to detect alpha-gamma PAC within visual area V1, we used a broad filter bandwidth, defined as ±0.4 times the amplitude center-frequency. Consequently, the alpha-gamma comodulograms will be unable to differentiate between adjacent gamma sub-bands, which have been proposed to fulfill differing neurocognitive roles (Buzsáki and Wang, 2012; Bosman et al., 2014), and patterns of PAC (Vaz et al., 2017). However, for the visual MEG data presented here, there was only an increase in gamma power within one band (40–70 Hz), and therefore the smearing of adjacent sub-bands is unlikely. Finally, we have focussed on PAC within the visual cortex, which is known to display highly sinusoidal alpha oscillations (Tort et al., 2010a). However, there are many examples of non-sinusoidal brain oscillations caused by physiological neuronal spiking patterns (Fontanini and Katz, 2005), including hippocampal theta (4–8 Hz) and sensorimotor mu (9–11 Hz) rhythms (Lozano-Soldevilla et al., 2016; Scheffer-Teixeira and Tort, 2016), which are indicative of behavior and disease states (Cole and Voytek, 2017). Therefore, whilst non-sinusoidal oscillations generate spurious PAC, this does not mean that these oscillations are uninteresting, but simply that common PAC algorithms, such as, the ones employed in this article, are ill-suited for these scenarios. Where non-sinusoidal oscillations
