> The first regressor from the bucket output for condition1/condition2 should be
> the unmodulated amplitudes, however, we see very different results when
> comparing all-subjects 3dttest++ output from this data to data not using AM2.
> Basically, we see all the expected task activation when not using AM2 and
> almost none when using AM2. Any idea why we are seeing reduced activation
> despite the first regressor being 'unmodulated'?

The word "unmodulated" is probably inaccurate or even misleading to some extent. The first sub-brick for each condition actually is associated with the effect estimate associated with the average value of your modulation numbers. One possibility for the "strange" group results is that the modulation numbers are skewed (e..g. with outliers) so much for some subjects that the average modulation value may not be a good representative as the average effect estimate across the trials. If this is case, one option is to specify the modulation value at which you would like to obtain the effect estimate. Read the details about option -stim_times_AM2 in 3dDeconvolve -help that is dated 12 Jul 2012.

> Are the gltsym contrasts being calculated using only the 'unmodulated' regressors
> from condition1 and condition2? If not, do we need to include an index (ex: -gltsym
> 'SYM: condition1'[0]' -control' -glt_label 1 condition1_control) or something else?

Try 'SYM: condition1[0] -control'

Gang