EEG: How Electroencephalography Measures Brain Electrical Activity
Author: Ukrainian Psychological Hub · Published: October 12, 2026 · Editorial Policy
Electroencephalography (EEG) measures changes in electrical potential at the scalp that arise in part from coordinated activity in large neuronal populations. It is used to investigate seizures, sleep and altered consciousness, and to study the timing of perception, attention, language, memory and behavior. EEG is unusually sensitive to when changes occur, often on the scale of milliseconds. Its interpretation also depends on how fields propagate through the head, which neural sources contribute, the chosen electrode reference and the presence of artifacts.
An electroencephalogram is the recorded signal, while electroencephalography names the method. The term EEG commonly describes both the test and its output. In clinical care, scalp EEG is typically painless and noninvasive. The electrodes sense voltage differences; they do not administer an electrical treatment, detect neurotransmitter concentrations or image the brain's anatomy. Official explanations are available from NINDS and MedlinePlus.
How does EEG work?
A technologist or researcher places electrodes on specified regions of the scalp, usually with conductive paste or gel, or through a validated dry-electrode interface. Positions commonly follow the international 10–20 system, which uses proportional distances between anatomical landmarks. Clinical standardization of placement and labeling is described in the American Clinical Neurophysiology Society (ACNS) electrode nomenclature guideline.
The apparatus records differences in electric potential between electrodes or between selected scalp electrodes and a reference. Sensitive amplifiers, appropriate filtering and sampling turn these small fluctuations into channels plotted as voltage against time. Each channel can be reviewed with other channels to assess field distribution, changes across tasks, reactions to stimulation or patterns that may have clinical relevance.
Clinicians interpret the recording alongside symptoms, medications, age, sleep state, history and sometimes synchronized video. Researchers connect traces with carefully timed events and statistical contrasts. This distinction is crucial: a voltage fluctuation is a physical measurement, while a conclusion about cognition or illness is an inference supported by the measurement and additional evidence.
What exactly do scalp EEG electrodes measure?
EEG sensors measure potential differences produced by electric fields at the surface of the head. Much of the conventional scalp signal reflects the combined extracellular fields associated with synchronized postsynaptic currents in oriented cortical neuronal populations, particularly pyramidal cells. Neural membrane activity gives rise to current flow in conductive tissue; spatially aligned currents can summate into fields that reach the scalp. Reviews of these physiological generators include Babiloni and colleagues and Hallez and colleagues.
Action potentials and other currents are part of the physiology, but the typical EEG channel is a measure of a macroscopic population field, not a direct count of individual neurons firing. Neurons can be very active yet produce weak scalp signals when their fields cancel; coherent activity can generate a larger detectable field even without an extraordinary number of neurons being active.
The scalp recording therefore has spatial and physiological selectivity. Geometry, alignment, synchrony, depth and tissue conductivity determine how much a neural process contributes. EEG does not sample every neuron equally, and it is not an exhaustive readout of total brain activity.
Why can EEG detect the brain's electrical fields through bone?
Brain tissue, cerebrospinal fluid, skull and scalp conduct electricity differently. Neural electrical fields spread through this conducting volume, with attenuation and spatial distortion. Multiple underlying current sources may affect a single sensor, and one source may affect many electrodes. A strong scalp signal over a particular location need not be generated directly beneath that electrode.
This creates the EEG inverse problem. The forward calculation predicts scalp voltages for a proposed set of sources and a particular head model. Source localization reverses that relationship, estimating which source configuration might explain recorded data. Because different configurations can produce similar scalp measurements, these estimates depend on assumptions, regularization, sensor locations and the head model. Hallez et al. review source modeling, while Kayser and Tenke explain consequences of volume conduction and reference dependence.
High-density arrays, anatomical MRI and improved modeling can sometimes refine the estimates. A source map remains an inferred model of the underlying currents. It should not be described as a direct photograph of a single activated structure. The corresponding systems perspective is developed in our English article on neural circuits and brain networks.
