Greetings All,
We're interested in estimating condition-specific IRF duration in an event-related paradigm. We've been running 3dTstat with the -duration option on IRFs rendered from 3dDeconvolve with the -iresp option. We're wondering if we might estimate something similar with the new duration modulation feature in 3dDeconvolve. Perhaps something like...
3dDeconvolve -input 3dplustime+orig -num_stimts 2 \
-stim_times_AM 1 cond1.1D 'dmBLOCK' -stim_label 1 'cond1' \
-stim_times_AM 2 cond2.1D 'dmBLOCK' -stim_label 2 'cond2' \
-gltsym 'SYM: +cond1 -cond2' -glt_label 1 'cond1 - cond2' \
cond1.1D = 4:12 18:12 30:12 ...
cond2.1D = 12:12 40:12 56:12 ...
I've used a constant duration value of 12 here (approx the duration of the HRF). Seems that the resulting duration betas would be some linear deformation of this "basis value" that would approximate IRF duration. Close at all? Your compassionately authoritative guidance will not go unheeded.
Best,
Paul