There are a number of methodological issues and open questions concerning connectivity estimation which need to be addressed. For instance, the question of whether increases in connectivity might be merely driven by increases in the amplitude of oscillatory activity, or whether the connectivity itself might induce increases in local amplitudes is still debated (Daffertshofer and van Wijk, 2011; Moon et al., 2015; Tewarie et al., 2019). Another methodological issue is that all connectivity methods, including those which have been designed to specifically handle artificial connectivity detections due to zero-lag correlations, are not safe from spurious connections caused by source leakage in the proximity of true interactions, i.e., the so-called “ghost interactions” (Palva et al., 2018). Finally, the relationship between some coupling measures and the underlying physiological mechanisms is not well-understood. For instance, cross-frequency coupling measures might be influenced by features in the signal that do not relate to interactions, such as the non-sinusoidal waveform of brain oscillations which generate the coupling between harmonically related frequencies (Aru et al., 2015; Lozano-Soldevilla et al., 2016; Deco et al., 2017).
