MindwaveNotes - mens-amplio/config GitHub Wiki

Current Status (Wed 2013-06-05) @Everett

Have the pi reading values from the headset just fine, and I recorded 20-minute readings while doing a bunch of different things (working, sleeping/napping, watching tv, etc) so we'd have a pile of data to work with. Unfortunately, the measurements appear to be all noise and no signal. They fluctuate wildly, and even when I manipulate them in R -- look at the distribution, smooth them over time, etc -- there is no meaningful correlation that I can deduce between any given activity and attention/meditation/any kind of brainwave. The repo has all the data in the "measurements" folder and some starter analysis code in "analysis/basicplots.R" (code courtesy of my friend Albert Kim (rudeboybert on github) who's a stats guy). Kudos to anyone who wants to take a shot of finding anything but spurious causality there.

Starter notes/links from Sarah

Initial tinkering with the device (Everett)

  • Paired with my android phone -- you have to slide and hold the headset switch to "pair" (past "on") for 3 seconds to enter pairing mode, light will blink in double-pulses of blue. Shows up as "MindWave Mobile" in bluetooth devices screen.
  • Downloaded "Brainwave visualizer" app (from NeuroSky, who makes the device). I tried to use their other app, "MindWave Mobile Tutorial" which supposedly walks you through the features, but the app seemed to think the headset was disconnected (false, since the other app worked and it was paired). So I gave up on that one.
  • Successfully ran brainwave visualizer. It outputs 8 bars showing the presence of 8 different types of brain waves, visualized in two ways: 8 vertical bars in the top right, and 8 points inside an octagon connected by a curve on the left. It also outputs 2 dials showing your level of "attention" and "meditation" on a 0-100% scale.
  • With the headset off my head, just sitting on the table, it shows no signal (thankfully). Wearing the headset, though, measurements vary pretty wildly. The levels of various types of brainwaves all seemed to fluctuate constantly despite the fact that I was doing the same thing the whole time. Same goes for the attention/meditation dials. I tried both concentrating and meditating, but the dials still swung up and down seemingly randomly.
  • Screenshot of visualizer: https://plus.google.com/109553468127314490845/posts/MUn1bQzfCpb
  • Video of the visualizer random movements (with me doing nothing but watching) for one minute here: https://www.youtube.com/watch?v=O3RJlZqdRxc
  • Fun tangent: apparently some Japanese engineers hooked up motorized cat ears to a MindWave that turn up or down in response to your mood. In theory. Pics: https://www.google.com/search?safe=off&site=&tbm=isch&source=hp&biw=1280&bih=600&q=neruowear&oq=neruowear&gs_l=img.3..0i10i24.394.1381.0.1542.9.8.0.0.0.0.96.410.7.7.0...0.0...1ac.1.15.img.MPkH1Ax6bPc

Rookie neuroscience wikipedia trip (slash conversation with Sarah)

  • EEGs measure electric potential in the brain. These are caused by neurons firing action potentials, and when enough of them fire in close physical proximity, it can produce a measurable potential difference between different points on the brain. When measure the potential over time, and break this signal out into its frequency spectrum, the relative activity of different frequencies can sometimes hint as to what the brain is doing. For example, low frequency "delta" waves are associated with deep sleep. Our MindWave device is surely using the relative activity of different types of waves to determine if you are meditating, concentrating, etc. See this chart for some ideas of the different kinds of waves and what they might mean: http://en.wikipedia.org/wiki/Electroencephalography#Comparison_table_of_EEG_rhythmic_activity_frequency_bands
  • Devices to measure this electric signal range in sophistication and are usually conducted in a lab/controlled environment with dozens of connections to different parts of the scalp with people shaving their heads and using conductive gooey gel and stuff. That's not really realistic for burning man. Our MindWave headset, by contrast, has only one connection point (the forehead), is cheap, and easy to slip on with no goo or haircuts or anything.
  • The downside of our simpler EEG method is that we get may not be pure, sweet, Columbian-grade brain signal. There will be noise from a lot of sources -- other bodily functions (eyes, heart, etc) as well as the environment around us add electric potentials that convolute the signal. It's hard to know what comes from the brain and what doesn't. There are varying levels of advanced algorithms to try to sift signal from noise, some freely available, but I haven't gone down that rabbit hole yet. http://en.wikipedia.org/wiki/Electroencephalography#Artifacts
  • The MindWave device is supposed to take care of this for us. In theory it cleans up the signal, filters out spurious crap, and gives us quality brainwave measurements. However, we have no way to know how "quality" it is. My next task is to get the Pi reading something from this headset over bluetooth, and explore the range of information it can gives us programatically.

MindWave working with the Pi