hi

I am running a gPPI analysis with three conditions (A,B and C).  When I look at the design matrix for the PPI analysis, the task regressor for condition A has been replaced by the sum of the regressors for A and C.  I am wondering why this is done--I had been doing the analysis in AFNI using the instructions here:

http://afni.nimh.nih.gov/sscc/gangc/CD-C...

and wanted to make sure that I got the same results when using your toolbox, since the procedure in AFNI is a bit byzantine.  However, in those instructions the task regressors are inserted in the PPI as is, whereas that does not appear to be the case in the toolbox.  Additionally, in the 2012 NI paper where you introduced gPPI it seems that the task regressors are also included in the model as is for gPPI.  So I really feel like I am missing something.

Interestingly, in the SPM.mat for the PPI analysis the onsets and durations for the conditions are all as specified for the basic GLM model (i.e., those in SPM.SessC(1).U) but the regressor for the condition in question (in SPM.xX.X) is not what I would have expected.

Any clarification you can give me would be greatly appreciated.

My param structure is:

clear P
P = struct();
P.subject = sprintf('s%s',subs{i});
P.directory = sprintf('~/Work/fMRI/transmem/CONN/s%s/',subs{i});
P.VOI = sprintf('/Users/keidel/Work/fMRI/transmem/CONN/s%s/LH_STS.nii',subs{i});
P.region = 'LH_STS';
P.analysis = 'psy';
P.method = 'cond';
P.CompContrasts = 1;
P.ConcatR = 1;
P.Estimate = 1;
P.equalroi = 0;
P.FLmask = 1;


thanks

James

PS. I was getting errors in running PPPI in a directory where the first-level SPM.mat had been specified but not estimated in SPM...does this make sense and if so perhaps it is worth mentioning it in the manual if has not been already?

James,

The first issue you are having is only applicable when you set concatR=1.
When you set concatR=0, this won't happen. I'm working on changing the
concatenation to not combine the tasks by accident as it did in your case.
It's matter of setting the correct regular expressions in my matlab code
for spm_estimate_ppi.

The second issue occurs whenever P.contrast is not defined OR when it does
not 0. I'll work on updating the manual and also updating the error dialog
to be more informative if the error is due to the first level model having
not been estimated.

-Donald