Electroencephalography, or EEG, remains one of the most widely used tools in neuroscience because it does something no other technique can match: it tracks brain activity as it happens, down to the millisecond. Since its invention a century ago, EEG has become indispensable in fields ranging from epilepsy diagnosis to cognitive psychology to brain-computer interface design. Its role in scientific inquiry is not simply historical; it continues to expand as new analytical methods, portable hardware, and machine-learning algorithms squeeze more information out of the electrical signals the brain naturally produces.
A Century of Recording Brain Waves
In 1924, a German psychiatrist named Hans Berger placed electrodes on a patient’s scalp at the University Hospital in Jena and recorded the first human brain electrical signals ever documented.1PubMed. Hans Berger (1873-1941): the German psychiatrist who recorded the first electrical brain signal in humans 100 years ago Berger coined the term “alpha waves” for the roughly 10-cycle-per-second rhythm that appeared clearly in many of his recordings, especially when the person’s eyes were closed.2PubMed Central. Forgotten rhythms? Revisiting the first evidence for rhythms in cognition He also foresaw clinical uses, including applications in patients with what was then called “senile dementia,” a vision that researchers have revisited extensively in the context of Alzheimer’s disease.3PubMed. Alpha rhythm and Alzheimer’s disease: Has Hans Berger’s dream come true?
What made Berger’s work transformative was not the technology itself, which was crude by modern standards, but the proof of principle: the brain’s electrical activity could be detected through the intact skull, non-invasively, in a living person. That insight opened the door to an entire field. Within decades, EEG was being used in operating rooms, sleep laboratories, and research centers around the world.
How EEG Captures Brain Activity
An EEG records tiny voltage fluctuations at the scalp, generated primarily by the synchronized activity of large populations of neurons in the brain’s outer layers. When thousands of neurons fire together in a coordinated way, the electrical fields they produce sum up enough to be detected by electrodes on the head’s surface. The resulting signals are small, typically measured in millionths of a volt, and they represent the aggregate behavior of neural populations rather than the firing of individual cells.
The standard method for placing electrodes is the International 10-20 system, which divides the scalp into proportional distances based on bony landmarks on the skull. The “10” and “20” refer to the percentages of those distances used to position each electrode site.4PubMed. Using the international 10-20 EEG system for positioning of transcranial magnetic stimulation Imaging studies using CT scans have confirmed that these electrode positions correspond reasonably well to specific brain areas mapped by classical anatomy, though some variability across individuals is unavoidable.5Electroencephalography and Clinical Neurophysiology. Cerebral location of international 10–20 system electrode placement More recent updates have expanded the original 21-electrode array into denser configurations based on the 10-10 system, increasing spatial coverage.6PubMed. The standardized EEG electrode array of the IFCN
What Brain Oscillations Tell Us
The most fundamental feature of the EEG signal is that it oscillates. The brain produces rhythmic activity at multiple frequencies simultaneously, and researchers have learned that different frequency bands tend to track different mental states and cognitive operations. Most cognitive processes have been linked to at least one of the traditional frequency bands: delta, theta, alpha, beta, and gamma.7PubMed. EEG oscillations: From correlation to causality
Delta waves are the slowest, associated with deep sleep and certain aspects of speech processing. Theta rhythms show up during memory encoding, navigation, and focused concentration. Alpha waves, the ones Berger first described, tend to increase when you close your eyes or relax and decrease when you actively engage with a task. Beta waves are linked to active thinking, motor planning, and alertness. Gamma waves, the fastest, have been connected to higher-order processes like binding sensory information together into a unified perception.
These bands do not operate in isolation. Research has shown that delta, theta, alpha, and gamma oscillations act as communication networks across large populations of neurons, playing a major role in memory and integrative brain functions.8PubMed. Gamma, alpha, delta, and theta oscillations govern cognitive processes A study of auditory sentence processing, for example, found that different bands indexed different levels of language comprehension: delta and theta oscillations tracked the sound-level analysis of speech, while gamma activity indexed meaning-level processing such as retrieving words from memory.9PubMed. Delta, theta, beta, and gamma brain oscillations index levels of auditory sentence processing This kind of finding illustrates why EEG is so useful for studying cognition: it can reveal the brain’s moment-by-moment engagement with a task in a way that slower imaging methods cannot.
