Epileptic seizures are caused by abnormal, excessive electrical discharges from cortical neurons and are diagnosed by their clinical pattern plus EEG or imaging evidence. Non-epileptic seizures are events that resemble seizures but arise from other causes—most often psychogenic (functional) attacks, but also syncope, movement disorders, or metabolic disturbances—and they do not respond to antiseizure medication. Telling them apart matters because the treatments diverge sharply and the wrong label causes harm.

The distinction is not cosmetic. A psychogenic event treated with escalating antiseizure drugs exposes a person to side effects without benefit. An epileptic event mistaken for anxiety can delay control of potentially dangerous seizures. And the two can coexist in the same person, which is where careful diagnosis earns its keep.

The electrophysiology that separates them

An epileptic seizure begins with a disturbance in neuronal excitability. Groups of cortical neurons fire synchronously when excitatory drive (glutamate) overwhelms inhibitory restraint (GABA), or when ion channel behavior shifts toward depolarization. That synchronous discharge spreads and produces the electrical signature clinicians look for on an electroencephalogram.

The mechanism explains the clinical pattern. When abnormal discharge stays in one region, the event is focal, with symptoms tied to that region's function—a temporal-lobe discharge may cause an odd rising sensation or automatisms. When discharge generalizes across both hemispheres, the result is a tonic-clonic convulsion with loss of consciousness. The electricity is primary; the movement is downstream.

Psychogenic non-epileptic seizures (PNES) work in the opposite direction. There is no abnormal cortical electrical discharge driving the event. The brain is generating movement through ordinary motor pathways, but the control is disrupted—typically linked to a stress response or a dissociative process. Consciousness may be altered, but the EEG does not show the evolving epileptiform pattern that defines an epileptic event.

This is why the two can look nearly identical from across a room and completely different on a recording. The surface behavior is a poor instrument; the electrical activity is the discriminating signal. That single fact drives almost every diagnostic decision that follows.

How often the two are confused, and why the stakes are high

Misclassification is common enough that it shapes how specialists approach any first seizure. A substantial share of people referred to epilepsy centers for medication-resistant seizures turn out to have PNES, and the reverse error—dismissing a true seizure as psychological—also occurs.

The harm runs both ways. People with PNES may spend years on antiseizure drugs that cannot help them, accumulating side effects and, in some cases, emergency intubation for prolonged events. People with unrecognized epilepsy face injury risk, driving restrictions, and untreated seizures. Getting the label right is the highest-value step in the entire pathway.

Age and context shift the probabilities. PNES most often emerges in adolescence and young adulthood and is more common in people with a history of trauma, though neither is required and neither confirms the diagnosis. Epilepsy has its own age distribution, with high incidence at the extremes of life. A clinician weighing a first event uses these base rates, then tests.

Where the picture gets complicated

The coexistence problem is the professional caveat that changes management. A meaningful fraction of people with PNES also have documented epilepsy, so finding one diagnosis does not rule out the other. Treating the psychogenic events as if they exclude epilepsy, or vice versa, is a recognized source of error.

Syncope is the great imitator on the other side. A faint from reduced cerebral blood flow can produce brief jerking (convulsive syncope) that looks like a seizure to bystanders. The mechanism is entirely different—a transient drop in perfusion, not a cortical discharge—and the workup is cardiac, not neurologic. Post-event confusion and duration help distinguish them.

Metabolic and toxic causes add another layer. Hypoglycemia, severe electrolyte disturbance, and drug or alcohol withdrawal can provoke genuine seizures that resolve when the underlying problem is corrected. These are epileptic in mechanism but not necessarily epilepsy as a chronic condition. That distinction alters whether long-term medication is appropriate.

Finally, movement disorders and sleep events can mimic seizures. Paroxysmal dyskinesias, tics, and certain parasomnias produce episodic movements without epileptiform activity. Each has its own diagnostic path.

Comparison table: features that help distinguish the two

Clinical features overlap heavily, so no single item settles the question. The table shows tendencies, not rules, and the EEG-video correlation remains the reference standard.

Feature Epileptic seizure Psychogenic non-epileptic seizure Syncope
Typical duration Seconds to a few minutes Often longer, sometimes many minutes Seconds to about a minute
Eye behavior May deviate, open or closed Often forcibly closed, resists opening Brief, then normal
Motor pattern Rhythmic, evolving, stereotyped Variable, asynchronous, out of phase Brief jerking possible if convulsive
Post-event state Confusion, fatigue, gradual recovery Variable, sometimes rapid return Rapid recovery, pale, clammy
EEG during event Evolving epileptiform discharge No epileptiform discharge Slowing from hypoperfusion
Response to antiseizure drugs Reduces seizures No effect on event frequency Not applicable
Typical injury pattern Tongue biting (lateral), bruising Injuries possible but less stereotyped Injuries from fall

The lateral tongue bite and the prolonged, variable motor pattern are among the more useful bedside clues, but none is decisive. A clinician who relies on appearance alone will be wrong often enough to matter.

