[BUG] Where are physio files placed when simultaneously recording (e.g. fMRI + EEG)

[oesteban]

Describe your problem in detail.

@effigies caught this issue while working on #1750 (https://github.com/bids-standard/bids-specification/pull/1750/files#r1541231904).

At the moment, the specs are quite unclear regarding where the physio recordings should be placed when they correspond to experiments where more than one single modality was obtained (e.g., fMRI and EEG).

Some solutions have been brainstormed, but this probably requires more visibility than a loose thread within a PR fixing something different.

Describe what you expected.

A clear exposition of what to do in such cases.

BIDS specification section

No response

[Remi-Gau]

My hot take on this.

Simultaneous recordings are usually "signified" / "detectable" by a proper use of scans.tsv where 2 files acquired at the same time will have the same acquisition time value.

If one of those datafile has a companion physio files, it should follow that the physio file should be valid for the other datatype.

I would say that even if my hot take is correct, this is nowhere explained in any of our doc.

[effigies]

I thought at one point there was guidance about simultaneous EEG/fMRI, but I don't know where it is, now. It is beyond the spec, in any case, and it seems unlikely that tools will depend on detecting it as opposed to just assuming it.

Either way, I would expect these files to have a common set of entities (sub, ses, task, run), which would allow them to find the associated physio.tsv.gz via the inheritance principle if it were pulled outside the func/ or eeg/ directories.