What do EEG amplitude, frequency and brain waves mean?
An EEG trace charts electrical potential differences over time, typically measured in microvolts. Amplitude describes the magnitude of the measured change. Frequency indicates how many repetitions of an oscillatory component occur per second, expressed in hertz. Investigators may also describe phase, spatial distribution, event timing and whether a rhythm persists or changes with behavior. For a fuller review of these frequencies and their contextual meaning, see our brain waves and neural oscillations guide.
Conventional frequency bands often include delta below about 4 Hz, theta near 4–8 Hz, alpha near 8–13 Hz, beta near 13–30 Hz and gamma above roughly 30 Hz. These boundaries are conventions rather than rigid biological categories; definitions differ across studies and developmental periods. Posterior alpha activity during relaxed eyes-closed wakefulness is a classical finding, but it changes with opening the eyes, engagement and other conditions.
Frequency labels cannot by themselves identify emotions, thought content or a clinical disorder. Slow rhythms can be characteristic of sleep yet signal a different concern in some waking clinical circumstances. High-frequency scalp activity can be contaminated by muscle signals. Spectral power describes energy or variance in a frequency range under a particular analysis method, not how much dopamine or serotonin the brain contains. See the EEG and brain rhythms review.
Why is EEG so useful for studying timing?
EEG can capture rapid changes in neuronal population fields, giving it excellent temporal resolution. At a sampling rate of 500 Hz, samples are recorded every 2 milliseconds; at 1,000 Hz, every millisecond. The scientifically meaningful timing precision also depends on acquisition filters, accurate event markers, noise and the size of an experimental effect.
A short sampling interval does not imply that every millisecond represents one discrete cognitive step. Nor does knowing when a waveform changes establish where it originated. A signal may be temporally precise and spatially ambiguous at the same time.
Functional MRI usually infers neural activity from blood-oxygenation-dependent signals evolving over seconds. EEG measures electrical potential differences much closer in time to the neural events that produce them. The modalities are complementary: a temporal question and an anatomical localization question may require different measurements. A useful scientific comparison appears in Fundamentals of EEG, MEG and fMRI.
What are EEG electrodes, references and montages?
An electrode is a conductive contact that couples the recording electronics to a point on the scalp. Each EEG channel displays a difference between two electric potentials. The selected reference matters because the apparent voltage, polarity and distribution of a waveform depend on how that difference is calculated.
A bipolar montage compares adjacent or otherwise paired electrodes. A referential montage compares channels to a specified reference or combination. Average-reference and related mathematical transformations are common in research. Changing a montage can substantially change the visible trace without changing the patient's underlying brain activity.
Clinical professionals may examine several montages to clarify suspicious activity, rule out artifacts and examine field consistency. The internationally defined electrode placement and standard montage practices are presented in the ACNS guidelines. A sensor's position is a sampling location, not a unique assignment of one psychological function to the cortex immediately beneath it.
What happens during a clinical EEG test?
The clinician determines why the test is needed; a trained technologist prepares the scalp, positions the electrodes, checks contact quality and obtains the recording. A patient generally reclines, sits or lies quietly. The technologist may request eye opening and closing, periods of drowsiness or sleep, and selected activation procedures such as deep breathing or intermittent photic stimulation when appropriate.
The test duration depends on its purpose. Routine testing involves preparation in addition to the active recording; prolonged monitoring may run for hours or days. Results are assessed by a qualified clinician in the context of the referral question and history. The NINDS diagnostic testing guide and Cleveland Clinic outline common patient procedures.
EEG recording does not involve ionizing radiation or routinely deliver current into the brain. Conductive gel or adhesive can cause temporary inconvenience or mild skin irritation. Selected activation procedures can provoke symptoms in susceptible individuals, which is why they are conducted according to clinical protocols rather than performed unsupervised.
How should someone prepare for an EEG?
