Hi Jinny,
Note that we recommend using the seed time series as a regressor
of no interest, so that PPI is not confused with simple correlation.
If you followed the steps on Gangs PPI page and used such a regressor,
you would probably not see much PPI activation in the seed region,
especially if the region is small. The voxels in your seed region
should have the vast amount of their variance explained by the seed
time series, not the interaction regressor.
- rick