Event-Related Potentials and the Timing of Thought
Beyond looking at ongoing oscillations, researchers can time-lock EEG recordings to specific events, such as the moment a person sees an image or hears a tone, and average across many repetitions. The result is an event-related potential, or ERP, a waveform that reveals the sequence of neural processing stages triggered by that event.10PubMed Central. Event-related potential: An overview
ERPs are the only non-invasive method that can resolve the dynamic pattern of brain events down to the millisecond range.11PubMed. Event-related potentials of the brain and cognitive processes: approaches and applications Early components of the ERP, appearing within the first couple hundred milliseconds, reflect basic sensory processing. Later components reveal higher-level operations. The P300 wave, a positive voltage peak roughly 300 milliseconds after a stimulus, has been studied extensively as a marker of attention reallocation and memory updating. Research has shown that the timing of earlier processing stages predicts the timing of later ones: if the initial visual analysis happens quickly, the subsequent steps of discrimination, attention updating, and motor response tend to follow sooner as well.12PubMed Central. Relationship between early and late stages of information processing: an event-related potential study
This millisecond precision is EEG’s greatest scientific strength. Brain imaging methods that rely on blood flow, like fMRI, take measurements on the order of seconds. EEG can distinguish between processes that happen 50 milliseconds apart, which matters enormously when you want to understand how the brain sequences its operations during something as rapid as reading a word or detecting a threat.
Clinical Workhorse Applications
EEG’s longest-running clinical role is in epilepsy. Because seizures are fundamentally electrical disturbances, EEG can detect them directly. Even between seizures, characteristic spikes and sharp waves in the EEG support a diagnosis of epilepsy when a seizure itself is not captured during recording. Continuous monitoring in intensive care settings has become essential for identifying non-convulsive seizures, which produce no visible symptoms but can cause ongoing brain injury.13PubMed. Electroencephalography in Epilepsy Evaluation
Sleep medicine depends on EEG just as heavily. The standard method for classifying sleep stages relies on interpreting the patterns of EEG signals alongside eye-movement and muscle-activity recordings. Each 30-second segment of the night is scored into a sleep stage: wakefulness, three stages of non-REM sleep (N1, N2, and N3, which is deep sleep), and REM sleep. This staging system, formalized by the American Academy of Sleep Medicine, provides objective markers for evaluating sleep complaints and diagnosing disorders.14PubMed Central. Spotlight on Sleep Stage Classification Based on EEG Measures like sleep efficiency, which is the percentage of time in bed actually spent sleeping, and the arousal index, which counts how often sleep is disrupted per hour, are all derived from EEG-based scoring.15Scientific Reports. Validation of sleep-staging accuracy for an in-home sleep electroencephalography device compared with simultaneous polysomnography in patients with obstructive sleep apnea
Anesthesiology is another area where EEG monitoring has grown. Assessing how deeply a patient is under general anesthesia is tricky, and commercial devices have been developed that analyze brain wave patterns to estimate the depth of unconsciousness in real time.16PubMed Central. Monitoring the depth of anaesthesia
Studying Working Memory and Cognition
EEG is not only a clinical tool; it plays a central role in cognitive neuroscience experiments. A common paradigm uses EEG to measure working memory, the brain’s ability to hold and manipulate information over short intervals. Research consistently shows that frontal theta power and parietal alpha power change in predictable ways as working memory load increases. Interestingly, these signatures are not purely cognitive: they can be influenced by the emotional content of the material being held in mind.17PubMed Central. Electroencephalography Based Analysis of Working Memory Load and Affective Valence in an N-back Task with Emotional Stimuli More detailed analyses have found that frontal theta power during the retention period of a memory task correlates with how precisely participants remember stored items, suggesting the signal is tracking actual memory quality, not just effort.18Scientific Reports. The association between working memory precision and the nonlinear dynamics of frontal and parieto-occipital EEG activity
EEG has also been used to study attention, language processing, decision-making, and developmental changes across the lifespan. In ADHD research, for instance, studies tracking children into adulthood have found shifting EEG profiles: children with ADHD tend to show elevated slow-wave activity (delta and theta) with reduced alpha, while adults with the disorder show a somewhat different pattern, with diminished frontal delta and persistently elevated global theta.19PubMed. EEG development in Attention Deficit Hyperactivity Disorder: From child to adult Findings like these help researchers understand how brain maturation interacts with clinical conditions over time.
