MEG sensor signals arise from the mixture of the activity of several neuronal pools. In fact, each MEG sensor measures, with different weights, the signal generated by all neuronal pools active at a given time. This results in the overlap, at sensor level, of the fields generated by each pool. Even if one would unrealistically assume that only one pool is active, and thus no actual brain connectivity is present, a fake connectivity profile between the sensors would be detected as the result of all sensors measuring the activity of the one brain source. Thus, assessing true brain connectivity directly from MEG signals at sensor level is challenging (Schoffelen and Gross, 2009).
