Had a researcher tell me my BCI calibration method was holding me back

I was using the same calibration routine for my OpenBCI headset for about 8 months. A neuroscientist at a meetup pointed out that I was only training on motor imagery data from one session. He said that was like learning a song by playing the same three notes over and over. I switched to running calibration across 5 different recording days with varied tasks like blinking and jaw clenching. My signal classification accuracy went from 72% to 89% after about 3 weeks. Has anyone else found that their setup works better with more diverse training data?

3 comments

3 Comments

alex52415d ago

Huh, that's funny you mention that. I had a buddy who builds little DIY EEG toys for fun, and he was stuck at like 60% accuracy for months. He'd just sit there in his chair at the same time each day, training on the same three motor imagery patterns. Then his girlfriend, who's a musician, told him he was basically just practicing the same riff over and over. So he started mixing it up - sometimes he'd train while watching action movies, sometimes while eating a bag of chips, and even once after a really stressful work call. After two weeks of that chaos, his accuracy jumped to 85% and he said the system actually felt more responsive in real time. It's wild how much your brain hates routine.

emeryfox13d agoTop Commenter

Jumping in here because I've messed with consumer EEG headsets myself and hit the exact same wall. What worked for me was treating training like brushing your teeth in a different part of the bathroom each day. One day in the morning with coffee, next day at night after a run, once while my neighbor was mowing the lawn. That forced my brain to actually lock onto the signal instead of memorizing the chair cushion. My favorite trick was doing a quick timed session before and after a cold shower throw everything off balance. Your buddy's jump to 85% sounds totally realistic to me based on my own trial and error.

spencer_moore3915d ago

Feel that story hard. My brain is exactly the same way with anything repetitive. I tried learning guitar a few years back and I would practice the same three chords at the exact same time every night. Got super stuck and almost quit. Then I started playing while my roommates were watching loud TV or after I got home from a long shift. It was like my brain finally woke up and actually had to pay attention instead of just going through the motions. It makes total sense that your buddy's EEG jumped so much when he forced his brain to adapt to different situations. Routine really does put your brain to sleep.