Dear all,

I understand this question has been brought up numerous times, but I am still finding it difficult to interpret gPPI results. I had initially conducted my analyses using GLM, but a reviewer on my grant commented that GLM was outdated and suggested using gPPI instead.

My study was a simple block design with TR=2s and task conditions: written word processing (combining 3 judgment tasks: rhyming, syllables, and identity), random pattern (i.e., a visual control task) and fixation periods. I am interested in examining the ROI-ROI functional connectivity (FC) between various ROIs during written word processing.

For the weighted GLM w/ bivariate correlations, I get results I'd expect based on the literature and my own separate activation analyses—strong positive FC between reading regions (visual word form area (VWFA), occipital word form area (OWFA), temporal parietal cortex (TPC), and inferior frontal gyrus (IFG) and negative FC between my orthographic ROIs (i.e., VWFA and OWFA) and auditory speech processing areas (auditory word form area (AWFA) and primary auditory cortex (A1)). The connectome ring results are attached.

However, when I switch to gPPI w/ bivariate regression—with the 3 above tasks included in my model—my results are harder to interpret. Even without multiple comparisons correction, I get no significant results with the gPPI when I select to examine the word processing condition (just as I had done for the weighted-GLM analysis above). If I choose to look at the contrast of word processing > fixation, I get a single negative FC result between the VWFA and TPC—this makes no sense to me.

In summary, I'm not sure if there's something wrong with how I'm setting up the gPPI analysis, or if I'm just not understanding how it should work. I would be grateful for any insights and suggestions.

Best,
Lillian