_task-movieX or _task-movie_run-X ?

[yarikoptic]

This question came up with our users after bids-validator started to complain about files for the same task having different lengths (I believe, forgot on which dataset ATM to give an example).

"Historically", for the same movie (e.g. Indiana Jones Raiders of the Lost Ark) split into multiple parts and presented across different runs, we just used `_task-movie_run-X`, which kinda made sense since it is "the same movie", thus kinda "the same task".

But we started to wonder either we better use `_task-movieX` for the part X (where X is the index 1,2,3,4...).  This would open possibilities to annotate for multiple presentations of the same portion of the movie with additional `_run-{1,2,...}` and also make it clear(er) that it was different stimulation and possibly different cognitive task, since neural responses to different parts of the movie might be dominated by different cognitive processes.  Then also validator wouldn't raise an alarm for runs of the same task having different lengths when movie is not split evenly.

What do others think?

[tibor...@gmail.com]

I would vote for the former (i.e. task-movieX), because the stimuli are not identical across runs.

 

Vale,

Tibor

 

Auer, Tibor (Ph '99)

--
We are all colleagues working together to shape brain imaging for tomorrow, please be respectful, gracious, and patient with your fellow group members.
---
You received this message because you are subscribed to the Google Groups "bids-discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bids-discussi...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/4ef330e8-43c1-4c7b-a729-28aa37c1c00e%40googlegroups.com.

[Chris Markiewicz]

Treating each section as its own task seems reasonable to me, on the whole. I agree that the brain is likely doing things correlated more to the specifics of the scene than to the movie identity.

It might have some analysis effects later on, though. If you're expecting to combine betas from different tasks, my suspicion is that this would be a more complicated model to encode for most software. But I would say software should accommodate a sensible organization of the data, not the other way around.

Chris

--

[yarikoptic]

On Monday, March 2, 2020 at 10:05:07 AM UTC-5, tibor.auer wrote:

I would vote for the former (i.e. task-movieX), because the stimuli are not identical across runs.

counter-argument could be: in a well designed experiment, "stimuli" should rarely if ever be the same across runs ;)  That of cause also then goes into what is the "stimuli" -- a set of individual stimuli (e.g. pictures, etc) or include the order of their presentation (think - randomization).  Even with a collection -- in many experiments those wouldn't be reused across runs of the same task.

[yarikoptic]

On Monday, March 2, 2020 at 10:07:16 AM UTC-5, Chris Markiewicz wrote:

Treating each section as its own task seems reasonable to me, on the whole. I agree that the brain is likely doing things correlated more to the specifics of the scene than to the movie identity.

It might have some analysis effects later on, though. If you're expecting to combine betas from different tasks, my suspicion is that this would be a more complicated model to encode for most software. 

Indeed a factor to keep in mind! thanks

[tibor...@gmail.com]

True.

However, “run-X” would suggest me that the runs can be considered equivalent and the variance across them is not of interest (apart from habitation effect, perhaps). On the other hand, different parts of a movie have usually significantly different content (unless it is a documentary or 10-hours chill-out “movie” with natural ambient sound effects 😊), and if you want to store the different characteristics/features, then you can do it only if you consider them different tasks. This way, you can still try to model the overall effect of ‘theme’, but you also have to option to model differences. It is event more important, if you split the movie according to some hypotheses rather than just time.

 

Vale,

Tibor

 

Auer, Tibor (Ph '99)

 

--

We are all colleagues working together to shape brain imaging for tomorrow, please be respectful, gracious, and patient with your fellow group members.
---
You received this message because you are subscribed to the Google Groups "bids-discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bids-discussi...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/bids-discussion/9237b5e5-3e2a-4d8f-a5bf-b411b248df92%40googlegroups.com.

[Elizabeth DuPre]

Just wanted to ping in a related discussion from GitHub issues: https://github.com/bids-standard/bids-specification/issues/146

I believe the conclusion there was that we're likely to just have to live with some ambiguity, here. But it might be worth re-opening that issue to link these discussions !

Elizabeth