An EEG trace is a mixture of rhythms at different frequencies. Splitting it into bands is the oldest and still the most useful way to summarise it. Alpha was the first rhythm Berger described, and bands remain the basis of most EEG features today.
Band edges are a convention, not a law of physiology, and different labs draw them a hertz or two apart. Individual peak frequencies also differ: one person's alpha may peak at 9 Hz and another's at 11. Whatever edges you use, record them.
The bands at a glance
These are the band edges BCILattice uses for its EEG preset. Band presets are shared across steps, so "alpha" means the same range in a topography, a time-frequency view and an extracted feature.
| Band | Range | Typically associated with |
|---|---|---|
| Delta | 0.5–3.5 Hz | Deep sleep; slow activity in infants; pathological in awake adults when focal. |
| Theta | 4–7 Hz | Drowsiness, and frontal-midline theta during working memory and cognitive control. |
| Alpha | 8–12 Hz | Relaxed wakefulness with eyes closed, strongest over occipital sites; drops when the eyes open or attention engages. |
| Mu | 8–12 Hz, over motor cortex | Same frequency as alpha, different source. Suppressed by real or imagined movement, which is the basis of motor imagery BCIs. |
| Beta | 13–30 Hz | Active thinking and motor control; beta suppresses during movement and rebounds after it. |
| Gamma | 30–80 Hz | Local cortical processing and perceptual binding. Hard to measure at the scalp because muscle activity overlaps it. |
What band power can and cannot tell you
A change in alpha power is a measurement, not a mood. "More alpha means relaxed" is true on average in some conditions and misleading as a rule: alpha rises over regions that are being actively ignored, and its level depends on the person, the electrode, the reference and whether the eyes are open.
Ratios such as theta/beta are popular in consumer products, and they are as fragile as the two powers that make them up. Treat a band feature as a hypothesis to test against a contrast, not as a readout of a mental state.
How band power is measured
BCILattice extracts all of these as named features, plus band peak frequency, spectral entropy and edge frequency, and the band ratios theta/beta, alpha/beta and the engagement index. See EEG feature extraction.
Absolute band power
Power in the band from a spectral estimate such as Welch's method, in µV²/Hz or µV².
Relative band power
Band power divided by total power, which reduces between-subject differences in overall amplitude.
Log band power
The log of band power, closer to normally distributed and the usual input to a linear classifier.
Over time
Time-frequency methods track band power through a trial. ERD/ERS is the drop or rise relative to a baseline.
Bands are signal-specific
Applying EEG bands to another signal is a common error. fNIRS needs haemodynamic bands far below 1 Hz, and EMG needs bands in the tens to hundreds of hertz. BCILattice ships separate band presets for each and reads the sampling rate from the recording, so a 10 Hz fNIRS file is not processed as if it were EEG.
Questions
What are the five types of brain waves?
Delta (about 0.5–4 Hz), theta (4–7 Hz), alpha (8–12 Hz), beta (13–30 Hz) and gamma (above 30 Hz). Mu is a sixth rhythm in the alpha range, generated over the motor cortex.
What is the difference between alpha and mu rhythms?
Both sit around 8–12 Hz, but alpha is strongest over the visual cortex and changes with eyes open or closed, while mu is generated over the sensorimotor cortex and is suppressed by real or imagined movement.
Are higher alpha waves good?
Not in a simple sense. Alpha power depends on the person, the region, eye state and the task. It rises over brain areas that are being disengaged, so a change is only meaningful against a controlled comparison.
Why is gamma hard to measure with EEG?
Gamma is small at the scalp, and muscle activity from the face, jaw and neck produces broadband high-frequency signal in the same range, so scalp gamma is easily contaminated.
Which band is used for motor imagery BCIs?
The mu (8–12 Hz) and beta (13–30 Hz) bands over the motor cortex. Imagining a movement suppresses their power over the opposite hemisphere.
Read the reference
Related
How EEG Works: What an EEG Measures
Electroencephalography (EEG) records the brain's electrical activity with electrodes on the scalp. It has millisecond timing and centimetre-scale blur, and knowing both is the key to using it well.
EEG Time-Frequency Analysis: ERSP and ITC
A spectrum tells you which frequencies are present. A time-frequency decomposition tells you when, and how power and phase change relative to an event.
EEG Feature Extraction: Which Features to Use
A feature is a reduction of a trial to a number. The right ones depend on where the signal lives (time, frequency, space or the links between channels), and the wrong ones multiply until the model overfits.
Motor Imagery BCI: EEG Classification Explained
Imagining a movement suppresses the mu and beta rhythms over the opposite motor cortex. A motor imagery BCI detects that suppression and turns it into a command, with no movement and no external stimulus.