Brain regions coordinate across distance, so measuring how they connect is central to reading brain function. Most connectivity measures do this through phase, whether two areas' rhythms line up in time, but phase methods get fooled by volume conduction (one source bleeding into several electrodes) and they throw away amplitude information. This paper offers a different lens: an extension of the wavelet scattering transform to multiple channels that looks at how the amplitude envelopes of nearby electrodes move together.

On a standard motor-imagery dataset, the method found consistent amplitude coupling in a central-parietal cluster across every subject, and a second-order version showed that coupling is periodically gated by slow rhythms, a cross-frequency effect. Tellingly, two phase-based measures found almost nothing overlapping with it, suggesting amplitude coupling is a genuinely separate connectivity signal.

This is based on a cut-off abstract, so the paper will carry the fuller validation.