[Eeglablist] Which is the best way to measure the "alpha" oscillation?

Thu May 14 10:25:14 PDT 2026

Hi Scott,

I completely agree with you. I was just mentioning a method if someone wants to obtain this oversimplified "IAF" measure, which tries to address the simple problems of split peaks, ambiguous peaks, etc. But still over an entire recording, and I agree that it is very misleading.
At the end of my email, I mentioned that your and Julie's IMA approach is the best. Or any method that, as you said, can model well the different central tendencies of alpha oscillations within session, within subjects, and across subjects. I hope to have an opportunity soon to try the IMA plugin and run this type of analysis.

- IMA takes an approach orthogonal to FOOOF. Wherease FOOOF considers the *form of the log spectrum* of a data source - in itself - IMA pays no attention to the mean log spectrum (removing it from consideration first of all). IMA then considers the following question: What maximally distinct modes of *log spectral variability* does the data source exhibit across time? IMA, for example, could possibly isolate multiple modes whose summed activities across time (i.e., in the grand mean spectrum) happened to cancel each other out at frequencies of interest. Here, FOOOF would not find any evidence of them.

This is a very interesting point from your previous email. I will make sure this is looked at in the ongoing community project on 1/f / fooof / aperiodic activity.

Thank you,

Cedric

Sent with [Proton Mail](https://urldefense.com/v3/__https://proton.me/mail/home__;!!Mih3wA!APd-2_tZlNmRq_OmhshOOp0zLZDgmYpfoM_PeYTkV4LPEM3O8TTvlkUNGYsPY5GKxTqelQPpCNC8a1ZMnxHY5_Rmrw$ ) secure email.

On Thursday, May 14th, 2026 at 10:11 AM, Cedric Cannard <ccannard at protonmail.com> wrote:

