 |
 |
Dear PPI (& DCM) Experts, I have a question regarding VOI definition for PPI (& DCM) and possible relation to results, -which I would like you to clarify for me. ;) So let me expand this a little, if there are errors, please correct me. 1.) What has been said clearly in the Literature (definition from 2ndLevel results & single subject deviations): In various papers (and the archives), like the "Ten simple rules for dynamic causal modeling" paper by Stephan et al., it was mentioned that VOIs can be defined from 2ndLevel Results (e.g. the start point for VOI definition) and anatomical constraints (maximum search Volume), GIVEN that the locations of activations between Single Subject may vary a little due to differences in functional architecture. All fine so far, I think. But, ... 2.) (My problem): what about VOIs that are defined from other Experiments/Studies or Literature in general? So if I do an experiment where I do find (on 2ndLevel) significant activations in area A and area B, but not in area C, can I still include(do a separate PPI for) area C if this area is described in the Literature "to be drived/modulated by" stimulations which are similar/equal to mine? Possibility a) "not alright" I have the feeling that this is not alright,because the area does not show modulation coinciding with my stimulus (not with my Psychological Variable), THUS any correlation/contribution that I find could be spurious, right? Could it be that I find a false positive contribution from area C to another area D, even though area C and area D do not respond to my experimental manipulation at all? e.g. could it be that I pick up intrinsic spontaneous correlation of area C with area D (i.e. I just picked up the parts of the correlation which do coincide with the timing of my Psychological Variable)? Possibility b) "not quite alright, but ..." It could be that the problem is only due to the significance threshold that I choose for my 2ndLevel Analysis, i.e. the area is actually modulated by the stimulus, however only weakly. In this case I could still get a significant correlation/contribution even though that is unlikely, isn't it? Can I then say this with confidence (i.e. is this publishable or will I be tourtured in the 10.000th layer of Hell reserved for bad/"overly eager" PhD Students)? How could I find out? (Or do I better shy away from this?) Kind regards, Rainer