Dear Russ,
> my guess is that the dt is there (within spm_hrf.m) to allow more
> accurate subtraction of the two gammas. when you create a model within
spm,
> it is oversampled by RT/fMRI_T in order to allow more accurate
recombination
> of the HRFs for each event (particularly when events happen at times other
> than every TR)....
This is absolutely right. To convolve the oversampled stimulus function you
need an oversampled hrf. This is simply spm_hrf(dt).
With very best wishes - Karl
|