Dear all,
I have performed the steps suggested in previous posts for producing a FIR-modelled peristimulus plot: (1) I applied an FIR analysis in FEAT (specifying a 20 sec window with 20 bins, 2 EVs and 3 contrasts - [1 0], [0 1] and [-1 1]); (2) then I applied fslmerge to the corresponding subset of PEs (1-20 for EV1 and 21-40 for EV2); (3) finally, with fslmeants I extracted the PE timecourse from various ROI masks (all registered to the native functional images).
When I plot the PE value within each bin, averaged over all subjects, the peristimulus plot doesn't look quite right. I do see a nice smooth peak occurring around 4-5sec (and even a small undershoot around 8-9sec), but the signal becomes extremely saw-toothed from 9-10 sec (see attachments); specifically, it alternates between positive and negative values on each subsequent bin. The same is true for both conditions (EVs) and all ROIs...
If I take plot only the even bins the signal looks fine, but why should this be? My TR=2 sec, but I have jittered my ITIs (25%=2sec, 25%=4sec and 50%=6sec). Have I incorrectly performed the FIR modelling/PE extraction?
I very confused, so any help would be appreciated.
Dan.
|