> ## Documentation Index
> Fetch the complete documentation index at: https://docs.elata.bio/llms.txt
> Use this file to discover all available pages before exploring further.

# EEG Headbands

> What consumer EEG headbands measure, how the Elata SDK processes it, and what calm and focus scores can and can't claim.

## The principle

Neurons communicate with tiny electrical currents. When large groups fire
together, the combined activity can be picked up by sensors on the scalp. That
recording is an **electroencephalogram (EEG)**. It's dominated by rhythms at
different frequencies:

| Band | Approx. range | Often associated with |
| - | - | - |
| Delta | 1–4 Hz | Deep sleep |
| Theta | 4–8 Hz | Drowsiness, some memory tasks |
| Alpha | 8–12 Hz | Relaxed wakefulness, especially with eyes closed |
| Beta | 13–30 Hz | Active thinking, alertness |
| Gamma | 30+ Hz | Fast processing; also heavily contaminated by muscle activity |

These associations are tendencies across groups of people, not rules for any
single moment.

***

## Consumer headbands

The Elata SDK supports the Muse family over Web Bluetooth:

| Device | EEG channels | Sensor positions |
| - | - | - |
| Muse 2, Muse S | 4 | Behind the ears (TP9, TP10) and on the forehead (AF7, AF8) |
| Muse S (Athena) | 8 | Additional channels on the newer protocol |

Research-grade systems use 32 to 256 wet electrodes in a cap. A headband uses a
handful of dry sensors in fixed positions. That makes it comfortable and
affordable, and it limits what it can measure.

***

## How the Elata SDK processes EEG

`eeg-web` runs in the browser using WebAssembly:

* **Filtering and spectra:** removes drift and line noise, and computes the frequency spectrum of each window
* **Band powers:** absolute, relative, and log power per band and channel
* **Quality flags:** flatline, clipping, extreme amplitude, and line noise, so bad windows can be excluded
* **Simple models:** for example, an alpha-rhythm detector and a calmness estimate built on band powers

***

## What affects the signal

| Factor | Effect |
| - | - |
| **Blinks and eye movement** | Large spikes on the forehead sensors |
| **Jaw clenching, talking, frowning** | Muscle activity that overwhelms higher frequencies |
| **Poor contact** | Hair, dry skin, or a loose fit cause noise or flat channels |
| **Movement** | Cable and sensor motion add slow drifts and bursts |
| **Few, fixed sensors** | Activity can't be located precisely, and some brain areas aren't covered |

Studies comparing consumer headbands with research systems find that they can
capture real brain signals. Event-related potentials have been recovered with
Muse, and four Muse electrodes performed close to a 32-electrode dataset on
one emotion-classification task. But signal quality varies by device and
setup: one 2024 comparison found Muse S had the weakest alignment with a
research-grade amplifier of the devices tested.

***

## What calm and focus scores can claim

A "calm" or "focus" score is a **product interpretation** built from band
powers and other inputs. It is not a direct readout of a mental state.

* Scores are most meaningful **relative to the same person's baseline**, not compared between people.
* They should come with **quality information** and be withheld when the signal is poor.
* Some popular measures don't hold up. For example, the SDK's Focus score deliberately does **not** use the theta/beta ratio, because that ratio isn't a valid measure of attention.

See [Metrics and scores](/overview/science/metrics-and-scores).

***

## Sources

Peer-reviewed literature, retrieved via PubMed:

* Krigolson OE, et al. Choosing MUSE: validation of a low-cost, portable EEG system for ERP research. *Frontiers in Neuroscience*, 2017. [doi:10.3389/fnins.2017.00109](https://doi.org/10.3389/fnins.2017.00109)
* Mikhaylov D, et al. Comparison of EEG signal spectral characteristics obtained with consumer- and research-grade devices. *Sensors*, 2024. [doi:10.3390/s24248108](https://doi.org/10.3390/s24248108)
* Garcia-Moreno FM, et al. EEG headbands vs caps: how many electrodes do I need to detect emotions? The case of the MUSE headband. *Computers in Biology and Medicine*, 2024. [doi:10.1016/j.compbiomed.2024.109463](https://doi.org/10.1016/j.compbiomed.2024.109463)

SDK implementation: [`eeg-web`](https://github.com/Elata-Biosciences/elata-bio-sdk/tree/main/packages/eeg-web) and [`biosignal-analytics`](https://github.com/Elata-Biosciences/elata-bio-sdk/tree/main/packages/biosignal-analytics) (public SDK repository).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.