Relationship between FC and power
We found a negative correlation between PPC connectivity and change in power in slow and fast gamma bands (Figure 6b). The reasons are unclear, but in line with other studies that have reported similarities between PSDs and network measures computed from FC measures such as PLV, and cautioned against treating them as completely separate measures (Demuruet al. , 2020). One possibility is that when the oscillatory signal is weak, the FC is governed by the background spectral power which may have a larger spatial spread. Another possibility is that the source of the oscillation is deeper, which would lead to lower power at the sensor level but nonetheless stronger FC across sensors since more sensors would capture the source activity.
On the other hand, several reports have shown that power and FC capture different features of the underlying neural activity. For example, differences in power versus FC with age and cognitive decline (MCI) reported here have also been observed in resting state EEG, although the strongest effect is observed at lower frequencies such as delta and theta bands (Meghdadi et al. , 2021). Similarly, FC could convey different information compared to PSDs about mental disorders. For example, patients with schizophrenia have higher EEG complexity at lower frequencies (Takahashi et al. , 2010), which is shown to reflect reduced FC (Friston, 1996). In addition, FC is shown to capture subject specific information like the demographic variables, while being consistent across a variety of tasks (Nentwich et al. , 2020). However, a recent study reported source and sensor space FC to be less reproducible than either the absolute or relative powers (Duan et al. , 2021). Note that most of these earlier studies used resting state EEG data. Our study extends these findings for stimulus-induced oscillations and shows that power and FC alter in distinct ways with aging and cognitive disorders, and the changes in FC can be observed even after accounting for the difference in power. Indeed, some studies have shown that compared to resting-state EEG, a paradigm involving visual stimulation/perception is more sensitive to AD disease effects (Barzegaran et al. , 2016).