So is it recommended to use mean before? To my understanding that might lead to some information getting lost.
I'm sorry, we don't have generic recommendations for the moment.
Connectivity analysis is a complicated topic, especially at source level.
For the moment, we offer the tools for users to play with without strict guidelines.
Another question popped up: if the matrix with MEAN / AFTER has values that are not 1 in the diagonal, why is the visual representation of the matrix with a diagonal = 1 as seen in the screenshot above?
It is not a 1 that you see in the diagonal, it's a ZERO, or more exactly an absence of data, so that it doesn't force scaling the colormap to 1 if the other values are much lower.
We are currently working on finalizing our connectivity tutorial, maybe some of these details will be adjusted as we go.
Would it be possible to just replace the diagonal with 1 in the exported file?
No, not if this is not consistent with the way the other values are computed.
The fact that you don't have a zero is informative of how much you lose by using this "after" option.