The stimulus at t=13 stays on the screen for 0 to 5s,
so that is indeed a variable duration stimulus, it is
not just the ITIs. A 0 to 5 second stimulus duration
will have an enormous effect on the shapes/magnitudes
of the responses.

Why not model the 0s stimulus with GAM or BLOCK(1.5,1),
say, and model the 13s stimulus with 16 seconds of
TENTs, say? There seems little reason to group the
2 events into one stimulus class. They are drastically
different.

Going forward with trying to compute an average response
(instead of IM), this seems like a good case for using
duration modulation. The 0s event could be modeled as
GAM or BLOCK(1,5.1) as noted above, and the 13s stimulus
could be modeled as dmUBLOCK, with durations between 0
and 5 seconds.

- rick