Epilepsy and the Brain: What EEG Reveals

The Storm Inside Your Skull

Imagine a stadium with 86 billion people in it. Under normal circumstances, they're having millions of small, organized conversations. Groups of a few thousand coordinate to solve a problem. Clusters in one section start a chant that ripples across to another section. The noise is constant, but it's structured. Meaningful. Productive.

Now imagine that one section of the stadium suddenly stops having individual conversations and begins screaming the same word, in perfect unison, at the top of their lungs. That synchronized roar overwhelms the normal conversations around it. It spreads. More sections join the unison screaming. Within seconds, tens of thousands or even millions of people are locked into the same rhythmic shout, drowning out everything else.

That is a seizure. And the stadium is your brain.

Epilepsy is a neurological condition defined by a tendency to have recurrent seizures. About 50 million people worldwide have it, making it one of the most common neurological disorders on the planet. And the tool that has defined our understanding of epilepsy for nearly a century, the only technology that can watch a seizure happen in real time, is EEG.

When Hans Berger recorded the first human EEG in 1929, one of the very first things he noticed was that certain patients produced dramatically abnormal electrical patterns. Huge, rhythmic voltage spikes that looked nothing like the gentle oscillations of a healthy brain. He was looking at epileptic activity, and he immediately understood the significance: the brain's electrical misbehavior could be observed, measured, and studied.

Nearly a hundred years later, EEG remains the single most important tool in epilepsy diagnosis, classification, and management. No other technology comes close. And understanding what EEG actually reveals about the epileptic brain is one of the most fascinating stories in all of neuroscience.

What Makes a Brain Epileptic?

To understand seizures, you first need to understand the balance that keeps them from happening in a healthy brain.

Every moment of every day, your brain maintains a delicate equilibrium between excitation and inhibition. Excitatory neurons (mostly using the neurotransmitter glutamate) push other neurons to fire. Inhibitory neurons (mostly using the neurotransmitter GABA) push other neurons to stay quiet. The ratio between these two forces determines everything about how your brain functions.

Think of it like driving a car. Glutamate is the gas pedal. GABA is the brake. Normal brain function requires constant, fine-grained adjustment of both. You need the gas to move, you need the brake to control the movement, and the two have to work in coordination.

In epilepsy, this balance is disrupted. The gas pedal gets stuck down, or the brake fails, or both. The result is runaway excitation. Neurons that should be firing individually start firing together, in lockstep. This hypersynchronous activity is the electrical signature of a seizure.

But what causes this imbalance? The answer depends on the type of epilepsy.

Structural/metabolic causes: A scar from a previous brain injury, a tumor, a developmental abnormality, a stroke, or an infection can create a focus of abnormally excitable tissue. Neurons at the edge of the damaged area often develop altered ion channel properties, making them more likely to fire and less responsive to inhibitory signals. This creates a "seizure focus," a small region that can trigger seizures that spread to the rest of the brain.

Genetic causes: Many epilepsy syndromes are caused by mutations in genes encoding ion channels or neurotransmitter receptors. A mutation that makes sodium channels stay open slightly longer than normal, for example, can push neurons past their firing threshold more easily. A mutation that reduces the effectiveness of GABA receptors weakens the brain's inhibitory brakes. These genetic epilepsies often affect the entire brain, which is why they tend to produce generalized (whole-brain) seizures rather than focal (localized) ones.

Unknown causes: In roughly half of all epilepsy cases, no specific cause can be identified. The structural imaging looks normal. No genetic mutation is found. The brain simply has a lower seizure threshold than average.

What Is the EEG Vocabulary of Epilepsy?

EEG doesn't just detect seizures. It has developed an entire vocabulary for describing the different types of abnormal electrical activity associated with epilepsy. Learning this vocabulary is essential for understanding what EEG actually reveals.

Interictal Epileptiform Discharges

"Interictal" means "between seizures." And this is the most diagnostically important thing EEG detects, because seizures themselves are relatively rare events. A person with poorly controlled epilepsy might have one seizure per week. That leaves 10,079 minutes of non-seizure time for every 1 minute of seizure.

But the epileptic brain doesn't just produce abnormal activity during seizures. Between seizures, the seizure focus continues to fire abnormally, producing brief, sharp electrical events called interictal epileptiform discharges (IEDs). These take two main forms:

Spikes: High-amplitude, very brief discharges lasting 20 to 70 milliseconds. On EEG, they look like sharp, pointed peaks that stand out dramatically from the background activity. They represent the synchronized firing of a few thousand to a few hundred thousand neurons in the seizure focus.

