How should I set the jitter in the fMRI experiment (event-related design)? What is the shortest acceptable mean ISI to ensure as many trials as possible (or as little total time as possible)?
I don’t know much about jitters since I have not designed any fMRI experiments as yet but maybe this will help guide you with setting jitters (you need E-Prime for this):
I don’t know if you came across this website already.
My best recommendations for material on how to design fMRI experiments is this link:
But the short answer is, yes, if you’re using an event-related design, then you should introduce some jitter or randomness into your design.
As to exactly how you do this, there are no hard-and-fast rules really. I tend to use a mean ITI of around 5-6s, if the stimuli/trials are relatively short (i.e. <5s). If the stimuli/trials are longer than you may want to extend the ITI somewhat as it’s the SOA (stimulus onset asynchrony) that’s really more important, rather than the ITI.
As for the distribution of ITIs, that’s even more variable. Some people use a flat distribution (i.e. equal numbers of ITIs of each duration), others use more sophisticated things like M-sequences, and many other approaches. I tend to rely on this old paper by Hagberg: Improved Detection of Event-Related Functional MRI Signals Using Probability Functions - ScienceDirect
In that they demonstrate that a Poisson-like distribution is perhaps the best option (i.e. a skewed distribution, with a few short ITIs of 2-3s, a large peak around the mean, and a longer right-hand tail with longer ITIs of up to 10-12s). I tend to go with that as I’ve always found it works well, but there are no right answers really…
Hope that’s helpful!
Another good source on experimental design for fMRI: http://www.mrc-cbu.cam.ac.uk/personal/rik.henson/personal/Henson_SPM_06_preprint.pdf