afni.nimh.nih.gov

> I'm still not sure what would be a proper way to use the two of them for the group analysis.

The second coefficient is typically ignored at the population level in the SPM world. Alternatively, you could combine the two coefficients by weight-averaging the two basis functions, but it would be a little complicated and tedious. We demonstrated the approach a few years ago: [dx.doi.org]

Gang