[ENH] add optional presentation software name, version, OS, and code to events.json

[Remi-Gau]

Mentioned in #558

Another low hanging fruit of metadata to add to any json for anything that involves presenting stimuli to a participant.

One question: should something like this be included in for beh json files out of internal consistency?

[sappelhoff]

@bids-standard/bep_leads please review this proposal

[robertoostenveld]

I like the proposal.

I do believe that StimulusPresentationXXX would be better (i.e. more commonly used) than StimuliPresentationXXX. That is also what returns more hits if you Google for an exact match, or if you search on PubMed for exact matches (like this).

should something like this be included in for beh json files out of internal consistency?

Yes, I think it should. However, there is more information missing from behavioural-only datasets, e.g. InstitutionName or TaskName are also missing. Since there is no JSON for the behavioural recording itself (only an optional data dictionary for the events table), it is not obvious to me where it should be represented.

[Remi-Gau]

I do believe that StimulusPresentationXXX would be better (i.e. more commonly used) than StimuliPresentationXXX. That is also what returns more hits if you Google for an exact match, or if you search on PubMed for exact matches (like this).

Good point. Will update.

[Remi-Gau]

Yes, I think it should. However, there is more information missing from behavioural-only datasets, e.g. InstitutionName or TaskName are also missing. Since there is no JSON for the behavioural recording itself (only an optional data dictionary for the events table), it is not obvious to me where it should be represented.

I believe all of this should be part of another PR, no?

[robertoostenveld]

I believe all of this should be part of another PR, no?

Indeed. I think that PR should consolidate Task Events (5) and Behavioural experiments (7) and make Behavioural data a first-class citizen.

[nicholst]

Agree with @robertoostenveld that the repetition of the Behavioural data descriptions is not good, but not sure what making it a first-class citizen entails.

Would all Behavioural data for all modalities live in "Modality specific files -> Behavioral data"?

[robertoostenveld]

I also don't know what it would entail but know that according to the spect "behaviour without fMRI" (and also without MEG, EEG and IEEG) is now allowed to be stored in a BIDS dataset. However, that now results in a very underspecified dataset with hardly any metadata. It would be relevant to consider how the relation of behavioural data to functional/biological data is optional, as is the joint (whether simultaneous or not) recording of multiple other data types (e.g. MEG with an anatomical MRI). This is a discussion (on "multi-modal") we had before on the mailing list but that has not resulted in visible improvements to the specification.

But I think that this discussion falls outside the scope of this PR and might better be discussed in the format of a full-fledged BEP.

[Remi-Gau]

But I think that this discussion falls outside the scope of this PR and might better be discussed in the format of a full-fledged BEP.

I agree. So we "just" need a volunteer to take the lead on that...

[poldrack]

note that there is a parallel effort called Psych-DS ( https://psych-ds.github.io/) that is working on a BIDS-like spec for behavioral data. so before going down the road of a full-on BEP we should consider whether we might instead try to adapt Psych-DS or at least work in parallel so that our efforts are aligned.

…

On Mon, Aug 17, 2020 at 4:22 AM Remi Gau ***@***.***> wrote: But I think that this discussion falls outside the scope of this PR and might better be discussed in the format of a full-fledged BEP. I agree. So we "just" need a volunteer to take the lead on that... — You are receiving this because you are on a team that was mentioned. Reply to this email directly, view it on GitHub <#573 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAGUVEGDMRBIK73VSL4CBWDSBEHIDANCNFSM4QAIJ7VA> .

-- Russell A. Poldrack Albert Ray Lang Professor of Psychology Building 420 Stanford University Stanford, CA 94305 poldrack@stanford.edu http://www.poldracklab.org/

[Remi-Gau]

note that there is a parallel effort called Psych-DS ( https://psych-ds.github.io/) that is working on a BIDS-like spec for behavioral data. so before going down the road of a full-on BEP we should consider whether we might instead try to adapt Psych-DS or at least work in parallel so that our efforts are aligned.

Yup I keep thinking of psych-DS every time we talk about behavior-BIDS. The way I would see the behavioral support in BIDS would have to be very minimalist IMHO and simply make sure that for example things that are included in all the JSON of all other modalities, should make be also supported for the behavior-BIDS. But might be worth even just clarify that amongst us and with Psych-DS.

[emdupre]

we should consider whether we might instead try to adapt Psych-DS or at least work in
parallel so that our efforts are aligned.

Strong +1 from me.

The way I would see the behavioral support in BIDS would have to be very minimalist IMHO and simply make sure that for example things that are included in all the JSON of all other modalities, should make be also supported for the behavior-BIDS.

I think this makes sense. But having a conversation earlier than later with PsychDS might be helpful, too. For example, the set of minimal meta-data we'd want for movie-watching stimuli (like the EAN of the used DVD) will be very different than for a standard cognitive task. And it's not an uncommon use case for BIDS !

