Overview¶
Due to an upcoming workshop I've been contemplating the world of wonder and intrigue that is DCM a little more frequently than usual. The paper by Friston and colleagues (Friston, 2014), is one of a series ((Razi, 2014), (Friston, 2014)) of 'technical notes' describing the most recent extensions of the DCM technique/behemoth/cottageindustry, namely modelling resting state fMRI cross-spectra. The spectral (fMRI) DCM ('spDCM') approach is interesting in itself as it represents something of a middle-ground, or maybe a cross-over, between the two dominant approaches for studying functional and effective connectivity in fMRI data - namely time series correlations and (other versions of) DCM, respectively. Relatedly, the two broad classes of techniques for studying studying causal influence in neuroimaging - phenomenological approaches and physiological approaches (Griffiths, 2015) are in a certain sense synthesized in the spDCM formulation [1]. However the main reason for picking out this paper is that it adds a number of other ingredients into the mix, which variously reflect new, long-standing, and not-long-forgotten theoretical perspectives in the field. Which makes for a spicy read. The principal idea is of course modes, and that will be largely what I am focusing on in these notes.
Background¶
What is DCM?¶
to do...
Linear Systems Analysis¶
to do...
The Paper¶
Fluctuations¶
to do...
DCM model¶
to do...
Modes¶
to do...
Principal Coordinates Analysis¶
to do...
See also¶
Demo in SPM¶
to do...
Light Relief¶
Footnotes¶
References
^ Friston, Karl J; Kahan, Joshua; Razi, Adeel; Stephan, Klaas Enno; Sporns, Olaf. 2014. On nodes and modes in resting state fMRI.. URL
^ Razi, Adeel; Kahan, Joshua; Rees, Geraint; Friston, Karl J.. 2014. Construct validation of a DCM for resting state fMRI. URL
^ Friston, Karl J; Kahan, Joshua; Biswal, Bharat; Razi, Adeel. 2014. A DCM for resting state fMRI.. URL
^ Griffiths, John D.. 2015. Causal influence in neural systems: Reconciling mechanistic-reductionist and statistical perspectives. URL