I am currently attempting to develop a task for fMRI with psychopy. I have read a lot of the posts regarding relative vs. non-slip timing, but I still have what I hope might be a basic question.
In my design, the duration of routines is variable (we have jittered ISIs and ITIs, as well as one routine that terminates with a button press, choice RT). So implementing the non-slip timing is not feasible with the current design.
I have run the experiment several times and have noticed as others have that there is a little bit of lag in the trial duration (trials that theoretically should take 19 seconds take 19 + 20-40ms). This problem seems to increase with the amount of things that get written to the screen (which also makes sense).
My question is whether the onset timestamps are themselves reliable? Said another way, I don’t necessarily care whether my routine lasts 4 seconds or 4 + 10ms if the onset timestamp of 4 +10ms is able to be trusted. If that’s the case I can simply use the onset timestamps to build MRI regressors knowing that the task will take about a second longer per BOLD run.
Is it correct that although lagged the onset times reflect when the stimuli were written on the screen?
Thanks in advance!