Continuous neural activity is not wasteful background noise—it is the physical basis of being alive, conscious, and ready to act. Some of the observe brain activity’s function can be isolated with controlled experiments and carefully designed tasks, but it is well-known that the bulk of this activity is not straightforwardly related to these tasks. This activity, at the whole-brain scale, is slow and has substantial spatiotemporal structure. On this grounds, how much this activity actually reflects cognition is debated and unclear. In this talk, I will propose an interpretative axis ranging from physical epiphenomenon to mental-content-realising. I will discuss, in light of what we know, where in this axis whole-brain activity may possibly lie, as we measure it with fMRI and magnetoencephalography (MEG), and as we characterise it with existing analytical methods. Finally, I will propose that when using analysis methods that preserve the phase of the signal, MEG patterns are likely to reflect context and variables that constrain the computations behind cognition; although we cannot conclude these to have any representational role.