Dear Donald:
> -----Original Message-----
> From: [log in to unmask]
> [mailto:[log in to unmask]] On Behalf Of Donald MCLAREN
> Sent: Tuesday, October 09, 2007 5:00 PM
> To: [log in to unmask]
> Cc: Sterling Johnson; Michele Ries; Guofan Xu; Shelly Fitzgerald
> Subject: PPI Analysis Questions
>
> Dear Dr. Gitelman,
>
> I have several questions about how PPI analysis are currently
> conducted and was hoping you could shed some light on the issues:
>
> First, I'd like to say I really like the general approach of
> the estimation of neural activity and the the reconvolution
> to get the interaction term (ppi) for psychophysiological
> interactions. However, it is unclear to me why there is one
> psychophysiological term, rather than one for each task.
>
> For example, one could construct a design matrix to test the
> interaction of task and region by having separate
> psychological and psychophysiological vectors for each task.
> The interaction would then be tested when you specify the
> T-contrast; which would be 1 and -1 over the
> psychophysiological vectors for a standard 2 condition
> experiment. Intuitively, this makes more sense since it would
> be more similar to the standard first level analysis. I don't
> think I've ever seen a design matrix that collapses two tasks
> into a single column, yet it is done in PPI analyses.
I think the issue is the level at which interactions are taking place. The
convolution of neural data with the HRF is what makes the difference.
The PPI machinery was designed to deconvolve the BOLD data so that the
interaction can be formed between the psychological factor(s) and the
"neural" data or as close as we can get to the neural data. By splitting up
the factors you are forming the correct interactions between neural and
individual psychological factors but then asking for the interactions
between these interactions at the BOLD level. I don't think this is correct.
The current ppi machinery gives you (e.g., with factors A and B)
PPI.ppi = conv((A-B)*xn,HRF) where A and B are your psychological factors
and xn your deconvolved BOLD data from the source region. What you are
asking for is
conv(A*xn,HRF) x conv(B*xn,HRF) but this is not the same as
conv((A-B)*xn,HRF). See the help section of spm_peb_ppi.
> To go one step further, in the case of three conditions the
> traditional psychophysiological vector would have to have a
> parametric modulation for all three conditions. I believe,
> although could be wrong, that adding a parametric modulation
> would be interpreted as looking for regions that show that
> particular difference pattern between the tasks since you are
> forcing the GLM to look for voxels that exhibit a specific pattern.
I'm not sure I understand what you want to do here. You could parametrically
modulate a factor and use that as the interacting term. I think the PPI
machinery deals with this.
> Does splitting the psychological and psychophysiological
> vectors and using contrasts at the statistical level change
> the interpretation of the data?
Yes. see above.
> Also, in my tests of the two
> methods, the statistical significant regions that show task
> dependent connectivity are different. Some times you get more
> regions by splitting up the vectors; however, you also lose a
> few areas. How would you interpret these findings? If
> necessary, I can send jpegs of the results.
they will be different because they are not equivalent.
Darren
>
> Any thoughts on this topic would be appreciated. Thank you
> very much for your time.
>
> --
> Best Regards, Donald McLaren
> =====================
> D.G. McLaren
> University of Wisconsin - Madison
> Neuroscience Training Program
> Tel: (773) 406 2464
> =====================
> This e-mail contains CONFIDENTIAL INFORMATION which may
> contain PROTECTED HEALTHCARE INFORMATION and may also be
> LEGALLY PRIVILEGED and which is intended only for the use of
> the individual or entity named above. If the reader of the
> e-mail is not the intended recipient or the employee or agent
> responsible for delivering it to the intended recipient, you
> are hereby notified that you are in possession of
> confidential and privileged information. Any unauthorized
> use, disclosure, copying or the taking of any action in
> reliance on the contents of this information is strictly
> prohibited and may be unlawful. If you have received this
> e-mail unintentionally, please immediately notify the sender
> via telephone at (773) 406 2464 or email.
>
|