The SPM code has been changed such that the psychological vectors
(when the condition is present) is not demeaned, so you can now use
dummy codes. This is exactly what is done in the generalized form of
PPI (gPPI).
See lines 498-500 spm_peb_ppi.m that note this change.
As for your comments on visual stimulation, you are exactly correct.
However, in the case of a block design with states A and B, it makes
no difference. If you have A, B, and C (where C is fixation) and you
want to see where A is different than C, then you need to use the
approach stated in the gPPI toolbox (see below), that will model A and
B separately and allow inferences to be made about A or B compared to
the null state. Otherwise, B would get lumped in with C with only
dummy coding A. A paper on this method and addressing these concerns
is currently under review.
Information about gPPI toolbox can be found in previous emails OR at:
http://brainmap.wisc.edu/gPPI;
http://www.martinos.org/~mclaren/ftp/Utilities_DGM/; and
http://www.martinos.org/~mclaren/ftp/Utilities_DGM/PPPI
Best Regards, Donald McLaren
=================
D.G. McLaren, Ph.D.
Postdoctoral Research Fellow, GRECC, Bedford VA
Research Fellow, Department of Neurology, Massachusetts General
Hospital and Harvard Medical School
Office: (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.
On Wed, Apr 6, 2011 at 2:14 PM, Shabnam Hakimi <[log in to unmask]> wrote:
> Dear Karl and others,
> I have a question about the PPI algorithm that is outlined in Friston et al.
> ("Psychological and modulatory interactions in
> neuroimaging," NeuroImage, 1997) and is implemented by SPM and many others.
> In the paper, the authors state that (page 220) the interaction term is
> computed using a "mean corrected variable of presentation rates and ... a
> similarly corrected vector of dummy variables taking the value of 1 or -1."
> What is the logic of mean correcting the psychological variable at this
> point? While demeaning both the physiological and psychological variables
> before computing the interaction term would remove some correlation between
> the terms, and is not problematic with these particular dummy variables
> (i.e., because their mean is 0), it seems that it would also introduce a
> potential confound in interpretation if the dummy variable only takes values
> 0 and 1. Is it a requirement of this method that the dummy variables only
> take values +/-1 so that they sum to zero?
> For instance, if you have a visual stimulation task with a flashing
> checkerboard and you are interested in the psychophysiological interaction
> during the visual stimulation only, you would specify your indicator
> function such that it has value 1 when the checkerboard is on and value 0
> when it is not. The mean of this term is nonzero, and therefore introduces a
> bias that results in a model where the period of time where the checkerboard
> is off (and connectivity is not of interest) takes a slightly negative value
> (rather than 0). This analysis seems to answer a slightly different question
> than that posed above, since, in this case, we are looking for regions where
> there is some correlation during the task of interest, but no correlation
> during the rest of the task. The algorithm implemented in the paper seems to
> find regions where there is a positive correlation during the task and a
> (slight) negative correlation at other time points.
> I'd greatly appreciate any clarification on this point.
> Thanks,
> Shabnam
> --
> Shabnam Hakimi
> Graduate Student
> Rangel Neuroeconomics Laboratory
> California Institute of Technology
> Baxter Hall, Room 332E
> Pasadena, CA 91125
> phone: 626-395-5988 | fax: 626-405-9841
> e-mail: [log in to unmask]
>
|