Brainwave units are in dB, but have they been scaled?


#1

I want to make sure I’m describing the headband’s readings as accurately a possible in the applications i’m working on. The documentation says the readings are
log band power (dB)

but i’m confused by exactly what that means (i suppose it might just be by the word “log”.) I’ve been reading that different waves generally have certain amplitudes, but the readings we are given all seem to be standardized somehow, fluctuating in the same range. Anyone know the exact origin of these numbers? I feel like i’m skewing something if i simply say, for example, “xx dB of alpha-band activity,” because that’s probably not the exact amplitude going on in the brain at that moment. Any insights?


#2

I suppose this has everything to do with understanding the complicated and enigmatic fourier transform process. Guess i’m going to go with “alpha-band power (dB)”, removing the word ‘log’ as it’s a given that decibels are a logarithmic measurement scale and it seemed redundant.


#3

Hi to_the_sun,

Just to clarify, the EEG units are in microvolts, not dB. The Brainwaves are EEG’s displayed and written out in microvolts.

The fft or fourier transform is a translation from the time domain into the frequency domain. That is to say in the frequency domain the x axis is frequency opposed to time when plotting raw brainwaves.

The frequency domain is commonly in logarithmic scale, known as dB, because the frequency powers degrade logarithmically. This means that as the frequency increases the order of magnitude shifts significantly. An order of magnitude is a multiplication of 10. 10dB is 10 times larger, 20dB is 100 times larger, 30dB is 1000 times larger.

The Delta, Theta, Alpha, Beta and Gamma outputs which we provide are given as relative metrics to their respective frequencies and history. Alpha score of 0.8 means you’re at the 80% alpha rating compared to your own brainwaves.

Hope that all makes sense to you.


#4

Hey Farough,

That all makes sense except for the last part. I was reading about this business of it being relative just now in other posts and it leads me to have a lot questions.

Why is it done relatively? Wouldn’t an absolute measurement be more useful?

How long of a history is it relative to? The past few moments? Days? Lifetime of the headband? What if multiple people are using it?

Does it mean that a score of 1.0 is the highest that’s been seen (in the history it’s keeping track of) and 0.0 is the lowest? What if you improve; would it be possible to see scores higher than 1.0? <edit> Or is it more of a ‘change of’ measurement, relative only to the immediately preceding reading?

I assume then that the way to get absolute measurements would be to conduct my own fft? <edit> Or are the raw fft readings absolute?

Thanks for the info


#5

Hi,

It’s done relatively because the brain is not only extremely complex, but also quite different between individuals. That is to say maybe person A has 8 times the output of Alpha Waves than person B. For this reason you want to score your alpha waves against your own brain. Therefore, the relative measurement to yourself is a lot more meaningful than an absolute reading of Alpha.

The history is fairly short, the history is stored locally on Muse-IO as it streams. Definitely does not exceed the duration which you’re streaming. It also is fairly robust to other people using it because the range will shift significantly. ie. Person A puts on Muse and they have an absolute alpha of 2-3 and then person B takes it with a range of 5-6. The drastic difference in the switch will cause the history to quickly tip into B’s range because we’ve detected such a significant shift. This is a simplified model, but the general idea persists.

The new highest entry will score at 1, the lowest at 0. Your improvements in one sitting will be adapted quickly, your improvements cross session will be unseen relative to these scores. The scores are meant to represent maintaining a certain level and trying to prevent yourself from dropping on your personal scale.

All the fft are absolute, feel free to use them to write your own DSP, scores and outputs of course. You can compute your own FFT’s as well if you please.


#6

Thanks, that’s very clarifying.

One question about what you said though

If so, wouldn’t one expect to see a lot of solid 1.0 and 0.0 readings when you first put the headband on (or at least a lot of extremes, while it settles down and hones in on your personal range)? If i have my program up and running when i turn the band on i should receive every osc message it puts out, but the very beginnings of my sessions don’t seem to look noticeably different than afterwards.

Also, if the history is fairly short, what if you wear the headband for a long time or keep a certain mental state for a long time? Wouldn’t it begin to “forget” your relative baseline and look like your score is petering out, even if it might not be?


#7

Hi Farough!..

So If I understand correctly : Does a 0.2 Alpha score compared to a 0.4 Alpha is a logarithmic progression (based on the subject frequencies and history) ?

Thx


#8

[I]if at certain time point, Alpha value is 0.2, does it mean at that moment Alpha/(Alpha+Theta+Alpha+Beta+Gamma) = 0.2 or [/I][I]Alpha/(the previous highest Alpha) = 0.2[/I][I]? [/I] Thanks.


#9

The scores are a linear progression.

Alpha is relative to itself, not the total power of the system. It’s not a ratio as much as a history.

Alpha of X will become Alpha of X + Y, where X is the previous state and Y can be both a positive and negative migration from the previous X.
ie X = 0.4 and Y = 0.1, X = 0.3 and Y = -0.2.