The bands are frequency ranges, not moods.

An EEG recording is a mixture of rhythms at different speeds happening at once. Decomposing it by frequency is the oldest and most useful way to describe it, and the five conventional bands have been in use — with slightly shifting boundaries — for the better part of a century.

The boundaries are conventions, not natural divisions. Different laboratories and different textbooks place them a hertz or two apart, and nothing physiological happens at exactly 8 Hz that stops happening at 7.9.

The five bands

Delta, roughly 0.5–4 Hz. The slowest and, in a healthy adult, the largest. Delta dominates deep non-REM sleep, and the slow waves that define that stage are delta waves. In a waking adult, prominent delta usually means an artefact rather than a state — eye movement, a shifting electrode, sweat under a contact. In infants it is normal while awake.

Theta, roughly 4–8 Hz. Present in drowsiness and light sleep, and prominent during the transition into sleep. It also appears in the waking brain during memory tasks and navigation, particularly in recordings closer to the hippocampus than a scalp electrode can get. Theta at the scalp during the day is hard to interpret and easy to over-interpret.

Alpha, roughly 8–12 Hz. The most reproducible phenomenon in EEG. Alpha rises when the eyes close and falls when they open, strongest over the back of the head where the visual cortex sits. Hans Berger described it in 1929 and you can still demonstrate it in about a minute on most people. It is often read as "relaxed", which is approximately right and frequently misleading — alpha is better understood as a cortical idling or gating rhythm than as a feeling.

Beta, roughly 12–30 Hz. Associated with alert, engaged wakefulness, and prominent over motor areas where it drops sharply just before and during movement. It is also where muscle activity starts to contaminate a scalp recording, which matters enormously for anything worn rather than applied by a technician.

Gamma, above roughly 30 Hz. The most interesting band scientifically and the hardest to trust from the scalp. Gamma has been linked to perception and to the binding of features into a single percept, and it also overlaps almost exactly with the frequency range of muscle activity from the scalp, jaw and neck. A great deal of published gamma has turned out on re-examination to be muscle, including from tiny eye movements.

Why band power is usually reported as a share

Absolute power depends on electrode contact, skin, hair, how the amplifier was calibrated and where the reference sat. A figure in microvolts squared is not comparable between two sessions, let alone two people, unless all of that was controlled.

Reporting each band as a proportion of the total removes some of that, because a change in contact scales every band together and leaves the ratios roughly intact. It introduces its own trap: shares are constrained to sum to one, so a band can appear to fall purely because another rose. "Alpha down" and "beta up" are frequently the same event described twice.

What the bands do not mean

The frequent claim that a particular band ratio measures focus, calm, meditation or cognitive load runs well ahead of what the evidence supports. Theta/beta ratio has been studied extensively in ADHD and did not survive as a diagnostic marker. Alpha asymmetry between the hemispheres has a real literature in emotion research and effect sizes far too small for an individual reading to mean anything.

The honest summary is that band powers describe the recording. Mapping them onto a state of mind is a separate claim, and it needs its own evidence every time.

The one you can check yourself

Alpha on eye closure is worth knowing about because it is verifiable by the person being measured. Sit still, keep the eyes open for twenty seconds, close them for twenty, and alpha should rise noticeably in the closed period and fall again on opening.

Two practical cautions. It is strongest at the back of the head, so a forehead-only device may see it weakly or not at all — on our own forehead recordings alpha sits close to the noise floor, and we would rather say so than dress it up. And closing the eyes also stops blinking, which changes the trace for reasons that have nothing to do with alpha.