma band activity captures disruptions in fast-timescale microcircuit function^{42, 45, 48, 49} . Gamma hyperconnectivity localized to sensory hubs is generally interpreted as a maladaptive process and overbinding of nociceptive representations, whereas gamma hypoconnectivity and reduced network efficiency are interpreted as fragmented fast-timescale integration. De Ridder et al⁴² reported increased gamma activity across sensory cortices, with the parahippocampus acting as a hub. Additionally, Ta Dinh et al⁴⁸ observed increased local frontal gamma but reduced global gamma network efficiency. In painful diabetic neuropathy, Topaz et al⁴⁹ identified an elevated gamma coherence (F1–F2) linked to pain severity. In contrast, Gao et al⁴⁵ observed widespread gamma hypoconnectivity across the frontal, central, parietal, and occipital regions in patients with postherpetic neuralgia, indicating impaired high-frequency integration and altered excitation–inhibition balance. Therefore, the gamma band findings should be interpreted with caution, because the high-frequency activity recorded from the scalp may be influenced by muscle activity and other non-brain artifacts.⁵⁵ In particular, gamma connectivity results should be evaluated based on the artifact removal steps, preprocessing methods, and frequency range used in each individual study. Future studies should clearly report how noise and artifacts can be removed. This may include independent component analysis or similar artifact removal methods. Studies should also include sensitivity analyses when interpreting gamma-band connectivity as a possible neural marker.

5. Oscillatory framework of chronic pain

Rather than replacing existing mechanistic models, the proposed framework is intended to aid these perspectives by providing an integrative interpretation of the frequency-specific findings recurring across studies. These oscillations are multifunctional and context-dependent, and their interpretation depends on the brain region, behavioral state, pain phenotype, medication exposure, and the analytical method used. Therefore, the labels used in this framework–learning, maintenance, stabilization, and fast plasticity–are intended as functional anchors to organize recurring EEG findings, rather than as direct mechanistic explanations. While descriptive synthesis can catalogue which frequency bands are altered in individual studies, it cannot fully explain how those alterations relate across different pain types, brain networks, or methods. As such, this framework is best understood as a hypothesis-generating map for future mechanistic and biomarker studies, rather than as a claim that frequency bands have fixed or exclusive biological meanings.

This interpretation is compatible with, rather than opposed to, other theoretical models of chronic pain. Predictive coding models frame chronic pain as an imbalance between the sensory input, prior expectations, and precision weighting.⁵⁶ Thalamocortical dysrhythmia is characterized by abnormal low-frequency thalamocortical activity, with associated changes in high-frequency oscillations.⁵⁷ Salience network models emphasize the persistent prioritization of pain-related signals, while excitation-inhibition imbalance models highlight disrupted cortical gain control and inhibitory regulation.⁵⁸

Converging mechanistic models have suggested that chronic pain may involve a slowing of dominant thalamocortical rhythms from alpha toward theta with concomitant changes in high-frequency activity, an organization commonly discussed under thalamocortical dysrhythmia (TCD), where reduced thalamic drive promotes low-frequency (theta-range) thalamocortical activity and may be accompanied by increased surrounding gamma “edge effects”.^57,59 In parallel, chronic pain has been framed as a disorder of cross-frequency interactions and frequency-specific communication channels. Slow rhythms (theta and alpha) shape large-scale gating and long-range coordination, while fast rhythms (beta and gamma) support rapid integration and precision. Accordingly, an imbalance across these systems may manifest as altered coupling and/or disproportionate engagement of slow versus fast bands, thereby providing a mechanistic rationale for the theta-alpha and beta-gamma shifts described in our framework.^60,61

