Frequency Bands

The brain has a number of more-or-less intrinsic frequency bands defined by the organization of neural circuitry and the properties of the neurons; a full discussion is beyond the scope of this web page, but there are many references out there for you to read.

There are almost as many definitions of these bands as there are authors, but a rough guide is as follows:

• Delta (2–4 Hz)
• Theta (4–8 Hz)
• Alpha (around 10 Hz)
• Beta (15–30 Hz)
  • Low beta (15–25 Hz)
• Gamma (30–50 Hz)
  • High gamma? (50–100 Hz)

High Gamma

What is the significance, both physical and cognitive, of higher gamma? Is there any reason to analyze it, or is it just noise?

Well, for one thing, it's very low amplitude, so it's harder to analyze. Also, you have to get rid of the 60 Hz line noise (that's easy, but then you are missing some data). Or, one idea is to NOT get rid of the 60 Hz, and assume that it won't bother SAM, because it's not a dipolar source. I'm not sure how well that works.

As for the meaning of such signals, well, one way to think about it is as broadband, asynchronous noise. That is, you have multiple sources like this:

....|......|.......|.|.....||.............|....|.....
 ..|......|....|......|.........||......|.|...|....|..
 ....|...|.......|......|...|......||......||....|.|..

and so on, where the "|" are post-synaptic events generating current flow. Then, if you add all those together, you get lots of events in a short time, and a broad high-frequency spectrum. That is, it's not an oscillation at that frequency, it's more like the noise of many (individually low frequency) sources overlapping.

But detecting changes in power levels in these bands is hard, because the individual signals are very small, well below the noise threshold of the sensors. You need a LOT of data to get a decent SNR.

With all of that said, go ahead and try it. You might see changes in activity level…