Hi Donald,

    I've been using gPPI toolbox, and everything goes well except the following warning message:

Warning: Divide by zero.
> In createVec at 757
In spm_contrasts_PPI at 126
In PPPI at 796
In g_PPI at 37
Valid Contrast


What may cause this warning message? The contrast images are created despite the warning message.

Best,
Chunliang

Chunliang,

What is the code in your version of createVec.m at 757?
Is it:
try
trialminimum=trialcountminper;
catch
trialminimum=trialcountmin./sum(tce>0); %Only count
event-related effects
end


Is this a block design or an event-related design?

If you are using a block design, this shouldn't be an issue.

Hi Donald,

     Thanks for your reply. Yes, createVec.m at 757 included the code that you pasted.

     It is an event-related design; however, duration of each event in the GLM was set to a non-zero value (i.e., boxcar regressor), and response time was added as a parametric regressor for each trial. The followings are part of scripts to set up tasks and contrasts.

P.Tasks={'0','condA_run1','condB_run1','condA_run2','condB_run2'};


P.Contrasts(1).left={'condA_run1' 'condA_run2'};
P.Contrasts(1).right={'none'};
P.Contrasts(1).STAT='T';
P.Contrasts(1).Weighted=0;
P.Contrasts(1).MinEvents=1;
P.Contrasts(1).name='condA';



Best,
Chunliang

Because you specified Weighted as 0, and your events are non-zero duration,
then the program treats the data as a block design. This just means that
when it averages across runs, each run gets an equal weighting - rather
than weighting based on the number of trials.

I will fix this piece of code to avoid the warning message in the future.

Also, RT as a PM doesn't make much sense because RT should change the
duration of the neural activity rather than just changing the amplitude.

In the future, you don't need to specify the run # in the condition name as
each run # is prepended to the task name in the GLM.

I see. Thanks again~~~

Best,
 Chunliang