[FieldTrip] a few questions on separating fractal and oscillatory components in the data

[Ivaylo Iotchev]

Jan Mathijs,

thank you dearly(!) for the fast response! Just to make my last thought
clear: I expected oscillatory power and aperiodic power to not be strongly
correlated with each other after separation, because I assume that
correlation would mean one is still somewhat inside the other. For example,
prior to separating them from each other, power in all bands likewise
strongly correlated with each other, and in the same direction (theta
with alpha, alpha with delta and so on...) which I understood as "the
aperiodic part is the biggest part in my signal, so it makes the
frequencies that float on top of it correlate with each other", does that
make any sense to you, logically speaking?

I would also value your opinion on *using division instead of subtraction*
(i.e. raw signal./aperiodic signal, instead of raw signal - aperiodic
signal). I think in my data, if I want to correct for aperiodic signals at
all, it makes more sense, because my trials' spectra do not form clean
straight lines in the log-log space as Gerster et al. recommend, so the
tiniest errors in fitting the slope of the aperiodic signal could eliminate
frequency information that is actually there (in my estimation the
proportion of aperiodic to periodic power in my data is magnitudes bigger
than already in humans, my animals have bigger muscles on the top of the
scalp and the EEG is extracranial, which is why without using either
relative power or aperiodically corrected power, frequency specific power
is highly intercorrelated).

If you have thoughts on any of these, let me know, otherwise thank you for
the time you already put in this

Bests,

Ivo

Am Mi., 25. Feb. 2026 um 10:55 Uhr schrieb Schoffelen, J.M. (Jan Mathijs)
via fieldtrip <fieldtrip at science.ru.nl>:

> Hi Ivaylo,
>
> It seems that you have certain expectations about correlations that are
> not met by your data. I am not sure whether this warrants the conclusion
> that ’the separation fails with the default approach’.
>
> It’s unclear what you correlate with each other, and why you expect these
> measures to be (linearly?) correlated or not. All in all this makes it hard
> to reply in a targeted fashion.
>
> Regarding your statement of using medians for integration: this is what
> the irasa folks came up with, so it’s standard practice in the irasa
> universe.
>
> Best wishes,
> Jan-Mathijs
>
>
> > On 23 Feb 2026, at 17:11, Ivaylo Iotchev via fieldtrip <
> fieldtrip at science.ru.nl> wrote:
> >
> > Dear community,
> >
> > this is a rather general question, because it is not limited to how
> IRASA is used within Fieldtrip, but whoever feels like answering feel free
> to tie your answers to Fieldtrip functions so that it is useful for
> everyone here.
> >
> > I came across the slope phenomenon rather recently, and the first
> general question is whether component separation always works and makes
> sense for every type of data? I noticed that my frequency specific
> components each remain positively correlated with the aperiodic part (and
> each other), strongly so (r > .90) thus I suspect that the separation
> failed with the default approach.
> >
> > For this reason I need to ask the legitimacy of a few alternative
> approaches, that make sense to me, but might be entirely wrong and I just
> don't see it
> >
> > 1. Divide the mixed spectrum by the aperiodic one, point by point. For
> me this reduces the correlation between oscillatory and aperiodic parts,
> but I do not see anyone else doing this
> >
> > 2. (in combination with the above) it seems to me that using medians to
> integrate power across bins is a good idea, but again this does not seem to
> be standard practice
> >
> > Just let me know what you think or know about this, and again feel free
> to incorporate Fieldtrip usage in your answer to make this more valuable
> for the community.
> >
> > Best wishes,
> >
> > --
> > Dr. Ivaylo Iotchev. Researcher. Active in the fields of ethology,
> psychology, and neuroscience
> > Eötvös Loránd University, Budapest
> > _______________________________________________
> > fieldtrip mailing list
> >
> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmailman.science.ru.nl%2Fmailman%2Flistinfo%2Ffieldtrip&data=05%7C02%7Cfieldtrip%40science.ru.nl%7Cc31e6076fbe2416e84aa08de7457d45a%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C639076117738994880%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=E%2F5gmER%2FkK4p%2FKETSBZYQvpjUbW3QnpIBTFYINGvc3Q%3D&reserved=0
> >
> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdoi.org%2F10.1371%2Fjournal.pcbi.1002202&data=05%7C02%7Cfieldtrip%40science.ru.nl%7Cc31e6076fbe2416e84aa08de7457d45a%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C639076117739014337%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=XnX6p7rc0OyED6HWIcWIvFf90njd9zkltUQXirOd1fI%3D&reserved=0
>
>
> _______________________________________________
> fieldtrip mailing list
> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmailman.science.ru.nl%2Fmailman%2Flistinfo%2Ffieldtrip&data=05%7C02%7Cfieldtrip%40science.ru.nl%7Cc31e6076fbe2416e84aa08de7457d45a%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C639076117739031770%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=7pactfQ1LBgs3wFxvcj0GMtXS8KpqG%2FxZpV4GR36HS4%3D&reserved=0
> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdoi.org%2F10.1371%2Fjournal.pcbi.1002202&data=05%7C02%7Cfieldtrip%40science.ru.nl%7Cc31e6076fbe2416e84aa08de7457d45a%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C639076117739049211%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=ZxW3F3Xjzug5ulvjlr7gHNRwtcFX%2FdViLqBULLBrR28%3D&reserved=0
>


-- 
Dr. Ivaylo Iotchev. Researcher. Active in the fields of ethology,
psychology, and neuroscience
Eötvös Loránd University, Budapest
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20260225/21929019/attachment.htm>