Dear all,

I have a question regarding the interpretation of ROI-to-ROI gPPI results in CONN.

I am using a gPPI analysis with an event-related design. My primary contrast compares two conditions (Condition A > Condition B). At the second level, I test both group differences (Patients > Healthy controls), as well as associations with continuous symptom measures (depression) disregarding the groups. To aid interpretation, I extracted the condition-specific first-level gPPI estimates and calculated group means for each condition.

For example, I obtained the following group effects:

ROI pair 1

There is a significant group difference (Patients > Healthy controls) for the A > B contrast --> β = -5 . If I understand correctly, this would mean connectivity decreases between this ROI pair in condition A compared to condition B. Am I understanding this correctly? 

The extracted condition-specific gPPI estimates from the first level were:

Patients:

• Condition A: -1.0


• Condition B: 0.05


• A-B contrast = -1,05

Healthy controls:

• Condition A: -0.4


• Condition B: -0.04


• A-B contrast = -0,36

I am not sure how to interpret this.

ROI pair 2

A second group effect (Patients > Healthy controls) for the A > B contrast --> β = 3

The extracted condition-specific estimates were:

Patients:

• Condition A: -0.9


• Condition B: -1.3


• A-B contrast = 0,4

Healthy controls:

• Condition A: 1.6


• Condition B: 1.8


• A-B contrast = -0,2

To summarise my questions:

1. How do I interpret the resulting beta from my gPPI analysis (e.g., from ROI pair 1, β = -5 . --> connectivity decreases between this ROI pair in condition A compared to condition B?)
   
   
   
   1. If this interpretation is correct, what exactly does "decreased connectivity" mean? Does it mean less connectivity, anticorrelation, or something else?
      
   
   
   
2. How do I interpret the condition-specific estimates?
   
   
   
   1. What does a negative/positive value mean?
      
   2. Which values should I compare: Between groups or between conditions? 
      
   3. What does this mean biologically? E.g., does a negative value indicate less connectivity between the ROI pair and a positive value indicate more connectivity? Are the ROIs less engaged or more engaged with each other?
      
   4. What do values around 0 mean?
      
   
   
   
3. For significant associations between a continuous symptom measure and the A > B contrast, what post-hoc analyses would you recommend to aid interpretation?
   

Thank you very much for your help.

Best wishes,