[Eeglablist] Critical pitfall of spectral power analysis?

[Gyurkovics, Mate]

Hi all,

this is a very interesting discussion! I am one of the authors on the paper, and would love to clarify our thinking here.

Makoto is correct - the main point of the paper is simply that the same narrowband change (e.g., a 5 mV increase in alpha) will come out to look much different using dB-correction (or any other type of baseline correction that is based on division) depending on the level of baseline activity in that range: people (or groups) with higher baseline activity (I'm avoiding the term "noise" here) will show an attenuated value compared to people (or groups) with less baseline activity. This makes sense only if we assume that there is a multiplicative relationship between baseline activity and the signal we are interested in (i.e., a 5 mV change is truly smaller if there is more activity because it represents a lesser change in "ongoing activity"). This is our major point - we do not say that divisive baseline correction is wrong, we just say that it only makes sense under a certain set of circumstances.

It is true, though, that we also say that an additive model makes more sense - i.e., where a signal would be added on top of the background activity, rather than be "born out of it" through some scaling factor. We also provide some evidence for additivity. However, our major point about dB correction is that it should be used with caution, and with proper justification especially in individual differences analyses (so where you want to make inferences about certain people or groups having larger or smaller signals). We did not make any points about cross-frequency comparisons, but that is certainly a direction very much worth exploring!

Now, one could argue that this multiplicative model is realistic for alpha, which is practically the only really oscillatory signal in surface recordings. It's easy to imagine that there is some amount of alpha activity (genuine, oscillatory alpha activity) before an event, and then after the event it scales up or down. However, even if that was the case a) this only applies to a very narrow slice of the spectrum, and more importantly, b) even if there is genuine alpha before an event, in the baseline period, there is also broadband noise at that frequency too. So baseline alpha power WILL be made up of noise + signal, and then we are dividing the post-event value (which is noise + bigger-signal or noise + smaller-signal) with this mixture.

My understanding of Daniele's very interesting point is that he suggests that 1/f activity may not represent broadband, non-oscillatory activity, but rather the amplitude modulation of lots of different oscillations and lots of different frequencies? I am familiar with this account, but I do not know of any good evidence for it. It is substantially less parsimonious than presuming that there is non-oscillatory activity going on in the brain which, for different reasons, would have this 1/f like shape in the PSD. In fact, I would almost go the opposite direction and say, there really is only evidence for alpha being oscillatory in surface recordings, true oscillations are super sparse, fairly rare (https://urldefense.com/v3/__https://www.nature.com/articles/s42003-024-06083-y__;!!Mih3wA!EmizTihlbL4yxQegaQdEAD71ZH220exY7LnjCG9XZYGy3qh_2LtAdNv3gSGmairQ6vSmixdnmnVXmqNsGmrc6hfH$ ). Why would the amplitude modulation of lots of (independent?) oscillations produce the typical 1/f shape, and not a wide variety of different shapes based on the state of the given oscillators? Is it just about the low-pass filtering properties of surface recordings?

I think there are different schools of thought on this, but our thinking is definitely in line with the FOOOF-type of thinking - there are broadband and narrowband activities, and these should be separated if we want to quantify either one.

Notably, others have made similar points since (i.e., that it is important to consider whether broadband/aperiodic/non-oscillatory activity and narrowband oscillations are independent or not when correcting for the former) - see e.g., https://urldefense.com/v3/__https://www.nature.com/articles/s41467-024-45922-8__;!!Mih3wA!EmizTihlbL4yxQegaQdEAD71ZH220exY7LnjCG9XZYGy3qh_2LtAdNv3gSGmairQ6vSmixdnmnVXmqNsGglRimuk$  and https://urldefense.com/v3/__https://www.jneurosci.org/content/43/37/6447.abstract__;!!Mih3wA!EmizTihlbL4yxQegaQdEAD71ZH220exY7LnjCG9XZYGy3qh_2LtAdNv3gSGmairQ6vSmixdnmnVXmqNsGpIbmSPp$ ).

Sorry, this email ended up longer than expected, but I hope I managed to clarify our points - we are not against dB correction, we are just saying that it should be used carefully in certain situations (when comparing groups or individuals).

Thanks,
Mate
________________________________
Feladó: eeglablist <eeglablist-bounces at sccn.ucsd.edu>, meghatalmazó: Makoto Miyakoshi via eeglablist <eeglablist at sccn.ucsd.edu>
Elküldve: 2025. augusztus 11., hétfő 17:59
Címzett: eeglablist at sccn.ucsd.edu <eeglablist at sccn.ucsd.edu>
Tárgy: Re: [Eeglablist] Critical pitfall of spectral power analysis?

Hi Jinwon and Daniele,

I've checked that paper recently but haven't read it. Let me guess what the
main problem is, and let me use a simple example below to share
understanding of it.

Subject 1: Baseline-period alpha power magnitude 10 microV^2/Hz, and a task
increased the power by 10 microV^2/Hz. Thus, the power change is 10
microV^2/Hz -> 20 microV^2/Hz, which is 3dB.
Subject 2: Baseline-period alpha power magnitude 100 microV^2/Hz, and a
task increased the power by 10 microV^2/Hz. Thus, the change is 100
microV^2/Hz -> 110 microV^2/Hz, which is 0.41dB.

Thus, even though both subjects showed the same 10 microV^2/Hz power
increase evoked by the task, dB-conversion showed one is +3dB while the
other is +0.41dB.
I guess this is the main point of the problem? I still do not see how the
source independence issue can relate here, but at least this is a part of
the problem and is legitimate, right?

This kind of '1/f slope + peak' conceptualization, together with concepts
such as 'oscillatory', 'non-oscillatory', reminds me of FOOOF (Donoghue et
al., 2020; Gao et al., 2017).

Here is my take:
If someone makes an assertion that that dB-converted calculation is the
ONLY VALID way of quantifying it, s/he is wrong.
Otherwise it is ok to use the dB-converted calculation. It just has
insensitivity in certain aspects. The calculation itself is valid.

A practical merit of using dB-conversion is that cross-frequency
normalization is automatically taken care of.
For example, if you observe 10 microV^2/Hz power increase in theta and
gamma bands, which is more prominent? The latter, right? It's because the
variance of the 'baseline signals' follows 1/f.
So, if you want, we can publish another paper saying that using microV^2/Hz
cannot show the significance of the same 10 microV^2/Hz power increases in
theta and gamma.

These are just thought experiments. My point is that there are usually
trade-offs in these approaches, and it is rare to find that one approach
turned out to be completely wrong. It's usually a matter of distribution of
sensitivities. Also, it should not be too difficult to use multiple
calculations and show the results in parallel. However, I do not know what
EEGLAB developers will do for this issue (will they ever recognize it as an
issue?)

I've also seen a criticism that our inter-trial phase coherence is biased.
https://urldefense.com/v3/__https://onlinelibrary.wiley.com/doi/abs/10.1002/sim.3132__;!!Mih3wA!DDPZ_YphIcjaVH3DU8jtzW_d9dhPFRD0nt95TF-cXHZrIJX7BX9VE_PO2zN4gVthY5GGOkVw0MwTB1uR7e3aMMJ-bQs$

%%%%%%%%%%%%%%%%%%%%%%%%
While writing this response, I saw Daniele's post. I'm curious to hear what
the 'substantial flaw' is in more detail. Looks like my quick and lazy
problem explained above is different from what he means.

By the way, I was happy to find that he wrote '1/f can be generated by
"pure oscillations" with nonuniform amplitude, among other things.' because
I once said exactly the same thing to express my dissatisfaction to hear
how the concept of 'aperiodic' had been misused in some communities!

Makoto

On Sat, Aug 9, 2025 at 1:25 PM 장진원 via eeglablist <eeglablist at sccn.ucsd.edu>
wrote:

> Hi all,
>
> Recently I found one interesting article that addresses the pitfall of
> baseline correction that many scientists have used to transform EEG to
> time-frequency domain. According to this article, power spectrum formation
> is highly exposed to subject-dependent noise that independently affects
> power spectrum regardless of signal. Because I am not an engineer who
> majors signal transformation, I wonder how eeglab could handle this issue
> in spectral power analysis because this article implies that using alpha
> (8-13Hz) or theta (4-8)Hz is totally unacceptable in clinical studies.
>
>
> Reference: Gyurkovics, M., Clements, G. M., Low, K. A., Fabiani, M., &
> Gratton, G. (2021). The impact of 1/f activity and baseline correction on
> the results and interpretation of time-frequency analyses of EEG/MEG data:
> A cautionary tale. NeuroImage, 237, 118192.
>
> https://urldefense.com/v3/__https://doi.org/10.1016/j.neuroimage.2021.118192__;!!Mih3wA!FUy2N9N5bZQJF1IM06-OIaXtDG8YvPWzfrSGxmJE6N_4DPqW9Irqgr9P4PajtadaJV9Jzo1Z9QWJsE2RPNZmbe-Mkw$
>
> Best regards,
> Jinwon Chang
> _______________________________________________
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu or visit
> https://urldefense.com/v3/__https://sccn.ucsd.edu/mailman/listinfo/eeglablist__;!!DZ3fjg!9LHMp1LILVQXqbdj8LiJbD-xKlDvDSIsfJE9YoyF_QRghxtE-v_3UQ8xUNrDoTtTsiMnNb5zDupEQkoQoT0Phvkr6mk$   .
>
_______________________________________________
To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu or visit https://urldefense.com/v3/__https://sccn.ucsd.edu/mailman/listinfo/eeglablist__;!!DZ3fjg!9LHMp1LILVQXqbdj8LiJbD-xKlDvDSIsfJE9YoyF_QRghxtE-v_3UQ8xUNrDoTtTsiMnNb5zDupEQkoQoT0Phvkr6mk$ .