Thursday, July 24, 2014

EEG of me blogging vs. feeding/cleaning


I recently acquired a MindFlex EEG-based toy (on ebay, for a total of $21 with shipping), which is based on a NeuroSky ThinkGear ASIC chip.  As a toy, it's not that great, but if you solder wires to the transmit and ground pins, and hook it up to a TTL-level serial port, you can read the data off it.  By default the data comes processed into a bunch of frequency domains (presumably by running an FFT on the raw potentials), though if you attach your serial port to the receive pin (I ended up shorting that pin to another and had to cut through the blog carefully afterwards; I'm not good at soldering), you can switch to raw data mode, though my bridge hardware isn't fast enough for that.  As a safety measure, it's a good idea to either have the computer be a laptop not plugged into mains or else to bridge from the ASIC to the laptop wirelessly.

I hooked up the mindflex to the BrainLink all-purpose-BlueTooth-interconnect device (coupon code SS72142 gives a 30% discount on everything at SurplusShed this week).  I had some initial technical difficulties.  For some reason, the BrainLink keeps on resetting when it gets a lot of incoming serial data, and so I had to write some custom communication code for it that un-reset when receiving data, instead of just using the standard BrainLink java library.  And I wrote some quick visualization code in java (it's a mess, as I've never written desktop java GUI code before).  If you want to play with the messy code, it's here (it's called brainflex).  (You probably don't have a BrainLink, but any serial-to-BlueTooth adapter should work, as long as you create a new implementation of DataLink, which should be quite easy.)

Anyway, while doing the preceding post, I had the BrainLink on, and recorded the processed EEG.  (The Attention and Meditation data is computed by the chip from the Fourier transform data.)  Here it is:

As control data, I then switched to feeding the toddler and cleaning her and after her:

There is discussion online on whether the BrainFlex toy actually works, or if it's just an illusion-of-control thing, though the NeuroSky folk have research data on their chip that suggests it does do something.

Note that all the regular frequency domains (not the Attention and Meditation, though) are normalized to sum to a constant total.

Beta and alpha seem much more active when blogging than when feeding/cleaning.  On the other hand, the chip's computed Attention value seems rather higher for the feeding/cleaning, which fits with how I felt: the blogging seemed fairly automatic, while the feeding/cleaning involved more conscious attention.

A confound for the experiment is that as I was blogging, I was on the computer, so there would have been electromagnetic interference from that.  I did make sure that the computer was not plugged into the mains, which may have reduced interference.  And another confound was that I knew I was recording myself.  This isn't real science!

No comments: