Hi Alfonso
 
Great and thanks. Now, I think I know how to set up everything.
 
However, for PPI (is this gPPI by the way?), I do have also a control condition. This means I could add onsets/durations for them too (pre_control and post_control), right?
 
If I use the design matrix which you suggest, I can look to PPI task effect > "baseline" (which is the whole scan -> 0 – inf). This should give some strong results.
 
However, then I am not controlling PPI results for any confounding effect (that is why we had an fMRI control condition).
 
As I wrote, I am interested in both background and task connectivity
 
Assuming to your recent mail (see below), background connectivity can be achieved by moving task conditions to confounds. But then you wrote later on "in addition to the same 'rest' effects as you were obtaining before". I don't understand this because in one case task effects are regressed out (first analysis), whereas in the second analysis "task" is not a confound. How can this lead to the same background connectivity result for analysis 1 and 2?
 
What is your opinion on the Fair approach: Can we argue, it regresses out task effects but still reflects "general task engagement", since we model from 0-inf (which includes all task periods)?
 
In any case, I do have to run 2 CONN analysis, I assume, one to extract background connectivity (analysis 1) and one to extract task connectivity (analysis 2), right?
 
"Dear Pravesh,
 
analysis 1
Yes, the Fair et al. method is appropriate both for event-related and block-designs. Typically you simply define two or more conditions (e.g. in your case one 'rest' condition with 0/inf onsets durations, and then 'taskA' and 'taskB' conditions with your task block onsets/durations), and then select your 'taskA' and 'taskB' conditions and click on 'condition tools -> move selected conditions to first-level covariates list (Fair et al.)'. That will leave only 'rest' as your single condition but it will create a new 'taskA' and 'taskB' first-level covariates which will be used to remove the task effects during the Denoising step before computing connectivity measures (simply make sure that in the Denoising tab the 'effect of taskA' and 'effect of taskB' entries are listed as part of the 'confounds' list)
 
analysis 2
If, on the other hand, you are interested in potential condition-specific connectivity effects, then simply leave the original three conditions (rest, taskA, and taskB) in the Setup.conditions list (do not move the last two to the first-level covariates list), and still have all of those 'effect of task*' effects included in the 'confounds' list during Denoising. In this case, in addition to the same 'rest' effects as you were obtaining before, you will also obtain task-specific ('taskA' and 'taskB') connectivity estimates for your second-level analyses"
 
Best and thanks
Lars