Abstract
Psychophysiological interaction (PPI) and beta series correlations (BSC) are two commonly used methods for studying task modulated connectivity on functional MRI (fMRI) data. So far there are no comprehensive tutorials to explain these two methods, and the relationships between these two have not been established. In the current paper, we explained in detail what the two methods measure, and how these two methods are related. We elucidated why the PPI approach always measures connectivity differences between conditions. This is in contrast with the BSC approach, which can measure the absolute connectivity in a specific task condition. By explaining the deconvolution process of the observed blood-oxygen-level dependent (BOLD) signals from fMRI with hemodynamic response function, we explicated that PPI can measure the differences of correlations of trial-by-trial variability in different conditions. Therefore, when comparing connectivity between different conditions, PPI and BSC methods could in principle generate similar results. In addition, we established that when modeling multiple conditions in PPI analysis, PPI models calculated from direct contrast between conditions could generate identical results as contrasting separate PPI terms coding each of the conditions (a.k.a. “generalized” PPI) if the models were defined correctly. We also reported empirical PPI and BSC analyses on fMRI data of a stop signal task to support our points.