These are my Brainwaves on Music


I didn’t see the pattern at first, but this struck me as very cool. I’ve been playing around with capturing my brain waves while doing different things lately, and a recent recording jumped out at me. You’re looking at my brainwaves captured using the emotiv, a consumer-grade EEG, while I played the guitar:

It’s a lot to take in. At first, I didn’t see the pattern. Then I started thinking back on the play session. I had played three songs. The first I had played four times, the second three times, and the last two times. Look again at the purple lines in the graph, thinking on that pattern:

If you cross your eyes a little, it’s not hard to see that 4x, 3x, 2x wave pattern in the purple. The pattern also seems to be present in the orange (interest) waves. I have yet to dig into the raw data, but it seems very probable that it would be possible to train a machine learning model on “fingerprints” of songs. If this theory holds, it would be possible to accurately know when a person had performed a given task.

This is a mind-bending proposition. I’m sure others have discovered it. It would not, in theory, be limited to music or songs. A well-trained task-fingerprint model could detect whenever you did arbitrary things throughout the day. It’s hard to say how such models might generalize to other individuals, but it’s entirely conceivable that a library of highly-trained fingerprints could provide the backbone for a tool of immense power.

Imagine: being able to know accurately how similarly two people had done a task. It could revolutionize education if a teacher could know how well he had conveyed the material to his subject. In education, we know that the most important variables are the timeliness and quality of feedback. Software which provided “neurofeedback,” like audio or images, could be used to guide students towards “perfect knowledge” of the topic or task in minimal time.

Other possibilities intrigue me as well. Could excitement (yellow) act as an indicator of interest levels throughout a given education/play-session? Might engagement be able to determine when a student should take a break?

There is a whole community of scientists and researchers out there, I know, and so far I have only stuck my toe in the water. I don’t know if or when I’ll have the opportunity to explore this data more completely. Still, I can’t help but wonder about the possibilities of a little computer science applied to this problem…

About the author

Add Comment

By zane

Recent Posts

Recent Comments




Recent Posts




Get in touch

Quickly communicate covalent niche markets for maintainable sources. Collaboratively harness resource sucking experiences whereas cost effective meta-services.