Hi Qing,

Just to clarify, you have one event at time 0s,
another at time 13s, a variable ITI, and then
the next event trial. Is that right?

If so, keep in mind that the variable ITI is
helpful in disentangling the signals, but from
a basis-function point of view, it is irrelevant.
That should have no impact on whether you
use BLOCK or GAM or pairs of them vs
TENT (except that without enough ITI time,
TENT is not possible, as you are seeing).

Do not say the trial length is varying between
22s and 28s if that variation is merely during
the ITIs. What matters is the actual stimulus
time, which I do no think you have mentioned.
The stimuli at time 0s and 13s are presented
for some duration, what is it?

- rick