[Eeglablist] Clarification Needed on EEG Analysis in EEGLAB

[Makoto Miyakoshi]

Hi Marjan,

> Secondly, regarding the SIFT software, I am concerned about the
implications of not meeting the white noise criterion. Would this pose a
problem, or can the issue be mitigated using surrogate criteria to
eliminate invalid connectivity?

You can find a list of 8 publications using (group)SIFT in this page.
https://urldefense.com/v3/__https://github.com/sccn/groupSIFT__;!!Mih3wA!CkVfWpP4Gb9QQEs4Ii4vF7qgP7wMcWoxCDQYuDv3n-8sRCQ6KyXNanKvGzkfBdgnQbYCNSvdL8ysEIDeN7tD3iXbFgA$ 
Note none of them passed the whiteness test. As the SIFT manual explains,
this test is stringent.

Not passing the test does not necessarily invalidate your results (sorry
for using triple negatives...); it just means not all signals are extracted.
Depending on your experimental hypothesis, achieving such a
perfectionistic goal may not be necessary,.
As Karl Friston wrote (2012), one observation would be sufficient to prove
the existence of a talking dog. So, as long as your 'talking dog' is
present in your results, you are fine. You may be missing programming cats
and driving rats, but it does not matter to your experimental conclusion.

Makoto



On Sun, Jan 26, 2025 at 11:09 AM m za via eeglablist <
eeglablist at sccn.ucsd.edu> wrote:

> Dear all,
>
>  I hope this message finds you well. I am reaching out to seek your
> guidance on two issues I am currently facing regarding EEG analysis using
> EEGLAB.
>
> First, I have a question regarding the channel statistics section. When we
> assess channel normality, I understand that if the channel deviates
> significantly from Gaussian distribution, we consider it non-Gaussian and,
> thus, not EEG. However, I would like to clarify the general assumption we
> make about data distribution when we do not have any distribution data,
> specifically whether we can safely assume it to be normal.
>
> Secondly, regarding the SIFT software, I am concerned about the
> implications of not meeting the white noise criterion. Would this pose a
> problem, or can the issue be mitigated using surrogate criteria to
> eliminate invalid connectivity?
>
> I would greatly appreciate your insights on these matters.
>
> Thank you for your help!
>
> Best regards,
>
> Marjan
> _______________________________________________
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu or visit
> https://sccn.ucsd.edu/mailman/listinfo/eeglablist.
>