> Hi Scott,
>
> I completely agree with you. I was just mentioning a method if someone wants to obtain this oversimplified "IAF" measure, which tries to address the simple problems of split peaks, ambiguous peaks, etc. But still over an entire recording, and I agree that it is very misleading.
>
> At the end of my email, I mentioned that your and Julie's IMA approach is the best. Or any method that, as you said, can model well the different central tendencies of alpha oscillations within session, within subjects, and across subjects. I hope to have an opportunity soon to try the IMA plugin and run this type of analysis.
>
> Cedric
>
> Sent with [Proton Mail](https://urldefense.com/v3/__https://proton.me/mail/home__;!!Mih3wA!APd-2_tZlNmRq_OmhshOOp0zLZDgmYpfoM_PeYTkV4LPEM3O8TTvlkUNGYsPY5GKxTqelQPpCNC8a1ZMnxHY5_Rmrw$ ) secure email.
>
> On Wednesday, May 13th, 2026 at 10:41 PM, Scott Makeig <smakeig at gmail.com> wrote:
>
>> Cedric -
>>
>> When you write of detecting (the) "Individual Alpha Frequency (IAF)", you reify the term [introduced 30 years ago by Klimesch](https://urldefense.com/v3/__https://journals.lww.com/clinicalneurophys/fulltext/1996/11000/Alpha_Frequency,_Reaction_Time,_and_the_Speed_of.6.aspx?casa_token=oAAhkPMByY0AAAAA:OXBNWqxRPYfRdk-oqwrPa9gGH8oibz3ujgYREp5YQ1kRsHCjfCozo8ChE6KN2k4MoP-x8NBacQQygifizS7MNJw__;!!Mih3wA!APd-2_tZlNmRq_OmhshOOp0zLZDgmYpfoM_PeYTkV4LPEM3O8TTvlkUNGYsPY5GKxTqelQPpCNC8a1ZMnxHEHfl6-g$ ) -- a term that our work (specifically, results shown in [this poster](https://sccn.ucsd.edu/~julie/AlphaPosterMini.pdf ) by Julie Onton) demonstrated to us was clearly a major oversimplification. Alpha range activities, whether from occipital/parietal ('alpha'), somatomotor ('mu'), auditory ('tau') cortex, or elsewhere, are in general not fixed within subject -- neither over space (cortical source location) nor over time (within session, as shown in [this poster)](https://sccn.ucsd.edu/~julie/AlphaIMposter.pdf ).
>>
>> I've often seen how, in science, giving some phenomenon a (singular) name can give rise to an uncritically held belief that what is being named is in fact a singular phenomenon -- e.g., 'the' (supposedly unitary) 'P300' ERP peak versus its other originally proposed designation ('Late Positive Complex (LCP)' summing distinct evoked activities in multiple cortical areas. In these papers, we showed a late positive peak in scalp ERPs (across a range of scalp channels) can be accounted as summing positive-going potentials (with differing time courses) from a number of cortical areas whose projected signals -- either [across the session](https://urldefense.com/v3/__https://journals.plos.org/plosone/article/file?id=10.1371*journal.pbio.0020176&type=printable__;Lw!!Mih3wA!APd-2_tZlNmRq_OmhshOOp0zLZDgmYpfoM_PeYTkV4LPEM3O8TTvlkUNGYsPY5GKxTqelQPpCNC8a1ZMnxFvimQzQQ$ ) or [across ERPs](https://urldefense.com/v3/__https://www.jneurosci.org/content/jneuro/19/7/2665.full.pdf__;!!Mih3wA!APd-2_tZlNmRq_OmhshOOp0zLZDgmYpfoM_PeYTkV4LPEM3O8TTvlkUNGYsPY5GKxTqelQPpCNC8a1ZMnxFcDeDXLQ$ ) each averaging event-related activity in one of the many task conditions -- are maximally distinct.
>>
>> Here, the example is the concept of 'the' IAF. giving it a unitary name ('the IAF') does not means it exists as such -- though the claim did build on early [visual observations](https://urldefense.com/v3/__https://journals.sagepub.com/doi/pdf/10.1177/003591575705001013__;!!Mih3wA!APd-2_tZlNmRq_OmhshOOp0zLZDgmYpfoM_PeYTkV4LPEM3O8TTvlkUNGYsPY5GKxTqelQPpCNC8a1ZMnxFfTadagA$ ) (more than 70 years ago) that alpha peak frequencies in EEG data recorded under similar conditions can and do differ between individuals. [Note interesting fact: the first EEG Fourier analysis was [reported by Grass](https://urldefense.com/v3/__https://journals.physiology.org/doi/pdf/10.1152/jn.1938.1.6.521__;!!Mih3wA!APd-2_tZlNmRq_OmhshOOp0zLZDgmYpfoM_PeYTkV4LPEM3O8TTvlkUNGYsPY5GKxTqelQPpCNC8a1ZMnxHE4SGSmQ$ ) nearly 90 years ago -- in 1938!]
>>
>> On Wed, May 13, 2026 at 7:31 PM Cedric Cannard via eeglablist <eeglablist at sccn.ucsd.edu> wrote:
>>
>>> There is also the non-parametric technique for detecting Individual Alpha Frequency (IAF) developed by Corcoran:
>>> https://urldefense.com/v3/__https://pubmed.ncbi.nlm.nih.gov/29357113/__;!!Mih3wA!Fo8w8N5UBFlY7iDyii253gF46GFSvgj4a3s8T2MvPfEjRiwul0MvwVpdQewM9tiwTQWymxmZ5e-GTAXmq2V2j4I-Ew$
>>> It offers both the alpha peak frequency, and center of gravity, to account for I individuals with split peaks, absent peaks, etc.
>>> And detects the insidious end frequency bounds from the data, so it is assumption free.
>>>
>>> -> This method is easily available via the BranBeats EEGLAB plugin (feature extraction mode): https://urldefense.com/v3/__https://github.com/amisepa/BrainBeats__;!!Mih3wA!Fo8w8N5UBFlY7iDyii253gF46GFSvgj4a3s8T2MvPfEjRiwul0MvwVpdQewM9tiwTQWymxmZ5e-GTAXmq2XDs00OGw$
>>>
>>> Although I think Scott’s IMAT recommendation is the strongest.
>>>
>>> Cedric
>>>
>>> Sent from Proton Mail for iOS.
>>>
>>> -------- Original Message --------
>>> On Sunday, 05/03/26 at 07:59 m za via eeglablist <eeglablist at sccn.ucsd.edu> wrote:
>>> Hi all,
>>>
>>> I think one key limitation of many traditional approaches is that they rely
>>> purely on stationary spectral analysis, while EEG is inherently
>>> non-stationary and dynamic.
>>> In that sense, time–frequency methods (e.g., wavelet-based approaches) can
>>> provide a more informative characterization of alpha by capturing its
>>> temporal variability, rather than relying only on averaged spectral power.
>>> At the same time, separating oscillatory peaks from the aperiodic
>>> background (e.g., using methods like FOOOF (Fitting Oscillations and One
>>> Over F)) is important to avoid confounds in alpha power estimation.
>>> This is particularly important given inter-individual variability, where
>>> using individualized peak frequencies and accounting for aperiodic activity
>>> can improve the reliability of alpha characterization. Adaptive
>>> decomposition methods may also offer complementary ways to capture
>>> subject-specific structure, although this requires further validation.
>>>
>>> On Thu, 30 Apr 2026, 06:10 장진원 via eeglablist, <eeglablist at sccn.ucsd.edu>
>>> wrote:
>>>
>>>> Hi all,
>>>>
>>>> There have been long controversies on measuring alpha frequency power. Some
>>>> researchers (especially in clinical fields where electrical engineering is
>>>> not familiar) use frequency bands (8-12Hz or 8-13Hz) with FFT or Welch's
>>>> method to obtain spectral power. Other behavioral scientists prefer
>>>> subdivisions such as lower alpha band (8-10Hz) and higher alpha band
>>>> (10-12Hz). Recent advancement on FOOOF also enables the isolation of
>>>> periodic components to discover individual frequency peaks. There are
>>>> numerous other techniques that could specify the regions of eeg activities.
>>>> Which do you think is the best way to characterize the neurophysiological
>>>> activity often represented as "alpha" oscillation?
>>>>
>>>> Best Regards,
>>>> Jinwon Chang
>>>> _______________________________________________
>>>> To unsubscribe, send an empty email to
>>>> eeglablist-unsubscribe at sccn.ucsd.edu or visit
>>>> https://sccn.ucsd.edu/mailman/listinfo/eeglablist  .
>>>>
>>> _______________________________________________
>>> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu or visit https://sccn.ucsd.edu/mailman/listinfo/eeglablist  .
>>> _______________________________________________
>>> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.ucsd.edu or visit https://sccn.ucsd.edu/mailman/listinfo/eeglablist .
>>
>> --
>>
>> Scott Makeig, Research Scientist and Director, Swartz Center for Computational Neuroscience, Institute for Neural Computation, University of California San Diego, La Jolla CA 92093-0559, [http://sccn.ucsd.edu/~scott%5D(http://sccn.ucsd.edu/%7Escott) 