Timeline: what happens from event to diagnosis

Diagnosis is a process, not a single test. The time course below reflects the typical sequence in a specialty setting; the limits of each step are noted.

Stage What happens Typical timing Key limitation
Event and bystander account Witness describes movements, duration, eyes, recovery At the event Recall is unreliable under stress
Initial clinical assessment History, medication review, metabolic and glucose testing Same day to days Overlapping features mislead
Routine EEG Brief recording between events Days to weeks A normal interictal EEG does not exclude epilepsy
Prolonged video-EEG monitoring Continuous recording with video, often with medication reduction Days of admission Requires provoking events to capture
Cardiac workup ECG, monitoring for syncope suspicion Days to weeks Arrhythmias may be intermittent
Integrated diagnosis and plan Synthesis of all data; discussion of results with the patient After monitoring Coexisting diagnoses complicate conclusions

The critical point is in the fourth row. A routine EEG captures a snapshot, and many people with epilepsy have normal interictal recordings. Video-EEG monitoring with capture of a typical event is the reference standard because it links behavior to electrical activity in real time.

Common misconceptions that derail diagnosis

The most persistent myth is that a normal EEG rules out epilepsy. It does not. Between-event recordings are frequently normal even in confirmed epilepsy, so a normal result changes probability rather than closing the question. Relying on it alone produces both false reassurance and false dismissal.

A second misconception holds that psychogenic seizures are "fake" or voluntary. They are neither. PNES are involuntary events with a real neurological basis in how the brain generates and controls movement, and framing them as fabrication damages the therapeutic relationship and reduces engagement with treatment.

A third error treats the two diagnoses as mutually exclusive. As noted, coexistence is well documented, and a single diagnostic label can mask the other condition. Clinicians who stop at the first explanation miss the second.

A fourth misconception concerns treatment response. The absence of benefit from antiseizure medication is sometimes taken as proof of PNES. That inference is unreliable, because roughly a third of people with epilepsy have medication-resistant seizures. Non-response shifts the odds but does not settle the cause.

The plain-language version: appearance is a hint, electricity is the evidence, and no single clue is a verdict.

Practical actions and thresholds

For anyone who has witnessed a first event, the sequence below reflects what specialist guidance generally supports. These are general steps, not a diagnosis or treatment plan for any individual.

Call emergency services for any seizure lasting longer than five minutes, for repeated seizures without recovery between them, for a first-ever seizure, or for breathing difficulty, injury, or pregnancy. Status epilepticus is a medical emergency. Protect the person from injury during the event: cushion the head, clear hard objects, and do not restrain or place anything in the mouth. Time the event. Duration above five minutes is the threshold that changes the response from observation to emergency care. Record what you observe: eye position, movement pattern, duration, and how quickly the person returned to baseline. This account is diagnostic gold. After recovery, arrange prompt neurologic evaluation rather than waiting for a second event, especially if the first was unwitnessed or atypical. Report any history of trauma, prior similar episodes, and all medications, since these shift the differential.

The referral threshold matters. A first unprovoked event warrants neurologic assessment, and events that continue despite adequate antiseizure medication should prompt referral to an epilepsy center for video-EEG evaluation. That step is where the epileptic versus non-epileptic question is most reliably answered.

Limits of the evidence and what remains uncertain

The reference standard itself has constraints. Video-EEG monitoring is only conclusive when a typical event is captured, and some admissions end without one. When the recording is inconclusive, diagnosis rests on the clinical synthesis, and residual uncertainty is real.

PNES is also a heterogeneous category, and the mechanisms are not fully mapped. The label describes a pattern of events, not a single pathophysiology, which is why treatment approaches vary and outcomes differ across patients. Research on the underlying processes is ongoing.

Coexistence remains incompletely quantified, and estimates vary by setting and referral population. What can be said with confidence is that the overlap is clinically significant and that clinicians should not treat one diagnosis as excluding the other.

Finally, no bedside feature or short test replaces integrated assessment. The tools—history, witnessed account, video-EEG, cardiac evaluation—each carry error rates, and the diagnosis is only as strong as the combination. Patients and families should expect a process that takes time, and should be cautious about any single-test answer to a question this consequential.