Dear all,
I'm a bit lost in how to interpret the results of a PPI analysis (SPM2,
incl. 4 regressors: psychological, physiological, interaction,
constant). There are a couple of questions:
1) Can directionality be inferred or not?
The paper by Friston et al. 1997 and most SPM mailing list comments
seem to suggest "yes", but there are some comments which say "no" and
there are quite a few research papers which do not interpret
directionality - they usually refer to other methods such as SEM and DCM
if you want to infer on directionality. Friston et al. (1997) noted that
PPI is equivalent to effective connectivity only if all areas (probably
meaning all voxel?) are considered, and that PPI has only a "tenuous"
relationship if a single area is investigated. Thus, they termed it
"contribution". But in practical terms, what if we have e.g. 8 ROIs for
which we calculate the PPI among each other - is this then close to
effective connectivity or still close to the "tenuous relationship"?
2) What exactly does "contribution" mean?
It is the influence one area ("seed") exerts over the other ("target"),
and the influence depends on the psychological factor. Could this be
interpreted as that the information flow of the seed to the target is
increased, depending on psychological state? Or is this
over-interpreted?
3) Karl Friston made a comment on the list which I can't follow:
https://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=SPM;41a9f073.0903
quote: "This is a very good question and the answer is a PPI can be
interpreted in both ways. You are absolutely right that the
second-order nature of a PPI breaks the symmetry and allows one to
infer a directed influence; however, the direction
of this influence has to be specified a priori as the alternative
hypothesis. This is because an increase in the regression
slope of area A on area B can be interpreted as an increase in the
effective connectivity (under an instantaneous and
linear model of effective connectivity) from B to A. However, one can
transpose the regression (i.e., switch the axes)
and interpret it as a decrease in effective connectivity from A to B."
I tried to follow this thought and made up some plain data (2
variables, both increase, i.e. are correlated). When the axes are
switched, the beta value changes, but not the sign: when A is dependent
variable, unstandardized B is 1.933 (standardized .984), when B is
dependent variable, unstand B is .501 (stand .984). Thus, in both cases
I would predict an increase of connectivity: When A is dependent from B
to A, and when B is dependent from A to B. The two attached images
illustrate this (SPSS output and Excel graph). Or is Karl's argument
specific to the interaction term?
4) What, then, is the advantage of SEM/DCM?
Apologies for this very basic question, but in the papers I've read so
far I haven't seen the clear advantage of these methods over PPI. So far
I thought it is the possibility of infering about directionality, but if
this can be derived from a PPI analysis as well? Isn't it a big
advantage of PPI that it works fully based on the data and does not
require a priori specification of models? Of course I don't expect a
tutorial about connectivity methods here, but maybe one or two keypoints
or a hint to a good reference would be greatly appreciated.
Kind Regards &
Thanks a lot in advance,
Andre
--
______________________________
Dr. Andre J. Szameitat
Department Psychologie
Neuro-Cognitive Psychology
Ludwig-Maximilians Universität
Leopoldstrasse 13
80802 München, Germany
Tel. +49-(0)89-2180 6778
Fax. +49-(0)89-2180 4866
www.psy.uni-muenchen.de/ncp
Office: Martiusstr. 4, Room 6
______________________________
|