Hi, I am interested in seeing how the interaction between EEG and BOLD is modulated by task.  Would this be possible to do with the gPPI toolbox?  Could you please advise me on what part of the code I would need to modify to use an EEG timecourse instead of the BOLD timecourse from a specified ROI.
Erika

> What do you mean by the interaction between EEG and BOLD is modulated by
> task. The relationship between EEG and BOLD should be constant as EEG
> should be measuring the underling BOLD signal? Do you mean you want to see
> if task modulates the EEG connectivity effects on BOLD? I'd have to look at
> the code again, but I think the first step is a deconvolution, which
> wouldn't be needed in the case of EEG. I think there is an option that I am
> working on that will be able to take a timeseries directly though. I'll
> post the update when the option is available.

What I want to know is if the correlation between EEG power and BOLD varies by condition (e.g. From Hanslmayr et. al., 2011 "Trial-by-trial correlations between EEG and BOLD activity showed that beta power correlated negatively with left inferior prefrontal cortex activity. This correlation was more pronounced for items that could later be successfully recalled compared to items later forgotten.").  As far as I know, PPI has not been done to test this before, but it seems reasonable to look at the interaction between the EEG timecourse and the BOLD signal for each condition and then to compare conditions.  Your toolbox seems to be well suited for such an analysis if there is a way to put in the EEG timecourse.

>
Erika,
There isn't currently a way to provide a non-BOLD timeseries to gPPI. Doing
so would require a number of changes in multiple places of the code, which
is not high on the priority list. However, it shouldn't be too hard to
construct the PPI for what you want to do.
STEP 1: Load in the EEG timeseries
STEP 2: Create a vector that labels the task at each EEG value
STEP 3: Convert these to dummy coded variables
STEP 4: Multiple the dummy coded variables by the EEG signal (this gives
you the task*EEG interactions)
STEP 5: Convolve the interaction term with the HRF, Also convolve the EEG
series
STEP 6: Downsample to the BOLD TR
STEP 7: Create a fMRI task model and add the outputs from Step 6 to the
fMRI model (this will now be the PPI model).
STEP 8: Estimate the model, create contrasts, and do group analyses.

Alternatively, you could create a Parametric Modulator and test whether the
PM is different by condition.