'custom_ref_applied' is True even if the offline rereferencing is an average

[nabilalibou]

Proposed documentation enhancement

Hello,

contrary to what is written in the doc, 'custom_ref_applied' is True event if the offline rereferencing is an average one.

set_eeg_reference(ref_channels="average") raises the flag 'custom_ref_applied' even so it is said for this Attribute on https://mne.tools/stable/generated/mne.Info.html:

custom_ref_applied: Whether a custom (=other than average) reference has been applied to the EEG data. This flag is checked by some algorithms that require an average reference to be set.

I find it strange that this flag is raised by an 'average' reref (which rereferencing does not raise it then ?) but I am guessing that there are reasons about this. if anyone could enlighten me as to the reason behind it, I'd be grateful :). But in any case i think that this little doc paragraphe should be changed to reflect how it really works. Since I'm already on this similar issue I volunteer to continue to spot and correct the little contradictions around rereferencing and projections in the doc.

[larsoner]

Yeah it should say "a custom (=other than average projector)` or similar. It's for inverse modeling reasons, we need to make sure an average reference over the remaining good channels is applied. Adding an average reference projector as the reference is a safe way to do this.

I find it strange that this flag is raised by an 'average' reref (which rereferencing does not raise it then ?) but I am guessing that there are reasons about this. if anyone could enlighten me as to the reason behind it, I'd be grateful :).

Let's discuss this over in #12369.

[nabilalibou]

Ok thanks for the link , I'll follow the discussion https://github.com/mne-tools/mne-python/issues/12369. I'll open a PR to add this detail on the doc

[larsoner]

I think this was taken care of in #12420 and #12426