Brain-Computer Interfaces
One of the most striking applications of EEG research is the development of brain-computer interfaces, or BCIs. These systems allow a person to control a device using brain signals alone, bypassing the body’s muscles entirely. EEG-based BCIs have been used to control computer cursors, drive word-processing software, operate robotic limbs, access the internet, and manage devices in the home environment. For people with severe motor disabilities, such as advanced amyotrophic lateral sclerosis, this technology can re-establish a degree of independence that would otherwise be impossible.20The Lancet Neurology. Brain–computer interfaces in neurological rehabilitation
BCI research also has implications for rehabilitation after stroke or traumatic brain injury. By feeding back information about the brain’s current activity state, EEG-based systems may help guide the brain’s own plasticity processes, essentially helping rewire damaged circuits by showing the patient what their brain is doing and training them to modify it. The practical engineering side of BCIs is advancing rapidly, with a shift toward dry electrodes that do not require conductive gel or extensive skin preparation, making the devices more comfortable and portable for use outside the laboratory.21PubMed Central. Recent Advances in Portable Dry Electrode EEG: Architecture and Applications in Brain-Computer Interfaces
Cleaning Up the Signal
One persistent challenge in EEG research is that the brain’s electrical signals are mixed in with noise from other sources. Eye blinks, eye movements, muscle tension in the jaw and forehead, heartbeat artifacts, and even ambient electrical interference from nearby equipment all contaminate the recording. Raw EEG data almost always needs cleaning before it can be analyzed meaningfully.
A breakthrough in this area came with independent component analysis, or ICA, a mathematical technique that separates the mixed recording into statistically independent components. Some of those components correspond to brain activity and some to artifacts like eye movements or heartbeats. Researchers can identify the artifact components and reconstruct the signal without them. Studies applying ICA to standard clinical EEG recordings demonstrated that it could effectively remove artifacts from heart rhythm, eye movements, muscle tension, and electrical interference, all without the drawbacks of traditional digital filters.22PubMed. Independent component analysis as a tool to eliminate artifacts in EEG: a quantitative study Earlier work showed that ICA could isolate pure eye activity in recordings and subtract it cleanly, reducing the amount of genuine brain signal that gets accidentally thrown out during artifact removal.23PubMed. Extraction of ocular artefacts from EEG using independent component analysis
The Inverse Problem and Spatial Resolution
EEG’s Achilles’ heel has always been spatial resolution. While it tracks brain activity with exquisite timing, pinpointing exactly where in the brain that activity originates is much harder. The signal at each scalp electrode reflects the summed electrical activity from many brain regions, distorted as it passes through the skull and other tissues. Working backward from the scalp recording to figure out which brain sources generated it is known as the inverse problem, and mathematically, it has no unique solution: an infinite number of source configurations could produce the same pattern on the scalp.24PubMed Central. Review on solving the inverse problem in EEG source analysis
Researchers handle this by making simplifying assumptions about the likely sources. Some models assume the signal comes from a small number of discrete sources (dipole models), while others distribute estimated activity across the entire cortical surface (distributed source models). Both require an accurate model of the head’s anatomy to account for how electrical signals spread through brain tissue, cerebrospinal fluid, skull, and skin.25PubMed Central. Electroencephalography source localization These approaches have improved dramatically with modern neuroimaging and computational power, but the fundamental limitation remains: EEG will never localize activity as precisely as fMRI can.
Combining EEG with Other Imaging Methods
The natural solution to EEG’s spatial weakness is to pair it with techniques that excel at localization. Simultaneous EEG-fMRI, in which both recordings are taken at the same time, has matured into a well-established research method over the past 25 years.26PubMed Central. Simultaneous EEG-fMRI: What Have We Learned and What Does the Future Hold? The idea is straightforward: let the fMRI tell you where things are happening and let the EEG tell you when. In practice, combining them introduces technical headaches, since the fMRI’s strong magnetic field induces artifacts in the EEG, but solutions have been developed and the approach is now used routinely in epilepsy research, sleep studies, and cognitive neuroscience.
