The sources in parentheses are links to the full references in the research review.
How it works
Electroencephalography (EEG) records, at the scalp, changes in electrical potential produced by the synchronized activity of large numbers of cortical neurons. The unit is the microvolt (µV). The signal is weak and easily swamped by other electrical signals such as eye movements, muscle activity and the heartbeat (Urigüen & Garcia-Zapirain, 2015). EEG can track changes on a millisecond scale, but the signal at the scalp mixes activity from many brain regions and cannot pinpoint where it comes from (Urigüen & Garcia-Zapirain, 2015; Cahn & Polich, 2006).
Meditation and relaxation research mostly uses spectral analysis, which splits the continuous signal into frequency bands and calculates the power in each band (Cahn & Polich, 2006; Lomas et al., 2015). Another common approach is event-related potentials, which average the responses that follow the same type of stimulus (Krigolson et al., 2017).
Frequency bands and how they are usually interpreted
Band boundaries vary slightly between studies, and the table shows common ranges. Alpha frequency differs from person to person, and some researchers recommend adjusting the bands to each person's individual alpha peak frequency (Klimesch, 1999).
| Band | Common range | Usual interpretation | Measurement caveats |
|---|---|---|---|
| Delta | About 0.5–4 Hz | Prominent in deep sleep; usually low in awake adults | In awake recordings, frontal delta is often an artifact of blinking or eye movement |
| Theta | About 4–8 Hz | Rises with drowsiness and while falling asleep; increases are common in meditation studies | A rise may simply mean sleepiness rather than a meditative state; also prone to eye movement artifacts |
| Alpha | About 8–12 Hz | Prominent during awake rest with the eyes closed; falls when the eyes open or attention turns to a task | Eyes-open and eyes-closed values differ greatly, so comparisons need the same condition |
| Beta | About 13–30 Hz | Associated with wakefulness and active thinking | Easily contaminated by facial and neck muscle activity; meditation findings are inconsistent |
| Gamma | About 30 Hz and above | Hypothesized to relate to higher-level integration of information | Scalp gamma is very easily contaminated by muscle activity such as jaw clenching or frowning |
Based on (Klimesch, 1999; Cahn & Polich, 2006; Lomas et al., 2015; Urigüen & Garcia-Zapirain, 2015).
Brain activity during meditation and relaxation
An early review reported that EEG activity tends to slow overall during meditation, and that theta and alpha activity relate to level of practice, although results vary widely across meditation styles (Cahn & Polich, 2006). A systematic review of 56 EEG studies of mindfulness found that, compared with eyes-closed rest, the most common finding during mindfulness was higher alpha and theta power. Not every study showed this, and beta, delta and gamma showed no consistent pattern (Lomas et al., 2015).
Studies of slow breathing mostly report higher alpha and lower theta, which runs in the opposite direction to the theta findings of some meditation studies (Zaccaro et al., 2018). Drowsiness also raises theta and lowers upper alpha, so a rise in theta during relaxation may simply mean the person is close to falling asleep (Klimesch, 1999).
No single accepted "relaxation EEG index" exists
Research often observes changes in certain bands, but the findings are not consistent. The EEG six-axis radar in Mandala Hub uses relative scores from 0 to 100 to show the direction of change for one person within one session. See Measurement and research.
Consumer headbands and medical-grade systems
| Aspect | Consumer headband | Medical-grade or research-grade system |
|---|---|---|
| Number and placement of electrodes | Few, fixed on the forehead or behind the ears | Usually many channels, placed at standard positions |
| Signal quality | Some studies found higher overall spectral power and greater test-retest variability | Better data quality and test-retest reliability |
| Artifacts | Forehead electrodes are more affected by blinking and frontal muscles | Also affected, but multiple channels allow more complete artifact separation |
| Research use | Can detect some event-related potential components; quick to set up | The methodological reference for spectral and event-related potential research |
The comparison studies above all had very small samples, so their conclusions should be read cautiously (Ratti et al., 2017; Krigolson et al., 2017). A reasonable position is that consumer headbands suit observing relative change in one person within one session under the same conditions. They are not suited to clinical interpretation or to direct comparison with values from other people or other devices.
Artifacts and eye movement interference
| Source | What it looks like | Main effect |
|---|---|---|
| Blinking and eye movement | Large, low-frequency waves at forehead electrodes | Delta and theta overestimated |
| Facial and neck muscles (jaw clenching, frowning, swallowing, talking) | Broadband, high-frequency noise | Beta and gamma overestimated |
| Heartbeat | Regularly recurring ECG waveforms | Periodic interference at some electrodes |
Common artifact removal methods need multiple channels. Headbands with few channels can usually only detect and discard contaminated segments (Urigüen & Garcia-Zapirain, 2015). Sitting or lying still, relaxing the jaw and brow, and not talking during measurement all reduce interference.
Factors that affect measurement
- Eyes open or closed: alpha is clearly higher with the eyes closed, so comparisons need the same condition (Klimesch, 1999).
- Drowsiness: theta rises and upper alpha falls (Klimesch, 1999).
- Age and individual alpha frequency: band boundaries and power differ from person to person (Klimesch, 1999).
- Meditation style and experience: EEG changes differ across styles and levels of experience (Cahn & Polich, 2006; Lomas et al., 2015).
- Device: spectral values and test-retest consistency differ between systems (Ratti et al., 2017).
Common misconceptions
| Common claim | A more accurate statement |
|---|---|
| High alpha means relaxed, high beta means stressed | Frequency bands do not map one to one onto mental states, and results depend on conditions, methods and individual differences (Klimesch, 1999; Lomas et al., 2015) |
| Rising theta means you have entered deep meditation | It may also mean drowsiness, or low-frequency artifacts from eye movement (Klimesch, 1999; Urigüen & Garcia-Zapirain, 2015) |
| The more gamma, the better | Scalp gamma is very easily contaminated by muscle activity, and meditation findings are inconsistent (Urigüen & Garcia-Zapirain, 2015; Lomas et al., 2015) |
| Headband EEG scores can be compared between people | Each person's alpha frequency differs, and values differ between devices (Klimesch, 1999; Ratti et al., 2017) |
| A consumer headband is equivalent to a hospital EEG | There are gaps in data quality, reliability and artifact handling (Ratti et al., 2017) |
EEG alongside HRV
The neurovisceral integration model places cardiac regulation and the brain's regulation of emotion within one framework, which is the theoretical basis for measuring HRV and EEG together (Thayer & Lane, 2000; Thayer et al., 2012). Slow breathing studies that report both higher HRV and higher alpha offer more direct evidence that the two signals change together, but there are few such studies and their methods vary widely (Zaccaro et al., 2018). Both signals are affected by breathing, posture, movement and drowsiness, so when they change together, these shared factors need to be ruled out first (Laborde et al., 2017; Klimesch, 1999; Urigüen & Garcia-Zapirain, 2015).