Dear RG Community,

I am performing an SPM analysis in which I've got 8 conditions plus the baseline in an event-related experiment. Each condition has 4 trials, totalling 32 trials per run. There are 4 runs for each participant. In each run, participants have to respond to stimuli and there is a reaction time which I would like to factor out, without cancelling processes of interest. I'm using parametric modulators to factor out the reaction time (RT).

Since I've got multiple conditions, I'm creating mat files with the multiple conditions as explained here: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=spm;d2ec8a0d.0711 I'm also including the structure pmod to specify the parametric modulators. I'm defining the parametric modulators as mean(RT) -RT, as explained here: https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=SPM;aa781174.2003

However, the parametric modulators are demeaned for each condition, and not each run, as I can see in SPM.xX.X. As I know that RT is longer for one condition than another, it doesn't make a lot of sense to have them demeaned (I think they should be demeaned for the run, instead). Also, as I have an uneven number of trials in a condition (I'm removing wrong answers), I'm defining one beta for each trial then collapsing them when I create contrasts. However, since parametric modulators are specified for each beta, the demeaning making the single parametric modulator to become zero and down the road I'm not factoring out RT at all.

As an idea of workaround, should I convolve my RT-based parametric modulator with the HRF (for each run) and remove its influence as a regressor of no interest?

Could anybody help with how to specify parametric modulators for multiple conditions without having them demeaned?

Thanks in advance!