Feb 21

Neurosky MindWave mobile headset review – how it may help with lucid dreaming


The headset is a AAA battery powered, connected by bluetooth and has a single metallic sensor

I’ve recently bought NeuroSky Mindwave Mobile headset to take a look at the commercial-grade EEG recorder. One of the biggest selling points of this headset is ease of use and ability to get raw EEG output at ~200Hz. My experiments have uncovered that there is potential to use this device or its modification for lucid-dream related tasks.

The NeuroSky Mindwave Mobile device

NeuroSky Mindwave Mobile is a black plastic headband with a single metal pad that touches the user’s forehead. A metal clip is attached to the ear to provide ground reference. The device looks like headphones, but fits a bit too snugly around my head. I was unable to find a comfortable fit – after a few minutes of use, my head feels “constricted” for a lack of a better word.

The device is powered by a AA battery and connects to a Mac, Android or an iOS device via Bluetooth. Setup was easy on both Mac and my iPhone. Windows users would likely require an extenral bluetooth dongle and have to go through the bluetooth stack installation.

Raw output – EEG, EMG, EOG, EKG what?

The NeuroSky Mindwave Mobile headband may be configured to provide you with a raw EEG data in arbitrary units. From what I calculated, it does so at a rate of about 200 points per second, which is good enough for a high-resolution graph.

EEG stands for ElectroEncephaloGram, a readout of electrical activity from the surface

Raw EEG output. Top image is zoomed in, bottom one is zoomed out

layers of the brain. There are other ExG out there:

  • EMG –  ElectroMyoGraphy – reads electrical activity of the muscles (in particular jaw muscles)
  • EOG – electro-oculogram – records movements of the eyes
  • ECG – ElectroCardioGram – the electrical readout from the heart

As I was looking at the raw output of NeuroSky Mindwave Mobile, I noticed that there’s quite a lot of high-amplitude noise on the signal. It is the sum of EEG and the components described above. The single sensor signal picks up everything – brainwaves, muscle movements, maybe even heart rate data. As such, it’s very hard to clean up the signal and get something useful.  For example, rolling my eyes or blinking produces a sharp ~1 second peak on the raw output graph. Moving my mouth produces additional artifacts. Holding two pads in separate hands lets me clearly see the heart

Brainwave frequency bands.

Having observed how the device’s raw output responds to motion and the kind of artifacts

2 sessions of brainwave band data, I saw a lot of randomness there

there is, I proceeded to examine the various brainwave frequencies that the headband outputs. Most likely the headband takes 5 seconds of motion data and does Fast Fourier Transform and just quantifies the coefficients of the transform into brainwave bands. They are from Delta to “High Gamma”, or about 3-50 Hz. From what I’ve seen, the output of these jumps quite unpredictably from one reading to the next, even with perfect connection to the head. Doing stuff like looking around (potentially looking at the device displaying the brainwaves) produces high-amplitude output spike, which screws the results of the brainwave analysis. For example, the Delta (slow wave sleep) reading is constantly high and is made even higher by motion.

The device produces two proprietary metrics (attention, or “mostly beta” waves) and meditation(“mostly alpha” waves), which I found to mildly respond to my actions. Meditation metric consistently stays high, and the app provided by neurosky labeled me a “Zen Master” within 15 seconds of starting the app. I must be that good…Or the meditation metric does not really work.

I had more success with the Attention metric, which appears to rise more in response to thoughts. As I was scanning the app I made to display the data and thinking of logic, the attention parameter did increase significantly.

Applications for Lucid Dream induction

NeuroSky Mindwave Mobile headband is not very accurate as an EEG device, and the brainwave frequency bands shift too randomly to be really useful for detecting states like REM or slow wave sleep. But one thing that I noticed about this device brings me hope – it is extremely responsive to eye movements. Rolling one’s eyes. Blinking, looking around. All of these are clearly visible as high amplitude peaks on the raw eeg output.

Being able to see one’s eye movements on the EEG output has these potential applications:

A typical problem of lucid dreaming experiments is lack of communication between the subject and the experimenter. The subject has no ability to influence the real world from within a dream. While it is too early to promise anything, it might be possible for a computerized algorithm using a consumer-grade device like NeuroSky to look for a series of blinks or eye movements executed by a dreamer and perform actions in the real world. For example, Michael Paul Coder, another lucid dreaming researcher already has a series of experiments using Neurosky’s headbands. His lucid scribe software has a potential to execute a mouse click on a computer in response to pre-determined eye movements. This can potentially play music while you sleep to truly see how brain perceives audio from within a lucid dream.

The potential of this technology is very exciting!