I am implementing a study with a mixed block/event-related design, and have a question
about the use of FIR basis functions for modelling the response to transient events.
My study has a TR of 2 sec, but stimuli are presented at intervals of 2.4 sec, with null
blocks providing jitter. I plan to use 8 time-bins to model the 16 sec after stimulus onset.
My question is whether the use of FIR is reasonable in this situation, where the onset of
stimuli are not synchronised to the start of each volume acquisition? Given that FIR is made
up of a basis set, and does not assume a shape for the HRF, I imagine it has the flexibility
to cope with this, but I just want to make sure. Perhaps I should set a different duration for
the bins? Are there any other considerations?