Good question. If you mean correct the inputs, then A will be the average connectivity of your two experimental conditions, and if you don't, it'll be the effect of faces. There are other factors which will cause differences - for example, the C parameters are likely to be smaller when mean-centring, leading to a lower complexity cost and so higher free energy.
From: SPM (Statistical Parametric Mapping) <[log in to unmask]> On Behalf Of Seda Sacu
Sent: 07 January 2019 05:38
To: [log in to unmask]
Subject: [SPM] Single input DCM
Dear DCM Experts,
I have a block design task that consists of two conditions: faces and shapes as control condition/baseline. Since there is no modulatory input, I would like to model A and C matrices only using faces condition as driving input. As far as I understood, A matrix represents effective connectivity across the time/experiment. I am interested in effective connectivity during face processing rather than all experiment. I saw in a previous discussion that I can interpret A matrix as an average of conditions (inputs) if I use mean-centring option during model specification :
For DCM analysis I set a GLM analysis with one column modelling faces(the other condition became baseline). If I use mean-centring option, will my A matrix be average connectivity across faces or average connectivity across faces relative to baseline(shapes). Secondly, I haven't come across any publication that only A and C matrices are used. If someone knows any publication as such, please let me know. I will be very happy to read it.