Those are very long blocks, which might make your
results noisy, especially as the polort degree gets
bigger.

Ignoring that, the basic point here is that you
should use duration modulation, probably via the
dmBLOCK(1) basis function.

1. The stimulus timing files would be sequences of
event:duration elements, where one row is a run, e.g.
41.028:138.801 292.071:82.955
... other runs ...

2. The stimulus types would be AM1 (either via the GUI
or via afni_proc.py -regress_stim_types).

3. The basis function might be 'dmBLOCK(1)'. There is
no 'choose' option for that in the GUI, but you can
still type it in the 'init basis funcs' box and hit
<enter>.


So the main work is to generate timing files that are
in the form START:DURATION, rather than START:END as
you are currently using.

- rick