Psychophysiological interaction analysis vs. Multivariate autoregressive model

Hi, I’m looking up literature on pros and cons (and research questions that can be answered) of using psychophysiological interaction analysis vs. multivariate autoregressive modeling. Specifically, I am thinking to justify my use of PPI over MAR, but I’m not very familiar with the latter approach. In case this is helpful – I am hoping to look at change of strength in functional connectivity between the orbitofrontal cortex and ventral striatum during anticipating reward vs. anticipating no reward.

I’d appreciate any input on this!


I don’t know much about MAR, but PPI (or generalized PPI/gPPI as the field is moving towards) is well suited for looking at changes in connectivity between two tasks, especially if the tasks are events based (as opposed to a simple block based design which a weighted GLM can handle fine). I found this video to be helpful, and it explains some things that will help you see a better effect size, such as making sure the seed region you choose has different expected activation strengths for the two tasks. gPPI may also be strengthened if you have a baseline / rest condition that you can model along with your two tasks of interest.