Hi Steve,
One basic point is that you should not be modeling the baseline.
Is that the purpose of your ITI regressor? If so, just remove it.
A polynomial baseline will be applied per run. That applies to
both commands, it seems.
---
Do your stimuli actually last 2.5 seconds, or is that just the
duration of your TR? If they do last that long, you could ponder
using 'BLOCK(2.5,1)' for the basis functions with -stim_times.
---
To view how well you are modeling the data, add -fitts options
to the 3dDeconvolve commands (e.g. "-fitts fitts_dset"). After
running it, view the data in afni:
underlay: all runs (input to 3dDeconvolve)
open graph window
in graph window: Opt -> Tran 1D -> Dataset #N
in Dataset #N plugin: choose the fitts_dset and apply
Maybe then select some statistic in the overlay and go to voxels
with high statistics. See how well the model fits the data.
- rick
