Dear Ganne,
If you want to use a FIR model simply switch from "Canonical HRF" to "Finite Impulse Responses" during fMRI model specification / "Basis functions". To start a few bins before the actual onsets you have to specify shifted onsets. Make sure that for a proper FIR model, the durations have to be set to 0 s, the number of time bin regressors and the length of the time bins is defined via "Window length" (in your case, the pre-stimulus window plus the post-stimulus window, for a brief stimulus e.g. 5 + 25 s to also cover the undershoot) and "Order" (e.g. 15, which would result in time bins of 2 s in case of a TR of 2 s). Note that the onsets will be shifted to the nearest full time bin, which might or might not be problematic depending on your design. If the onsets of a particular condition are too close, then the FIR regressors will overlap in time (e.g. the 15th time bin for trial 1 and the 1st time bin of trial 2). Consider https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=spm;f525092b.1508 then to turn off the automatic orthogonalization within condition.
Hope this helps
Helmut
|