[poldrack]

looping in @mekline to start the conversation

[mekline]

Hi all, thanks for tagging me in. I would be very open to making PsychDS work well for BIDS, or otherwise aligning as far as possible! PsychDS was originally inspired by BIDS, so I'm very happy to see there's some interest here! A couple initial notes, and I'd love to set up a longer conversation as well.

• The primary goal of PsychDS is to make minimal metadata and data structuring accessible to psychologists without extensive training in a particular programming language, and to provide a common starting point for people developing experimental/analytic software. The 'modal' user might be someone who runs experiments without a presentation script or scripted data collection, but instead e.g. hand-coding of responses into a spreadsheet (or alternately, someone who works with idiosyncratically structured data from a 3rd party and needs to build a robust analysis pipeline).

• Compared to BIDS, the types of data I've been encountering are both much smaller, and much less regular (in experimental structure, types of information captured, and level of 'processing' in the initial available state) than fMRI data. We do aim for this behavioral output (whatever it might be) to be the 'first class citizen' of the specification, if that's a useful distinction.

• We are focusing right now on building the validator app, and we could really use help with it! The basic technical infrastructure is a javascript core that will power both an R package and a browser version (https://github.com/psych-ds/validator-prototype).

[effigies]

Small comments... Haven't reviewed in general.

[Remi-Gau]

Hi all, thanks for tagging me in. I would be very open to making PsychDS work well for BIDS, or otherwise aligning as far as possible! PsychDS was originally inspired by BIDS, so I'm very happy to see there's some interest here!

@mekline Thanks for the update. 😄

I am thinking that maybe members of the BIDS and psych-DS community who are interested could have a video meeting to maybe figure out how to best coordinate our efforts, no?

[mekline]

Definitely - I"m guessing this github issue isn't the best channel to reach all interested parties - should I write to bids-discussion, or is there some better place to reach likely interested people?

…

On Tue, Aug 18, 2020 at 5:23 AM Remi Gau ***@***.***> wrote: Hi all, thanks for tagging me in. I would be very open to making PsychDS work well for BIDS, or otherwise aligning as far as possible! PsychDS was originally inspired by BIDS, so I'm very happy to see there's some interest here! @mekline <https://github.com/mekline> Thanks for the update. 😄 I am thinking that maybe members of the BIDS and psych-DS community who are interested could have a video meeting to maybe figure out how to best coordinate our efforts, no? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#573 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABUA75JW2LDXHET4IF7OCA3SBJCBTANCNFSM4QAIJ7VA> .

[Remi-Gau]

Definitely - I"m guessing this github issue isn't the best channel to reach all interested parties - should I write to bids-discussion, or is there some better place to reach likely interested people?

Yup I think that the bids-discussion is the best AFAICT and then we can spread the words in other channels if we set up a video call or something. :-)

[Remi-Gau]

@effigies I believe that the number of files changed is now equal to 1. 😉

[effigies]

Looks good. Mostly formatting suggestions. Significant rewording suggestion for one field.

[CPernet]

@Remi-Gau @poldrack @robertoostenveld @mekline is there another thread/doc open somewhere?
happy to help structure better behavioural data -- I, for instance, made a tool to read spreadsheets and spit out participants.tsv and particpants.json to indeed help as much as possible people (easier to edit particpants.json that to create without making mistakes, like missing {} , ; etc)

[Remi-Gau]

@Remi-Gau @poldrack @robertoostenveld @mekline is there another thread/doc open somewhere?
happy to help structure better behavioural data -- I, for instance, made a tool to read spreadsheets and spit out participants.tsv and particpants.json to indeed help as much as possible people (easier to edit particpants.json that to create without making mistakes, like missing {} , ; etc)

There is no other public thread as far I know. Will open one on the BIDS google group now to keep things separated because this PR is starting to resemble to my kitchen after I try making bread.

[ChristophePhillips]

@CPernet Happy to help on participants information import.
Got to deal with that too... sharing resources and maintaining some efficient tool seems a good way to proceed.

[Remi-Gau]

@CPernet Happy to help on participants information import.
Got to deal with that too... sharing resources and maintaining some efficient tool seems a good way to proceed.

@CPernet @ChristophePhillips might be worth seeing if some of that can be integrated into bids-matlab

[Remi-Gau]

• Updating the glossary of the BIDS starter kit so we can link to specific words: ✔️
• Moving URI link in the link section: ✔️
• Updating DOI: ✔️
• Nuking one of @sappelhoff comments because it somehow showed twice on my browser: ✔️ (let me know if you want me to get it back)

Not sure what's left on that one. Let me know.

[sappelhoff]

I am +1 for merging. The uri link can be improved upon in the following PR related to #601

[sappelhoff]

@hoechenberger fine with you?

[hoechenberger]

LGTM!

[sappelhoff]

Thanks for your patience and persistence and for bringing this up in the first place @Remi-Gau :-)

Thanks also to the many reviewers!