Hello,
With your last screen capture, I understand better what happens and why your 10Hz peak disappears.
You extracted the scouts time series with the process “Extract > Scouts time series”, which does different things for different types of source models. In the surface/constrained case, there is one signal per source point and the sign is preserved in the output signal. In the volume/unconstrained case, this process takes the vectorial NORM of the three orientations: sqrt(x^2+y^2+z^2).
If you are familiar with these concepts, please read carefully the following sections in the tutorials. These are complicated notions, but necessary to understand what happens in your case.
http://neuroimage.usc.edu/brainstorm/Tutorials/SourceEstimation#Sign_of_constrained_maps
http://neuroimage.usc.edu/brainstorm/Tutorials/Scouts#Option:Absolute.2F_relative
In the volume/unconstrained case, the sign of the signals is lost (the signal for each orientation is squared), causing important changes in the spectrum of the signal. The rectification of the signals causes the oscillations at 10Hz around 0 to be transformed in 20Hz patterns. Something similar would happen if you observe the spectrum of a function abs(sin(t)) instead of sin(t).
We still don’t really know how to group these three orientations in the connectivity analysis. Computing connectivity measures for unconstrained source models is tricky. But you can already get better results by running the coherence process directly on the source files, and selecting the scouts you are interested in. Select “Apply scout function: BEFORE”, otherwise it will try to estimate a 30000x30000 connectivity matrix, which will crash your computer.
What the process will do is compute separately the connectivity measure for each orientation and each scout (3x3=9 values for each pair of scouts), then get the maximum measure for each pair (max of 9 values). This is arbitrary and unstable, but this is what is done for lack of a better solution at this time.
Working on the surface and with constrained orientations definitely make things easier to compute, display and undrestand.
The corresponding code is in bst_connectivity, line 567:
Cheers,
Francois