Dear Anne,

The following is speculation:

Perhaps there is some mismatch between i) your task specification in P.Name; and ii) your contrast specification P.Contrast(1...) which you have resolved by removing P.Weighted from the localiser and the subsequent contrasts. Since you are not using a localiser (P.Contrast=0), you could try re-running to with only P.Weighted removed (keep P.Weighted=0) for the contrasts P.Contrast(1).Weighted etc. This could support the mismatch in task/contrast specification. I agree that the suspicion that a session issue may be a factor in this mismatch is a possibility.

I don't know whether your fix violates any assumptions of gPPI.

Looking at P.Tasks you seem to specify 'invalid trials session 2' twice. Perhaps this is an issue. Another thing that you could possibly try.
P.Tasks = {'0' 'condE' 'condR' 'invalid trials session 2'...
'condM' 'condC' 'invalid trials session 2' };

If you are not already aware, then the following forum query seems to pertain to a similar issue (invalid task gPPI contrasts).

https://www.nitrc.org/forum/forum.php?th...

Is your study event-related or block in design?

Many thanks,

Mark