Just got a simple tip from a retired researcher that changed my BCI testing

I was talking to Dr. Miller at a conference in Chicago last month, and he asked why I was using a 128-channel setup for basic motor imagery. He said he got better results with just 8 channels and a good spatial filter. I tried it on my last batch of test subjects and the signal clarity was much better. Has anyone else found that fewer channels actually work better for certain tasks?

2 comments

2 Comments

alice2421mo ago

Wait, didn't a bunch of brain-computer interface folks at that Stanford workshop last spring talk about this exact thing? Someone there showed how just 4 channels over the motor cortex gave them almost the same accuracy as a full cap for basic left/right hand imagery. They said the noise cancellation was better because there was less cross-talk between electrodes. I started stripping down my setup too after hearing that, and yeah, the signal processing is way cleaner with fewer channels for simple tasks. It makes me wonder if we've all been overcomplicating things for years because the hardware companies keep pushing bigger arrays on us.

olivers281mo ago

I get what you're saying, @alice242, and I don't totally disagree on the noise point. But I've been doing this long enough to remember when everyone swore 64 channels was the sweet spot for anything useful. That Stanford talk was interesting, but most of those demos were tightly controlled tasks in a lab setting. Real world use, like when someone's moving their head or the cap shifts just a little, those 4 channels can drop out fast. I need the redundancy of more channels to keep the signal stable over a full session. Plus, if you want to move beyond just left/right hand stuff, you start needing those extra electrodes to pick up finer patterns. The hardware companies might push big arrays, but they also sell smaller ones for a reason.