Preparation instructions differ by clinical indication. Clean, dry hair and avoiding styling products may facilitate electrode contact. Tell the care team about current medicines, recent sleep, caffeine consumption, possible seizure episodes and relevant medical conditions. Follow any tailored instructions about sleep, food or activity.
If a sleep-deprived study is planned, the facility specifies how much sleep to obtain beforehand and any transport precautions. Patients should not stop prescribed medicines simply to change the recording; medication adjustments require explicit medical instruction. After a routine examination, gel may need to be washed out.
Routine EEG, ambulatory EEG and video EEG
Routine EEG samples a relatively brief interval. It may reveal epileptiform discharges, background abnormalities or a seizure that happens to occur during the recording. It cannot observe an event occurring hours later or guarantee that an intermittent pattern will be captured.
Ambulatory EEG records over an extended period, including selected everyday activities. Longer sampling may help when the question concerns occasional episodes, although movement and electrode artifacts can complicate analysis. Video EEG adds synchronized visual behavior, allowing a specialist to compare the appearance of an episode with its electrical correlate. In specialized hospital units, this pairing can be especially valuable when characterizing events and evaluating seizure disorders.
Sleep EEG records physiological activity during sleep or attempted sleep. Polysomnography combines EEG with other measures, potentially including eye movements, muscle activity, breathing and oxygenation. The interpretation of sleep stages or sleep-related disorders follows the full recording configuration, not a lone waveform. NINDS describes these distinct recording settings.
What can EEG reveal about epilepsy and seizures?
EEG is a major tool in seizure evaluation. Specialists may identify patterns considered epileptiform, capture an electrographic seizure, characterize background activity and relate findings to the reported events. Such evidence can support diagnosis and clinical decision-making when interpreted by an appropriately qualified professional.
A normal EEG does not rule out epilepsy. The recording may miss intermittent activity or a seizure may be generated in a location with limited scalp detectability. Likewise, an isolated abnormal waveform does not automatically demonstrate that an individual has epilepsy. The medical question, history and full semiology of the episode remain essential. The MedlinePlus EEG overview explicitly addresses this limitation.
Continuous monitoring may reveal electrographic seizures in critically ill patients who have few obvious external signs. In some patients evaluated for epilepsy surgery, invasive electrodes are used to record closer to candidate seizure-generating tissue. These are specialized procedures with a different balance of spatial information, coverage and risk from ordinary scalp EEG.
EEG in intensive care, encephalopathy and anesthesia
EEG can contribute to assessing altered mental status, certain encephalopathies, seizures without prominent physical movements and changes during critical illness. Continuous monitoring permits more sustained observation than a short outpatient study. Interpretation requires careful attention to medications, sedation, temperature, systemic illness and other possible causes of EEG changes.
EEG-derived measures may also support aspects of anesthesia monitoring. An algorithm's single index, however, is a processed representation of selected signals, and neither an entire measure of the brain's state nor a direct substitute for clinical assessment. Processing choices and pharmacological context affect its relevance.
Can EEG determine brain death?
The 2023 US multisociety pediatric and adult brain-death/death-by-neurologic-criteria consensus guideline does not accept EEG as an ancillary test for establishing brain death. EEG records cortical electrical activity but does not adequately evaluate brainstem function. The determination requires the prescribed clinical process, with approved ancillary testing only under the applicable conditions. The primary standard is the American Academy of Neurology-led consensus guideline.
Statements such as 'a flat EEG proves brain death' are therefore incorrect under this guideline. A very low-amplitude or apparently inactive tracing also requires technical scrutiny and cannot replace the specified neurological examination and protocol.
What are event-related potentials in EEG research?
Event-related potentials, or ERPs, are measurable voltage patterns temporally related to events such as a flash, a spoken word, an error or a decision cue. Researchers align data to the event and often average repetitions to estimate the consistent event-related response against ongoing background variation.