Although the present framework focuses specifically on learning, maintenance, stabilization, and rapid plasticity changes, oscillatory activity is multifunctional and relies on context.⁶⁰ Therefore, interpretation of the theta, alpha, beta, and gamma band results should be viewed as heuristic, rather than exclusive. Other possible explanations for this include ideas from the chronic pain literature, such as predictive coding concepts, which interpret chronic pain as aberrant precision weighting of nociceptive signals; thalamocortical dysrhythmia models, which emphasize the pathological slowing of thalamocortical rhythms; salience-network dysfunction models; and excitation-inhibition imbalance theories.⁶²

Taken together, the findings from the investigated connectivity and oscillatory studies have identified theta band activity as a recurring feature across chronic pain conditions. The observed pattern of increased theta synchronization within sensorylimbic pathways and reduced engagement of cognitive control networks is broadly consistent with theoretical models linking theta oscillations to network updating and adaptive plasticity. Hyperconnectivity within pain pathways, such as strengthened coupling between sensory cortices and the parahippocampus,⁴² enhanced frontal theta activity,⁴⁸ increased pgACC–SSC synchrony in deafferentation pain,⁵¹ and elevated theta connectivity in painful diabetic neuropathy,⁴⁹ suggests that theta-driven plasticity is preferentially allocated to nociceptive and affective-salience networks. Conversely, theta hypoconnectivity in the frontotemporal and frontoparietal circuits reflects the reduced learning-related engagement of the cognitive control systems needed for the regulation, reframing, or extinction of pain signals.^43,47 The behavioral relevance of this imbalance is supported by findings such as those of Ueno et al,⁵⁰ who demonstrated that prefrontal theta power scales positively correlate with pain intensity.

Oscillatory studies further reinforced the role of theta as a compensatory maladaptive learning signal. Across multiple populations, theta increases appeared when inhibitory tone or corticospinal function was compromised, indicating an attempt to recalibrate or stabilize the impaired circuits. In amputees, greater frontal-central-parietal theta correlated positively with short-interval intracortical inhibition (SICI), indicating upregulated theta engagement when inhibition is reduced.⁶³ In spinal cord injury, theta power is negatively associated with motor evoked potential (MEP) amplitude across cortical regions, again reflecting heightened slow-frequency plasticity when the corticospinal drive is weakened.⁶⁴ In chronic neuropathic pain, a higher resting theta, particularly in the central regions, is associated with lower pain intensity,⁶⁵ indicating a potentially protective or salutogenic learning response. In addition, stimulus-evoked theta event-related synchronization (ERS) predicts better symptom profiles,²⁵ consistent with a model in which theta-driven plasticity initially provides compensation, but may become maladaptive or insufficient when chronically overloaded.

In contrast to the proposed role of theta oscillations in network updating, alpha oscillations have been widely associated with inhibitory control, the maintenance of ongoing network states, and the suppression of task-irrelevant information.^66–68 Similar to Klimesch’s inhibition-timing framework,^67,68 this perspective provides a useful basis for interpreting the bidirectional alpha findings observed across chronic pain populations.

Event-related studies further support this interpretation. In knee osteoarthritis, Marques et al⁶⁹ revealed that the alpha ERS, an inhibitory and organizational marker, was reduced in individuals with greater pain severity, longer pain chronicity, and lower pain thresholds, indicating that insufficient alpha-mediated stabilization was associated with a weaker cortical inhibitory response. Their finding that pain correlates with lower-alpha ERS aligns with a broader pattern of inhibitory failure, in which chronic pain reduces the brain’s ability to maintain organized top-down control over sensory and motor representations. This view is consistent with the results from fibromyalgia and neuropathic pain studies, in which task-evoked alpha restoration predicted better symptom profiles, whereas reduced resting alpha was linked to affective and cognitive symptoms. These findings support a model in which chronic pain disrupts the role of alpha in maintaining and protecting network states, leading to diminished inhibitory tone and unstable maintenance of the sensorimotor and affective representations. In this framework, chronic pain represents not only excessive updating within nociceptive circuits (theta), but also an insufficient alpha-mediated maintenance, allowing aberrant pain signals to persist without adequate top-down regulation.

