Hey Rick,

One of my labmates is also using dmBlock with trials of different durations and is under the impression that AM2 is better to use than AM1. I've been using AM1 based on an earlier comment you made in this thread that AM1 was more likely a better fit for my design. Now I'm not sure which is better to use, especially based on this commentary in the 3dDeconvolve help page thta suggests that AM2 is the option one should probably use and that AM1 is for special cases and not really for fMRI tasks.

[-stim_times_AM2 k tname Rmodel]
Similar, but generates 2 response models: one with the mean
amplitude and one with the differences from the mean.
*** Please note that 'AM2' is the option you should probably use!
*** 'AM1' is for special cases, and normally should not be used
for FMRI task activation analyses!!
*** 'AM2' will give you the ability to detect voxels that activate
but do not change proportional to the amplitude factor, as well
as provide a direct measure of the proportionality of the
activation to changes in the input amplitude factors. 'AM1'
will do neither of these things.
*** Normally, 3dDeconvolve removes the mean of the auxiliary
parameter(s) from the modulated regressor(s). However, if you
set environment variable AFNI_3dDeconvolve_rawAM2 to YES, then
the mean will NOT be removed from the auxiliary parameter(s).
This ability is provided for users who want to center their
parameters using their own method.

Can you clarify when it's better to use AM1 and when it's better to use AM2??

Thanks!