false positive errors on a beh.json file

[Remi-Gau]

Some colleagues tries to use a beh folder to organize their data and has been getting some errors on the json content that look like false positive.

I have been able to reproduce the error here:

• https://github.com/Remi-Gau/dummy_beh_ds/tree/main/sub-01/beh

Here I copied an events.tsv and its side car and renamed it into beh.tsv

Apparently the specifications says nothing of how the content of beh.json should be organized.

So it looks like the validator is complaining as soon as one puts anything in the json file (an empty {} is ok though).

Also note that the validator is not picking up that the beh.tsv has both onset and duration and should then probably be an events.tsv - though see this case in the bids-example.

sub-01_task-TEST_beh.json
0.215 KB | application/json
Location:

dummy_beh_ds/sub-01/beh/sub-01_task-TEST_beh.json
Reason:

Invalid JSON file. The file is not formatted according the schema.
Evidence:

should NOT have additional properties

---

Suggestions:

• lenient: do not put any restriction on how the content of beh.json files should be organized
• strict and more "agressive": assume that beh.json files should have the same structure as events.json (will require an update of the spec)

[sappelhoff]

without having cross-checked this:

• I thought beh.tsv is a tabular file where you write whatever columns you need to represent your behavioral experiment data
• By extension, I thought beh.json is the dictionary, where all columns from beh.tsv are described, following the standard BIDS convention of Description, Levels, etc.

So I think your example should definitely pass.

+1 for being lenient (your 1st suggestion)

I think a check for onset+duration in beh.tsv and a contingent recommendation (warning) that beh.tsv might actually be an events.tsv sounds like a potential (low priority) enhancement 🤔

[Remi-Gau]

Actually how about this change: we recommend in the spec that beh.json should be organized as events.json but do not make it required? This way it at least gives users some "guidance".

[sappelhoff]

Yeah 👍 ... as of now, beh.json is not even explicitly mentioned: https://bids-specification.readthedocs.io/en/latest/04-modality-specific-files/07-behavioral-experiments.html

making a comparison that "similar to events tsv+json, the beh tsv should have a (similar) json file" ... and that that's where all the metadata (cited at the bottom of the page) can/should be reported