> For the complete documentation index, see [llms.txt](https://docs.anthriq.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.anthriq.com/xbud/experiments/eyes-closed-alpha.md).

# Eyes-closed alpha

Your first EEG experiment. Two 30-second recordings, eyes open and then eyes closed. The difference between them is the alpha rhythm: a peak near 10 Hz that appears over the back of the head when the visual cortex has nothing to look at.

The effect is reliable and visible within seconds, which makes it the best first recording to attempt. Allow about 10 minutes, and work with two people if you can: one subject, one operator.

> **Note:** Complete [Get started with xBud](/xbud/technical-guide.md) first. This guide assumes xBud is powered, wired to the DAQ, and connected in BXI Studio.

## Check the setup

Confirm the kit is running before the subject sits down.

* xBud is powered and the DAQ is connected in BXI Studio.
* The analog inputs are configured, with the sampling rate set to 250 Hz or higher.
* Channels are labelled for the sites you are about to use: `O1`, `O2`, `Fpz`, `REF`.
* Tapping an electrode produces a visible jump on the live trace.

Have ready: the EEG cap or headband, the electrode set, a quiet room, and a chair facing away from the screen. Put the laptop on battery and unplug its charger.

> **Tip:** Decide the file names before you start, for example `sub01_eyes_open` and `sub01_eyes_closed`. Naming files at the end of a session is easy to get wrong.

## Prepare the subject

### Fit the band

Fit the cap or headband level, above the ears and eyebrows, and seated over the back of the head. Snug, not tight. Part the hair out from under each contact so the combs reach the scalp.

### Find the occipital sites

Alpha is strongest over the visual cortex, so the two recording sites sit at the back of the head.

1. Find the **inion**, the bony ridge where the skull meets the neck.
2. Move about 4 cm up the midline from it. That is **Oz**.
3. **O1** and **O2** sit about 5 cm to the left and right of Oz, at the same height.

Place `EEG 0` on O1 and `EEG 1` on O2.

### Place the control and the reference

Place `EEG 2` at **Fpz** on the forehead. This is the control site: alpha is weak there, which is what makes the occipital result convincing.

Place `REF` at **Cz**, on top of the head midway between the ears. Place `AGND` and `DRL` on the two temples, T7 and T8, or on the two mastoids.

![10-20 head map with O1, O2 and Fpz marked as recording sites, Cz as reference, and T7 and T8 as ground](https://2761951758-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FhtBZnSGOC4EMhU3nc5oM%2Fuploads%2Fgit-blob-673056498214f470445932786ec9c849387408cd%2Fmontage-10-20.svg?alt=media)

> **Tip:** Gel the AGND and DRL contacts. Neither carries a recorded signal, so gelling them improves the common-mode path and reduces mains hum.

### Check the contacts

Open the **Real Time Plot** and confirm every channel shows a trace. Ask the subject to blink: Fpz jumps. Tap O1: that channel jumps. A flat trace or a thick 50 Hz saw-tooth means poor contact, so re-seat that electrode before continuing.

## Run the experiment

Use a bandpass filter of roughly 1–40 Hz and a 50 Hz bandstop for the live view, and open the **Fast Fourier Transform** alongside the **Real Time Plot**.

### Record 30 seconds with eyes open

Select **Start Plotting**, then **Start Recording**. The subject sits still, relaxed, eyes open on a fixation point. After 30 seconds, stop and save as `sub01_eyes_open`.

The Fast Fourier Transform shows a smooth falling curve: power drops as frequency rises, with no bump.

### Record 30 seconds with eyes closed

Start a new recording. The subject closes their eyes and stays still. After about 30 seconds, ask them to open their eyes, then stop and save as `sub01_eyes_closed`.

Ask the subject to stay relaxed rather than concentrating: alpha is strongest when the visual cortex is idle, and mental effort suppresses it.

## Read the result

Open the eyes-closed recording and look at the **Fast Fourier Transform** for O1 and O2. A **peak near 10 Hz** stands above the falling curve. That is alpha.

![Occipital spectra from O1 and O2 between 1 and 45 Hz. The eyes-closed trace carries a large peak near 9 Hz that is absent from the eyes-open trace.](https://2761951758-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FhtBZnSGOC4EMhU3nc5oM%2Fuploads%2Fgit-blob-6edf67ea70a82e1bb73047699ebdc4c5d8570cd1%2Falpha-eyes-closed-vs-open.svg?alt=media)

The figure above is a recording made with this montage on xBud. Both conditions are plotted on the same axes over 1–45 Hz, which is the band the front-end passes cleanly and keeps mains interference out of the view.

> **Note:** The y-axis is logarithmic. A peak that looks modest on this scale is a large change in power, and a ratio between two conditions becomes a subtraction in dB. Set the same **Window (s)** for both recordings, or the two spectra are not comparable.

### What to expect

The numbers below come from the recording plotted above, which used O1 and O2 only. Yours will differ, since individual peak frequency and alpha strength vary considerably between people, but the pattern should hold.

| Measure                     | O1      | O2      |
| --------------------------- | ------- | ------- |
| Peak frequency, eyes closed | 9.25 Hz | 9.25 Hz |
| 8–12 Hz power, eyes closed  | 130 µV² | 82 µV²  |
| 8–12 Hz power, eyes open    | 14 µV²  | 13 µV²  |
| Ratio, closed to open       | 9.2×    | 6.2×    |

A ratio above about 2× is a clear result. If yours sits near 1×, the contacts are the first thing to check, followed by whether the subject was actually relaxed.

Three comparisons make the point:

| Compare                                     | Expected                                  |
| ------------------------------------------- | ----------------------------------------- |
| Eyes closed against eyes open, on O1 and O2 | The peak is present only with eyes closed |
| O1 and O2 against Fpz, eyes closed          | The peak is small or absent frontally     |
| Across the eyes-open file                   | No peak on any channel                    |

Alpha is occipital and appears with the eyes closed. Those two recordings demonstrate both.

## Take it further

**Alpha reactivity.** Measure band power between 8 and 12 Hz on O1 and O2 in each file, then divide the eyes-closed value by the eyes-open value. Ratios of 2× or more are common; the recording above gives 9.2× on O1. A normalised form, `(closed − open) / (closed + open)`, keeps the value between −1 and 1 and is easier to pool across a class: the same recording gives 0.80 on O1 and 0.72 on O2.

**Individual peak frequency.** Read the frequency at the top of the eyes-closed peak. It falls between 8 and 12 Hz, differs between people, and is stable for one person across sessions.

> **Note:** The 8–12 Hz window is the convention used here. Some texts define the alpha band as 8–13 Hz, which shifts the computed power slightly. State the band you used when you report a result.

Export both recordings as CSV to carry this into Python or a spreadsheet. See [Exporting a dataset](/bxi-studio/streams/exporting-a-dataset.md).

## Next steps

* [Experiment Builder](/bxi-studio/experiments/experiment-builder.md): turn this into a repeatable paradigm with timed epochs and markers
* [Troubleshoot xBud](/xbud/maintenance-and-troubleshooting.md): if the trace did not look right