Common ERP names include P300, often studied in selective attention and context updating; mismatch negativity, associated with responses to deviations in regularity; and N400, widely investigated in processing meaning. The response depends on the paradigm, recording and analysis. A component is neither a single brain area nor a one-to-one marker of a unique mental faculty.
The influential synthesis by Kutas and Federmeier examines what N400 research reveals about semantic processing. Helfrich and Knight review event-related methods and mechanisms, while a clinical methodological review identifies limits in translation to individual diagnostics. A robust group-level experimental effect does not automatically serve as a validated diagnostic test for a single person.
EEG and sleep: what can the rhythms establish?
Oscillatory activity changes with vigilance, development and sleep state. The analysis of sleep-related EEG helps characterize sleep architecture when integrated with relevant physiological channels. A slow rhythm that is typical in one sleep stage may carry a very different meaning in a waking clinical recording.
Popular formulas that associate theta with creativity, alpha with relaxation, beta with concentration and gamma with conscious awareness are incomplete and potentially misleading. The same frequency band can be present in multiple psychological contexts. The measurable scientific question is how a predefined signal feature varies across carefully compared conditions and populations.
What is EEG source localization?
EEG source localization uses a mathematical model to estimate the brain generators compatible with recorded scalp potentials. It combines sensor measurements with assumptions about possible source locations, tissue conductivity and source orientation. With dense sensors and subject-specific anatomical information, some estimates become more spatially informative.
The inverse solution is not uniquely determined from the measured scalp voltages alone. Source-estimation software may generate apparently precise color images whose precision exceeds what the data justify. Validation against independent techniques and sensitivity analysis of modeling choices are therefore essential. The EEG source-analysis review describes these issues, and Kayser and Tenke explain volume-conduction effects.
Why artifacts can resemble brain signals
Scalp sensors also detect voltage changes caused by eye blinks, eye movement, jaw or neck muscles, heartbeat, electrode motion, cable problems and environmental electrical interference. Some artifacts can be far larger than the neural signal of interest. Muscle activity can particularly complicate high-frequency interpretation.
Preprocessing may include filtering, correcting poorly functioning channels, rejecting contaminated intervals, re-referencing and component-based artifact identification. Independent component analysis is useful in many settings but is not a universally reliable separator of neural and non-neural activity. Overaggressive preprocessing can remove genuine brain signals or distort the shapes and timing of waves.
Scientific reports should describe acquisition parameters, exclusion criteria, filtering and artifact procedures. The peer-reviewed EEG artifact removal review discusses competing techniques and their limitations. A visually smooth EEG plot alone does not demonstrate that a recording is valid.
What do EEG time-frequency and connectivity analyses mean?
Time-domain methods compare voltages and waveform characteristics across intervals and experimental conditions. Spectral analyses estimate the distribution of oscillatory power across frequencies, while time-frequency methods track how such distributions change. Results depend on the reference, the analysis window, baseline conventions, preprocessing and statistical choices.
Connectivity measures may evaluate whether signals from different locations show related fluctuations or phase patterns. Volume conduction can create statistical associations between scalp electrodes even when the underlying neuronal populations are not directly communicating. Shared inputs, artifacts and filtering also affect connectivity estimates.
Accordingly, EEG coherence or phase coupling does not independently prove that two named brain regions causally exchange information. Research on networks requires converging methods, explicit controls and replication, as explained by the physics of scalp fields and the volume conduction review.
EEG compared with MRI, fMRI, PET, MEG and ECG
Structural MRI primarily describes tissue anatomy. Most fMRI measures blood-oxygenation-related signals associated indirectly with neuronal activity. PET detects emissions from administered radiotracers to address selected metabolic or molecular questions. MEG detects magnetic fields associated with neural currents. EEG detects electric potential differences at the scalp. For the magnetic counterpart of these measurements, explore our MEG methods guide.
EEG is portable in many implementations, is often cheaper than large scanning systems and is useful for temporally fine-grained or prolonged measurements. Its anatomical localization is limited by source mixing. MEG is also temporally precise but differs in field sensitivity and technical demands; fMRI offers complementary spatial and hemodynamic information. See the methods overview and our comparison of MRI, fMRI, PET and EEG.