Sharp waves: Similar to spikes but slightly longer, lasting 70 to 200 milliseconds. They reflect the same underlying pathology but involve a slightly larger volume of tissue or a slightly less synchronous discharge.

Both spikes and sharp waves are typically followed by a slow wave, a broad, rounded deflection that represents the inhibitory response of surrounding tissue trying to contain the abnormal discharge. This spike-and-slow-wave pattern is the single most recognizable signature of epilepsy on EEG.

Ictal Patterns: Watching a Seizure Happen

When an actual seizure occurs during an EEG recording, the pattern depends on the type of seizure. And this is where EEG's diagnostic power really shines, because different types of epilepsy produce dramatically different seizure patterns.

Generalized Tonic-Clonic Seizure (Grand Mal): The "classic" seizure starts with a sudden burst of high-amplitude, fast activity across the entire scalp (the tonic phase, corresponding to whole-body stiffening). This transitions to rhythmic spike-and-wave discharges that gradually slow in frequency (the clonic phase, corresponding to rhythmic jerking). The seizure typically lasts 1 to 3 minutes and is followed by a period of diffuse slowing (postictal suppression) as the brain recovers.

Absence Seizure (Petit Mal): This produces one of the most distinctive patterns in all of EEG. A sudden onset of 3 Hz (three cycles per second) generalized spike-and-wave discharges that are bilateral, synchronous, and rhythmically precise. The pattern starts and stops abruptly. During the discharge, which typically lasts 5 to 30 seconds, the person stares blankly and is unresponsive. Then the discharge stops and they resume whatever they were doing, often unaware that anything happened.

Focal Seizure: The EEG shows rhythmic activity beginning in one region and evolving over time, often increasing in amplitude, changing in frequency, and potentially spreading to involve larger areas. If the seizure generalizes (spreads to the whole brain), the focal onset pattern is eventually replaced by bilateral rhythmic discharges.

How EEG Localizes the Source

For the roughly one-third of epilepsy patients whose seizures don't respond to medication, surgery to remove the seizure focus can be life-changing. But only if the surgeons can pinpoint exactly where the seizures begin. Cut out too little and the seizures continue. Cut out too much and you risk removing functional brain tissue.

This is where EEG's localizing power becomes critical.

The first step is scalp EEG with a standard 10-20 electrode placement (or a high-density array with 64 to 256 electrodes). The location of interictal spikes and the pattern of seizure onset provide an initial map. If spikes are consistently maximal over the left temporal region, and seizures show rhythmic theta activity beginning at the same electrodes, the seizure focus is likely in the left temporal lobe.

But scalp EEG has limitations. The skull and scalp act as spatial filters, smearing the electrical signal so that precise localization is difficult. A spike that appears over the left temporal region could be generated by tissue anywhere within a 2-3 centimeter radius. For surgery, that's not precise enough.

This is where intracranial EEG enters the picture. In a presurgical evaluation, neurosurgeons implant electrodes directly on the brain's surface (subdural grids and strips) or within the brain tissue itself (depth electrodes). These electrodes sit millimeters from the seizure-generating neurons, with no skull or scalp to blur the signal.

Intracranial EEG provides extraordinary resolution. It can identify the exact cortical area where seizures begin, map the direction and speed of seizure propagation, and identify eloquent cortex (brain regions responsible for critical functions like language and movement) that must be spared during surgery.

The combination of scalp and intracranial EEG has made epilepsy surgery one of the most successful neurosurgical procedures. In patients with temporal lobe epilepsy (the most common surgical candidate), roughly 60 to 70% achieve seizure freedom after surgery, a remarkable outcome for a condition that by definition has already failed medication treatment.

The Crown captures brainwave data at 256Hz across 8 channels. All processing happens on-device. Build with JavaScript or Python SDKs.

Explore the Crown

The Seizure Prediction Problem

Here's the question that haunts every person with epilepsy: when will the next seizure happen?

Seizures feel unpredictable. They strike without warning, disrupting work, social situations, sleep, and driving. The unpredictability is often described by patients as more disabling than the seizures themselves. The constant anxiety of not knowing when the next one will hit creates a psychological burden that medications don't touch.

But are seizures truly unpredictable? For over two decades, researchers have been asking whether EEG changes preceding a seizure could be detected in time to issue a warning. And the answer, increasingly, is yes.