Analysis of simultaneous recordings can go in both directions. An EEG-informed fMRI analysis uses the brain’s electrical timing information to select moments of interest in the fMRI data. An fMRI-informed EEG analysis uses the spatial maps from fMRI to constrain the source-localization problem, narrowing down the infinite possible sources to the ones that are plausible given the fMRI results.27Frontiers in Systems Neuroscience. Simultaneous electroencephalography-functional magnetic resonance imaging for assessment of human brain function Combining EEG with magnetoencephalography (MEG), which detects the magnetic counterpart of the brain’s electrical signals, offers another path. MEG and EEG together have been shown to improve the ability to localize brain activity compared with either method alone.28PubMed Central. The advantage of combining MEG and EEG: comparison to fMRI in focally stimulated visual cortex
EEG as a Psychiatric Biomarker
One of the more active frontiers in EEG research is the search for biomarkers that could help diagnose psychiatric conditions. Unlike epilepsy, where EEG abnormalities are dramatic and directly diagnostic, psychiatric disorders tend to produce subtle shifts in the EEG that overlap considerably across conditions. Still, patterns are emerging. In depression, one well-studied marker is frontal alpha asymmetry, a difference in alpha power between the left and right frontal regions. However, recent research suggests this marker is not as consistent as early studies implied.29Journal of Yeungnam Science. Advances, challenges, and prospects of electroencephalography-based biomarkers for psychiatric disorders: a narrative review – Section: Investigating potential electroencephalography biomarkers for major psychiatric disorders: insights into depression, bipolar disorder, and schizophrenia More recently, increased gamma-band power has been proposed as a novel depression biomarker, along with complexity measures that suggest the depressed brain shows different patterns of signal irregularity.
In schizophrenia, the EEG profile tends to include excess slow-wave activity in the delta and theta bands and reduced alpha power, pointing to disrupted cortical inhibition. Machine-learning models trained on these kinds of features have shown promising classification accuracy. One study using supervised learning achieved roughly 73% accuracy in distinguishing between schizophrenia patients and healthy controls, with reduced central alpha power emerging as the most consistently predictive feature for both schizophrenia and major depression.30Schizophrenia Bulletin. EEG-based Signatures of Schizophrenia, Depression, and Aberrant Aging: A Supervised Machine Learning Investigation Another approach using dynamic functional connectivity, which looks at how brain-network patterns shift over time rather than treating the recording as a static snapshot, achieved about 73% accuracy in a four-way classification among healthy controls, non-psychotic depression, psychotic depression, and schizophrenia.31Molecular Psychiatry. Resting-state EEG dynamic functional connectivity distinguishes non-psychotic major depression, psychotic major depression and schizophrenia
These accuracy levels are not yet high enough for stand-alone diagnosis, but they are well above chance and improving. The appeal of EEG-based diagnostics is that EEG is inexpensive, widely available, and completely non-invasive, which makes it a realistic candidate for routine clinical screening in a way that brain scans involving large magnets or radioactive tracers are not.
Machine Learning and Large-Scale EEG Data
The explosion of deep learning has changed what researchers can extract from EEG data. Traditional analysis required human experts to define which features of the signal mattered, then build statistical models around those features. Deep-learning architectures can learn features directly from raw data. Reviews of the field have found that convolutional neural networks and recurrent neural networks consistently outperform older approaches in EEG classification tasks.32PubMed. Deep learning for electroencephalogram (EEG) classification tasks: a review This matters for practical applications: robust automatic classification of EEG signals is a critical step toward making EEG-based tools less dependent on trained specialists and more usable in everyday clinical and consumer settings.
Privacy and Ethics of Brain Data
As EEG moves out of the research lab and into consumer devices, wearable headbands, meditation apps, and gaming peripherals, questions about the ethics of collecting brain data are becoming urgent. A systematic review of ethical challenges in EEG neurofeedback identified seven recurring concerns: informed consent issues, psychological risks, equitable access, the ethics of placebo-controlled study designs, data privacy and governance of brain-derived information, vulnerabilities in specific populations like children, and neuroethical questions around using the technology for cognitive enhancement rather than treatment.33PubMed Central. Ethical challenges in EEG neurofeedback: a systematic review of gaps, risks, and responsibilities
The data privacy question is particularly thorny. EEG recordings carry information about a person’s mental states, cognitive patterns, and neurological health. As algorithms improve, the amount of personal information extractable from a brain recording is likely to grow. Unlike a blood test result, which tells you about your chemistry at one point in time, EEG data can contain temporally rich, continuous information about how your brain processes the world. Whether existing privacy frameworks are adequate for this kind of data is an open and unresolved question, one that technologists and policymakers are only beginning to address seriously.