ECG, or electrocardiography, also records changing electrical potential but primarily measures cardiac activity. Its waveforms are associated with the heart, whereas EEG's target is neural population activity. EEG may contain cardiac interference, which should be recognized as such.
EEG versus electrocorticography and intracranial EEG
Electrocorticography records electrical signals from electrodes placed on or near the cerebral cortex, while depth-electrode recording methods such as stereoelectroencephalography can sample selected intracranial regions. These techniques may provide signals less spatially blurred by the skull and are used for particular clinical questions, especially some presurgical epilepsy evaluations.
They also involve invasive procedures and deliberately limited spatial coverage. Greater proximity to particular neural tissue does not mean that every brain structure or every neuron is observed. The clinical indication determines when their additional information is worth the accompanying risk. The NINDS guide describes intracranial recordings in epilepsy care.
Can EEG measure attention, emotions or intelligence?
Researchers can detect electrophysiological features associated with aspects of perception, attention, language processing, decision tasks and arousal. With a controlled experiment, a difference between responses can support the conclusion that a particular task manipulation affects the measured neural process at a particular time.
An observed correlate is not a direct universal measurement of attention, intelligence or emotion. Similar scalp patterns can result from different processes, and each psychological construct involves context and multiple mechanisms. Conventional EEG is not an intelligence test and cannot independently establish a psychiatric diagnosis from a proprietary concentration or stress score.
The distinction between experimental association and individual diagnosis is particularly important for vulnerable people. The clinical usefulness of any proposed EEG biomarker must be demonstrated in its intended population, with credible sensitivity, specificity, calibration, external validation and evidence of patient benefit.
Can EEG read minds or control a computer?
Some brain-computer interfaces use event-related or oscillatory EEG features to select symbols, assist communication or control predefined actions. Such systems typically rely on calibration, restricted choice sets, task training and signal-processing algorithms. They may achieve useful function for selected users, but the decoded output is a constrained estimate of a designed command rather than an unrestricted transcript of private thinking.
Machine learning can also classify EEG signals in medical research. High accuracy in one dataset may reflect patient-specific patterns, recording conditions or training-test leakage, so external performance must be demonstrated. A 2026 systematic review of EEG-based real-time seizure prediction reported that cross-dataset external validation was lacking in the included studies and that operational metrics were often incompletely reported.
Are consumer EEG headbands reliable?
Consumer headsets can measure meaningful electrical features in some conditions and are useful for education, interface experiments and certain research questions. Their quality varies with electrode count, placement, contact method, movement tolerance and access to raw data. A 2024 scoping review documented extensive consumer EEG use in published research.
Validation varies considerably. A 2025 device-comparison study identified signal-quality differences across tested products. A 2026 evaluation found that several consumer devices could detect expected alpha-related features under a structured protocol. Measuring a known rhythm in controlled conditions does not independently validate an app's prediction of anxiety, attention or cognitive performance.
Proprietary 'focus' or 'meditation' metrics should be judged against independent evidence of what they predict, across settings and users. Users should also review how neural recordings and derived behavioral scores are stored, processed and shared.
What is EEG neurofeedback, and what does the treatment evidence show?
Neurofeedback presents selected brain-signal features as feedback that a person tries to change. The ability to influence a feedback score and the ability to improve a clinically meaningful outcome are distinct empirical questions.
A 2025 systematic review and meta-analysis in JAMA Psychiatry pooled 38 randomized clinical trials with 2,472 participants for ADHD neurofeedback. In probably blinded outcome assessments, neurofeedback did not produce a meaningful average improvement in core ADHD symptoms as a stand-alone treatment; selected subgroup or processing-speed results were more favorable but limited. Results from one condition cannot establish effectiveness for unrelated disorders, and consumer EEG scores are not evidence of treatment efficacy.