The period before a seizure is called the preictal state. Research using long-term intracranial EEG recordings (months of continuous data from implanted electrodes) has revealed that the brain's electrical activity begins shifting minutes to hours before a seizure.

The changes are subtle and variable between patients, but several consistent patterns have emerged. Increases in high-frequency oscillations (ripples and fast ripples above 80 Hz) in the seizure onset zone often appear 30 to 60 minutes before a seizure. Changes in the phase-amplitude coupling between slow and fast oscillations can be detected up to several hours before. And shifts in the long-range coherence between brain regions, suggesting network-level instability, may precede seizures by even longer.

A landmark 2021 study published in The Lancet Neurology demonstrated that a seizure forecasting algorithm, trained on intracranial EEG data from individual patients, could predict seizures with sensitivity above 85% and a false-positive rate below 20% in a group of patients with refractory epilepsy. The forecasts were clinically useful, providing warnings ranging from minutes to hours before seizure onset.

The challenge is transitioning from intracranial to scalp-based or wearable EEG. Scalp EEG has lower signal-to-noise ratio and can't detect the high-frequency oscillations that are among the strongest preictal markers. But advances in machine learning and the development of personalized forecasting models (trained on each individual's EEG patterns) are closing the gap.

Consumer EEG and the Epilepsy Monitoring Gap

There's a glaring gap in current epilepsy management. Clinical EEG recordings happen in a hospital, typically for 20 minutes to a few days. But epilepsy is a 24/7 condition. Seizures happen at home, at work, during sleep. The vast majority of seizure activity is never recorded.

Patients are asked to keep seizure diaries, manually logging when they think they had a seizure. Studies comparing diary reports to continuous EEG monitoring have found that patients miss roughly 50% of their seizures. They miss seizures that happen during sleep. They miss brief absence seizures. They don't recognize focal seizures that manifest as subtle cognitive changes rather than dramatic convulsions.

This is where wearable, consumer-grade EEG has the potential to change the equation.

An 8-channel EEG device with 256Hz sampling rate has the technical specifications to detect many of the EEG signatures associated with seizures and interictal activity. The generalized spike-and-wave patterns of absence seizures, the rhythmic theta of temporal lobe seizures, and the sharp transients of interictal spikes all occur within the frequency range and amplitude that consumer-grade sensors can capture.

The Neurosity Crown, with its 8 electrodes at CP3, C3, F5, PO3, PO4, F6, C4, and CP4, doesn't have the spatial resolution of a clinical 21-electrode system, but it covers both hemispheres across frontal, central, and parietal regions. This is enough to detect lateralized seizure activity (which hemisphere), identify generalized patterns, and capture the spectral shifts that characterize both ictal and interictal states.

More importantly, the Crown's open SDKs (JavaScript and Python) and compatibility with BrainFlow and Lab Streaming Layer make it a platform for building the next generation of seizure detection algorithms. Researchers can access raw EEG data at 256Hz, build and test real-time detection models, and integrate with alerting systems that notify caregivers or log events automatically.

The Electrical Disorder in an Electrical Era

Epilepsy holds a unique position in neurology. It is, in the most literal sense, a disorder of the brain's electrical system. The neurons are structurally intact (in most cases). The chemistry is fine. The problem is timing. Synchronization. The electrical coordination that makes normal brain function possible runs off the rails, and millions of neurons get locked into a rhythm they can't break out of.

And EEG is the perfect tool for this disorder precisely because it speaks the same language. EEG measures voltage over time. Seizures are abnormal voltage over time. The match is so direct, so fundamental, that no other neuroimaging modality has ever seriously challenged EEG's dominance in epilepsy.

MRI can show structural lesions that cause epilepsy. PET can show metabolic abnormalities. MEG can localize seizure foci with high spatial precision. But only EEG watches the seizure happen. Only EEG captures the millisecond-by-millisecond evolution of an electrical storm from its first spark to its dying oscillation. Only EEG can be recorded continuously, for days or weeks, catching seizures that happen at 3 AM when no one is watching.

Roughly 50 million people in the world live with epilepsy. A third of them don't respond adequately to medication. Many of them have never had a complete picture of their seizure frequency because they've only been recorded for a day or two in a hospital setting.

The technology to change this exists. Eight channels. Two hundred fifty-six samples per second. Electrodes that sit on your head like a pair of headphones. Open software that lets anyone build detection algorithms and monitoring tools.

The brain's electrical language has been speaking for nearly a hundred years since Berger first listened. The question is no longer whether we can hear it. It's whether we'll give everyone access to the translator.