First level model - model event duration for event related design on Nilearn

Software Support
1

Summary of what happened:

Dear NeuroStars Community, I was wondering which duration you would put in the events file for the first level model when you have event related design. I heard 0 or 1 is good, however, when I model with 0 or 1 I have different results for some subjects…
Targets are presented for 1 second (so I thought about putting 1), however, it disappears as soon as the participant answers (so for some they see the stimuli for 500ms, some 700ms etc…
Should I put in duration the RT to each trial per participant for example?

Thank you!

2

Yes you should probably put the RT. Remember that the amplitude of the BOLD response will be proportional to this duration: it means that when the RT is doubles, the assumed BOLD reponse is modeled.
One of the key questions is whether you think that the BOLD amplitude should be the same for all trials or not. Having RT as duration will make it different across trials.
Does it makes sense ?
Best,
Bertrand