Dear Alfonso, 

I would like to use parametric analysis to estimate the association between task-related connectivity (PPI) for a specific condition and a vector of parametric modulators (pmods) with a value per block. My dataset is comprised of multiple runs (10) and only a few blocks of that condition per run (4); therefore, I would like to estimate the association considering all runs (40 datapoints per subject). However, assuming that CONN works in the same way as in SPM, the pmods would be demeaned for each run, i.e., the pmods [1 2 3 4] for one run would be same as the pmods [5 6 7 8] in another run. 

I thought about first concatenating the runs then specifying the runs as covariates to not demean the pmods for each run, but that wouldn't be an efficient solution in CONN. Would there be a strategy to use the pmods not averaged for each run?

If I may also ask, I found in another thread that, in order to estimate the association between PPI and pmods, one should select "other temporal-modulation effects" in Analyses (1st-level) --> Analysis type. But it's not clear to me what should be my interaction factor. Is it the pmod time-series that I previously defined? (yes, I created a txt file with the pmod time-series in Setup --> Covariates (1st-level) --> Covariates) Then, where do I specify the condition of interest? (it's clear to me that, for categorical analysis, one selects the condition in a pop-up window after choosing "task-modulation effects (gPPI)" in Analysis type).

Please let me know if something is not clear in the question. 

Thank you,
Gustavo