Beta oscillations are widely characterized as a rhythm that stabilizes the current sensorimotor and cognitive set, signaling maintenance of the “status quo” and resistance to neural change.^53,70 Across chronic pain studies, beta abnormalities were found to be broadly consistent with this perspective, revealing patterns of altered stabilization within affective, salience, and regulatory networks. In knee osteoarthritis, Simis et al⁷¹ revealed that frontocentral high-beta power increased with pain intensity and osteoarthritis (OA) severity, while reduced theta power accompanied these elevations, suggesting a shift toward rigid, antiplastic cortical dynamics in more severe phenotypes. Complementing this, Ueno et al⁵⁰ reported reduced beta-3 connectivity across the middle temporal gyrus-posterior cingulate cortex-ventromedial prefrontal cortex (MTG-PCC-vmPFC) networks in chronic low back pain, indicative of weakened maintenance in regulatory hubs, alongside increased beta-1 connectivity between the right DLPFC and auditory cortex proportional to pain intensity, pointing to selective stabilization of maladaptive perceptual states. In fibromyalgia, two independent studies revealed the same duality: Alves et al⁴¹ demonstrated beta-3 hyperconnectivity across the DLPFC–ACC and bilateral insula linked to pain disability and central sensitization, while Makowka et al⁴⁶ identified high-beta hypoconnectivity in the mesiotemporal–amygdala region and increased low-beta power in the prefrontal cortex tracking ongoing pain, indicating unstable limbic fast-frequency processing alongside excessive prefrontal stabilization. Neuropathic pain echoes this pattern. Topaz et al⁴⁹ revealed that beta connectivity features robustly distinguished painful from nonpainful diabetic neuropathy, highlighting beta synchrony as a marker of entrenched nociceptive states. Current evidence indicates that chronic pain disrupts the normative role of beta by promoting the overstabilization of affective-sensory circuits and insufficient maintenance of regulatory pathways, yielding an oscillatory profile with anti-learning, reduced flexibility, and reinforcement of persistent pain representations.

Gamma oscillations represent the fastest coordinated neural rhythms and have been widely associated with rapid neural communication, synaptic plasticity, and dynamic integration across different distributed cortical networks.^71–74 Additionally, gamma abnormalities have been linked to disruptions in fast timescale cortical processing and network integration during chronic pain. De Ridder et al⁴² revealed increased gamma activity across the sensory cortices, with the parahippocampal region acting as a hub, which is consistent with the heightened sensory gain and excessive Hebbian strengthening of pain-related ensembles. Similarly, Ta Dinh et al⁴⁸ similarly reported local frontal gamma increases but reduces global gamma efficiency, indicating fragmented fast timescale integration: overly synchronized microcircuits embedded within a poorly integrated large-scale network. In painful diabetic neuropathy, Topaz et al⁴⁹ identified gamma hyperconnectivity (F1–F2) correlated with pain severity, aligning with the role of gamma in amplifying the precision of nociceptive coding. In contrast, Gao et al⁴⁵ found widespread gamma hypoconnectivity across the frontal, central, parietal, and occipital regions in patients with postherpetic neuralgia, indicating the collapse of high-frequency communication and impaired excitation/inhibition (E/I) balance, specifically involving the microcircuit instability expected when gamma-mediated binding fails. Supporting the role of gamma rays in fast adaptive reorganization, noise electrical stimulation during motor learning has been shown to reduce beta and gamma corticomuscular coherence while accelerating learning, demonstrating that reducing excessive high-frequency synchrony can enable more efficient plasticity in motor circuits.⁷⁵ Together, these results suggest that chronic pain disrupts the normative role of gamma rays in rapid plasticity, flexible binding, and high-precision sensory integration, thereby producing a spectrum ranging from maladaptive hyperprecision (overbinding of pain-related assemblies) to microcircuit fragmentation (loss of fast-timescale coherence). In this framework, gamma abnormalities reflect the misallocation of fast plasticity, whereby nociceptive representations are strengthened, and global integrative precision is degraded.

