We have two types of stimuli (A and B) in our event-related design. They are each 4 seconds long (TR 2sec). Due to some Psyscope timing problems, by the end of the 6 minute task, trials are beginning in the middle of a TR.

We are using stim_times to specify the onset times for stimuli A and B. Subjects generally performed very well, but we want to look at only correct trials.
Is it better to:

1.) In our .1D files for stimuli A and B only list the onset times of trials answered correctly? This would result in incorrect performance being completely absent from the model.

2.) Have performance be it's own stimulus type and list the onset times of incorrect performance (aka Perf)?
-stim_times 1 TaskA.1D GAM -stim_label 1 TaskA -iresp 1 TaskA_irf
-stim_times 2 TaskB.1D GAM -stim_label 2 TaskB -iresp 2 TaskB_irf
-stim_times 3 Perf.1D GAM -stim_label 3 Perf
(Probably need a GOFORIT 3).

3.) Use stim_timesAM2 for performance (on the message boards people have done this for reaction time, etc.)

4.) Put it in the baseline model like motion - but I don't think I can do -stim_base with -stim_times yet.

I saw the bootcamp slide indicating use of a stim_file (with 0's for correct trials) to model performance, but because our Psyscope trials slowly drifted, earlier trials begin at the onset of a TR whereas later trials begin halfway into a TR.

Thank you for your help,
Erin Moore Parks