Hi Thalia,
RSFgen only randomizes trial sequences, whereas 3dDeconvolve (via
-nodata) measures the normalized standard deviation. So neither
takes any cognitive processing into consideration.
Having an ISI that is a little longer than the return to baseline
is fine, but as you said, there is probably little to gain in an
ISI above 20 seconds.
Maybe Gang will have more insight to provide after he returns.
- rick