6. Translational requirements for electroencephalography biomarkers in chronic pain

Future studies should go beyond group-level case-control comparisons of EEG functional connectivity measures to identify clinically useful biomarkers. Candidate biomarkers should show test-retest reliability, reproducibility across laboratories and EEG systems, and diagnostic specificity across both painful and non-painful conditions. They should additionally be robust after accounting for medication use, sleep quality, psychiatric comorbidities, and other clinical confounding factors.

Importantly, EEG markers must provide values beyond those of standard clinical assessments, pain intensity ratings, questionnaires, quantitative sensory testing, and existing neurophysiological measures. Therefore, prospective validation in independent cohorts is necessary before EEG connectivity measures can be considered clinically actionable for the diagnosis, prognosis, treatment stratification, or monitoring of therapeutic response.

7. Limitations and conclusion

Overall, EEG studies on chronic pain face several methodological constraints that complicate inference and comparability. Most of the existing evidence remains cross-sectional and underpowered, thereby limiting causal inference regarding whether spectral/connectivity changes reflect drivers of pain or compensatory responses; longitudinal and interventional designs (analgesics, cognitive behavioral therapy [CBT], neuromodulation) with preregistered hypotheses are rare. The methods are heterogeneous and include band definitions, filter/artifact handling, sensor- and source-space choices, leakage control (e.g., wPLI/imaginary/lagged), and graph thresholds, which impede replication and meta-analysis. Furthermore, the reliability and external validation have seldom been reported. Moreover, anatomical precision is limited by template head models/regions of interest (ROIs), different EEG systems, and the number of channels rather than by individualized anatomy.

Another systematic review of resting-state quantitative electroencephalography (qEEG) in fibromyalgia faced similar limitations, as EEG findings remain heterogeneous and may differ across different clinical phenotypes. For example, Silva-Passadouro et al⁷⁶ reported trends toward reduced low-frequency activity and increased beta activity in fibromyalgia while highlighting methodological variability and the need for further phenotype-specific research.

To improve reproducibility, future studies should investigate their hypotheses and analytical plans, clearly describe all of the preprocessing steps, and provide precise definitions of all relevant frequency bands and connectivity metrics. When ethically feasible, authors should share codes and de-identified data. In addition, standardized reporting templates should be used for EEG acquisition, artifact correction, source reconstruction, and graph theory analysis. Multi-site studies should include harmonized EEG protocols and external validation cohorts.

Overall, the currently available evidence suggests that chronic pain may be associated with frequency-specific changes in EEG network dynamics across the theta, alpha, beta, and gamma bands. Within this hypothesis-generating framework, theta alterations may reflect biased engagement in learning or salience-related circuits. Changes in alpha bands may indicate altered inhibitory control and sensory gating. Additionally, beta abnormalities may reflect the maladaptive stabilization of pain-related states or reduced regulatory control, while gamma findings suggest disruptions in fast-timescale neural integration or sensory precision.

Instead of establishing a single EEG mechanism for chronic pain, the proposed framework provides a structured and testable model for interpreting EEG connectivity findings. This framework can guide future longitudinal, interventional, and phenotype-specific research by linking recurring theta, alpha, beta, and gamma alterations to candidate neuroplastic processes. Such studies would help to determine whether these oscillatory patterns are causal, compensatory, epiphenomenal, or clinically useful biomarkers. However, these interpretations are theoretical only. These limitations should be carefully considered because the current literature is limited by cross-sectional study designs, small sample sizes, methodological heterogeneity, and limited external validation. Future longitudinal, interventional, phenotype-specific, and reproducible studies are required to determine whether these EEG connectivity patterns represent causal mechanisms, compensatory responses, or clinically useful biomarkers of chronic pain.