Hello,

I have problems with applying the default setting P.contrast={'Omnibus F-test for PPI analyses'}.

I want to analyze a simple emotional face matching paradigm with three task regressors (Instruction, Forms, Faces) and the classical six realignment parameters (which I want to control for). 

Analysis with P.contrast=0 runs flawlessly. Turning to the default setting, the following error message appears, no matter whether I create the effect of interest myself, or let gPPI build the contrast automatically:

Not enough input arguments.

Error in beta (line 19)
y = exp(betaln(z,w));
Error in timeseries_extract (line 370)
y=y-spm_FcUtil('Y0',SPM.xCon(xY.Ic),SPM.xX.xKXs,beta);
Error in PPPI (line 461)

Following the instructions from the manual, I do not see which input I miss to define.

Thanks in advance for your help,

Marcel

Hi Marcel,

Do you get the same error with the sample dataset? This would help determine if it is unique to the current dataset/model or the underlying code/software version.

Thanks,
Donald

Originally posted by Marcel Daamen:

Dear Donald,

I have changed the runOnSampleData.m accordingly, and get the same error message (even though it additionally generates a long series of additional warnings about invalid contrasts before, see attachment).

Since I had problems with running the sample data using the original package download, I tried some of the fixes that you posted inbetween (for spm_estimate_ppi, spm_spm_WB, spm_contrasts_ppi), I don't know whether this may cause the problem.

I run gPPI with SPM12 in Matlab 2019a on Windows 10.

Thank you for your help,

Marcel

Originally posted by Donald McLaren:

Dear Donald,

postscript: I solved the problem. Obviously, one of the updates I found in the forum did not fix, but cause problems. 

I rebuilt my PPPI installation from scratch with the current download, only (a) modifying the SPM version in PPPI.m, correcting the problem with the "contains" function with recent Matlab versions, and replacing the spm_contrasts_gPPI. Now the sample dataset runs without errors.

Anyway: Thank you for your efforts!

Best regards,

Marcel