[FIX] Clarify use of acq and task parameters in EEG, MEG, and iEEG

[sappelhoff]

While @robertoostenveld and I went over #173, we noticed several minor inaccuracies with the entities that recording modalities can have.

This PR mainly targets a consistent and sensible use of the acq parameter, as also discussed in #194 (comment)

Furthermore, one superfluous task label is deleted from EEG electrodes.

In Robert's words:

My confusion is only with task for headshape/photo/electrodes. Those are measurements which are task-independent. See https://bids-specification.readthedocs.io/en/stable/04-modality-specific-files/03-electroencephalography.html#electrodes-description-_electrodestsv which includes task as required. I think that is incorrect. Electrode data is not recorded while the subject performs a task (except for "sit still!")

[choldgraf]

A minor typo in there. I'm a little bit confused about this change, but I'll defer to your judgment. Is the problem that a particular shape of electrodes cannot be task-dependent? (as opposed to session-dependent?)

Also can you confirm that this is only updating the text to adhere to the current BIDS spec, and not changing the spec itself?

[sappelhoff]

Is the problem that a particular shape of electrodes cannot be task-dependent? (as opposed to session-dependent?

Fair point - I was assuming that electrodes are set up on the participant only once per session, because I never heard of another design. But of course that doesn't mean it does not exist ... what do you say @robertoostenveld ?

Also can you confirm that this is only updating the text to adhere to the current BIDS spec, and not changing the spec itself?

Given that iEEG electrodes don't have a task label either (see here), I have the impression that in EEG this task label was erroneously copy pasted.

[choldgraf]

Just to be clear, those points weren't pushing back on either of those assumptions (they seem reasonable to me), just wanted to clarify them!

[robertoostenveld]

My review aims at the general logic, not at the consistency between this table and the documentation elsewhere.

• if acq is optional for MEG/EEG/iEEG, it should also be optional for channels.tsv. Otherwise you could have two recordings that only differ in the acq label for which you cannot distinguish the channel details. This might e.g. be the case with clinical iEEG, which sometimes gets "spliced" into different amplifiers.

• task should not be required for eeg/ieeg electrodes (it is just like headshape and photo)

Other than that it all looks good.

[robertoostenveld]

Is the problem that a particular shape of electrodes cannot be task-dependent? (as opposed to session-dependent?

Fair point - I was assuming that electrodes are set up on the participant only once per session, because I never heard of another design. But of course that doesn't mean it does not exist ... what do you say @robertoostenveld ?

Agreed, I would be fine with the statement that an EEG "session" corresponds to a single application of electrodes on the subject. That does not cover all cases, but should (at this moment) cover >95%. Gel-based EEG electrodes usually require some cleanup. But an example that is not covered: using dry EEG electrodes, e.g. in a BCI experimental setup, and then reapplying the cap or using another cap.

[sappelhoff]

if acq is optional for MEG/EEG/iEEG

The point of this PR is to remove acq for MEG/EEG/iEEG time series, because the parameter does not make sense for data that is not stable (as opposed to a picture of a face, or the T1 scan of a brain)

Thus I would argue that channels.tsv also does not need an acq parameter

task should not be required for eeg/ieeg electrodes (it is just like headshape and photo)

yep, I removed the task parameter for electrodes. see the changes

Example that is not covered: using dry EEG electrodes, e.g. in a BCI experimental setup, and then reapplying the cap or using another cap.

I think we can work with these issues when they pop up.