The calculated GC value has a negative number. Is it normal?

[neurochenhe]

Dear developer, thank you for such a great toolkit for brain region connectivity calculation.

I am using pipeline_connectivity.m and calculating the GC in the example dataset, I found that the value stored in the EEG.roi.GC variable after the calculation has a negative value, is this an anomaly, and should the Granger causality value be greater than 0, and should not have a negative value.

When plotting the schematic via pop_roi_connectplot(EEG, 'measure', 'GC', 'plotmatrix', 'on'); you can see that the colorbar starts with a negative value. Thank you for your answer!

[nguyen-td]

Hey there,

Thanks for using our toolbox!

@stefanhaufe Please correct me if I am wrong. GC is a directional connectivity metric, meaning that it indicates a flow from region A to B. A negative value then indicates a flow from B to A.

[stefanhaufe]

It is correct that GC is a non-negative quantity. However, we are not looking at original GC but so-called net GC, which is the difference GC_i->j - GC_j->i . This quantity is always antisymmetric, that is netGC_i->j = - netGC_j->i and this is what is shown in the plot.

netGC is already much more robust wrt. source mixing than original GC. To be fully robust, we recommend using time reversed GC (TRGC), which is the difference netGC_i->j - netGCR_i->j, where netGCR_i->j is netGC estimated on time reversed signals.