Established findings, emerging methods and unresolved questions
Established findings include the physiological basis of scalp potentials in neuronal population currents, EEG's high temporal resolution, the prominence of volume conduction and artifacts, and the value of EEG in appropriately selected seizure and critical-care evaluation. These conclusions are supported by longstanding physiology, clinical standards and reproducible research.
More qualified findings include precise EEG source imaging in difficult geometries, decoding complex individual psychological states from scalp data, transferable machine-learning biomarkers, and the clinical utility of various consumer and neurofeedback systems. Some are scientifically promising, but each specific use requires independent performance evaluation. A result from a laboratory dataset should not be promoted into an individual clinical test by description alone.
Important open problems include robust artifact handling in natural environments, reliable multimodal source estimation, generalization across people and devices, longitudinal stability, neural-data privacy and methods for distinguishing genuine functional interactions from shared field effects. Improving the sophistication of an algorithm does not eliminate the physical properties of the original measurement.
Common EEG myths
'EEG records every neuron firing.' Conventional scalp EEG records macroscopic field contributions from populations, shaped by geometry and synchrony.
'The electrode shows what the brain region underneath it is doing.' Electric fields spread and mix through the head, so scalp location does not uniquely identify a neural generator.
'Each brain-wave band corresponds to one mental state.' Frequency ranges are descriptive; context, sleep state, age, task and artifact determine their interpretation.
'A normal EEG rules out epilepsy.' Seizures and interictal activity may be absent from the sampled period or difficult to detect at the scalp.
'EEG directly measures neurotransmitter levels.' Electrical potential differences do not directly determine molecular concentrations.
'A wearable EEG is an automatic mental-health diagnostic tool.' Diagnostic validity must be established for the device, algorithm, population and intended clinical purpose.
'EEG can reconstruct private thoughts.' Restricted experimental classification does not establish unconstrained thought reading.
'An electrically silent EEG independently proves brain death.' The current US brain-death guideline excludes EEG as an acceptable ancillary test; see Greer et al..
Frequently asked questions
Is an EEG painful or dangerous?
Routine scalp EEG is generally painless and low risk. Electrodes measure voltage without routinely introducing electrical current into the brain. Minor skin irritation is possible, and particular clinical activation procedures are used with appropriate precautions.
What is the difference between EEG and an electroencephalogram?
EEG is used for both electroencephalography, the method, and an electroencephalogram, the recording made by that method.
Can EEG diagnose epilepsy?
EEG can provide important supporting evidence, reveal epileptiform patterns or capture seizures. Diagnosis requires interpretation with the person's symptoms and history, and a normal routine EEG does not exclude epilepsy.
Can someone sleep during EEG?
Yes. Some recordings intentionally include drowsiness or sleep. A sleep study may also use EEG together with other physiological measurements.
How long does an EEG take?
The duration depends on the indication and recording protocol. A routine appointment often takes around an hour including preparation, while ambulatory or inpatient monitoring may continue for much longer.
Can EEG detect anxiety, depression or ADHD?
Some research reports group-level electrophysiological associations, but a standalone routine EEG is not a validated individual diagnostic substitute for standard clinical assessment of these conditions.
Can EEG locate the exact origin of a brain signal?
Methods can estimate plausible sources using mathematical and anatomical assumptions. Scalp measurements alone do not uniquely specify the true underlying generator.
Is EEG more accurate than fMRI?
They measure different physiological quantities. EEG has high temporal precision, whereas fMRI provides spatially informative blood-oxygenation measures. Accuracy depends on the question and the method used.
Can EEG identify a person's thoughts?
EEG-based systems can sometimes classify restricted tasks or choices after calibration. They cannot reconstruct arbitrary private thoughts from an ordinary scalp recording.
Do EEG waves mean the same thing in every person?
No. Age, sleep, medication, anatomy, signal quality, reference choices and task context affect waveform appearance and interpretation.
Why are eye blinks visible on EEG?
