> Stimuli are presented for 12 sec, but most responses occur before 8 sec. I'm
> having trouble deciding how long to model the response for, and how many
> tent functions to use.
How are you going to model the response? As an instantaneous event at the moment when each response occurs, or as a block from the beginning of stimulus presentation to the response? Either way, you may allow an extended BOLD response period of 10-12 seconds from the subject's response, and see if the model catches it properly.
Gang