The eye and its movement create electrical fields that reach scalp sensors. These non-neural contributions may be much larger than some of the brain signals under study.
Does an abnormal EEG automatically mean a serious disorder?
No. An abnormal finding may be nonspecific or artifactual, and interpretation requires the pattern's characteristics, clinical history and qualified review.
Conclusion
EEG provides a temporally precise view of changes in electric potential recorded at the scalp. It can answer important questions about neural timing and can provide clinically useful evidence when the recording protocol and interpretation match the medical problem. Its signals are composite, reference-dependent and sensitive to artifacts; localization, diagnosis and psychological interpretation therefore require explicit evidence and carefully bounded inference.
The most reliable EEG knowledge integrates physiology, standardized acquisition, experimental design, context, transparent analysis and independent validation. Those foundations make EEG a powerful tool in behavioral and cognitive neuroscience while keeping its conclusions aligned with what it actually measures.
Related Articles
References
Acharya, J. N., Hani, A., Cheek, J., Thirumala, P., & Tsuchida, T. N. (2016). Guidelines for standard electrode position nomenclature. American Clinical Neurophysiology Society, Guideline 2. Official PDF.
American Clinical Neurophysiology Society. Guidelines and consensus statements. Official standards.
Babiloni, C., et al. (2009). Fundamentals of electroencephalography, magnetoencephalography and functional magnetic resonance imaging. International Review of Neurobiology. DOI: 10.1016/S0074-7742(09)86005-4.
Cleveland Clinic. (2025). Electroencephalogram (EEG): What it is, procedure and results. Clinical overview.
Greer, D. M., et al. (2023). Pediatric and adult brain death/death by neurologic criteria consensus guideline. Neurology, 101, 1112–1132. DOI: 10.1212/WNL.0000000000207740.
Hallez, H., et al. (2007). Review on solving the forward problem in EEG source analysis. Journal of NeuroEngineering and Rehabilitation, 4, 46. DOI: 10.1186/1743-0003-4-46.
Helfrich, R. F., & Knight, R. T. (2019). Cognitive neurophysiology: Event-related potentials. Handbook of Clinical Neurology, 160, 543–558. DOI: 10.1016/B978-0-444-64032-1.00036-9.
Kayser, J., & Tenke, C. E. (2012). Generator localization by current source density: Implications of volume conduction and field closure at intracranial and scalp resolutions. Clinical Neurophysiology, 123, 2328–2345. DOI: 10.1016/j.clinph.2012.06.005.
Kutas, M., & Federmeier, K. D. (2011). Thirty years and counting: Finding meaning in the N400 component of the event-related brain potential. Annual Review of Psychology, 62, 621–647. DOI: 10.1146/annurev.psych.093008.131123.
MedlinePlus Medical Encyclopedia. EEG. U.S. National Library of Medicine. Official source.
National Institute of Neurological Disorders and Stroke. Neurological diagnostic tests and procedures. National Institutes of Health. Official source.
Sabio, J., Williams, N. S., McArthur, G. M., & Badcock, N. A. (2024). A scoping review on the use of consumer-grade EEG devices for research. PLOS ONE, 19, e0291186. DOI: 10.1371/journal.pone.0291186.
Song, Z., Arrowsmith, K., Henare, D., & Pedersen, M. (2026). Real-time EEG-based epileptic seizure prediction using artificial intelligence: A systematic review. Artificial Intelligence in Medicine. DOI: 10.1016/j.artmed.2026.103476.
Urigüen, J. A., & Garcia-Zapirain, B. (2015). EEG artifact removal—State of the art and guidelines. Journal of Neural Engineering, 12, 031001. DOI: 10.1088/1741-2560/12/3/031001.
Westwood, S. J., et al. (2025). Neurofeedback for attention-deficit/hyperactivity disorder: A systematic review and meta-analysis. JAMA Psychiatry, 82, 118–129. DOI: 10.1001/jamapsychiatry.2024.3702.