Data Availability statement
All spectral analyses and connectivity analyses were performed using Fieldtrip toolbox available at https://www.fieldtriptoolbox.org/. Figure data and codes used here are available at https://github.com/supratimray/TLSAEEGProjectPrograms under the connectivityProject folder. The raw data that support the findings of this study are openly available in the OSF repository at https://osf.io/ebryn/
Introduction
Neural coordination referred as functional connectivity (FC) (Fingelkurts et al. , 2005), refers to the statistical relationship between spatially distant neurophysiological events which is generally considered to be mediated by neuronal assemblies (Mcintosh, 1999). FC can be studied at various spatial and temporal scales, using brain signals such as fMRI, electro/magneto-encephalogram (EEG/MEG) or neuronal firing activity, and is quantified using a variety of techniques such as coherence (Srinivasan et al. , 2007), imaginary part of coherence (Nolte et al. , 2004), phase locking value (PLV; (Lachaux et al. , 2000)), pairwise phase consistency (PPC; (Vincket al. , 2010)) and so on. Many studies have reported changes in FC in different frequency bands with cognitive tasks such as working memory visual discrimination (Lachaux et al. , 1998), emotional face processing (Li et al. , 2015), fixation tasks (Murty et al. , 2018), pain perception tasks (Alba et al. , 2022), motor learning tasks (Heitger et al. , 2012), aging and with brain disorders (Ishii et al. , 2017). For example, reduction in alpha (8-12 Hz) FC was reported with healthy aging in resting state EEG (Moezzi et al. , 2019), while gamma (30-55 Hz) FC reduction was reported in autism (Safar et al. , 2020), and in Alzheimer’s Disease (AD) and schizophrenia (Uhlhaas & Singer, 2006). Interestingly, some studies have also reported an increase in gamma FC in subjects with mild-cognitive-impairment (MCI) (Rodinskaia et al. , 2022) and AD (Stam et al. , 2009).
Majority of EEG FC studies that describe the effect of aging and cognitive disorder are during resting state, where many factors can alter the data, such as specific instructions, cognitive state before recording, caffeine intake, or random episodic spontaneous thoughts (van Diessen et al. , 2015). Having a stimulus dependent paradigm restricts subject specific variations. Such responses are termed as either evoked or induced, based on their trial-wise relationship with stimulus onset (Tallon-Baudry & Bertrand, 1999). During periods of spontaneous activity, while alpha (Gómez et al. , 2013; Scallyet al. , 2018) and beta rhythms are prominent, with distinct bumps in the power spectral density (PSD) of brain signals, there is rarely any periodic activity in the gamma range. On the other hand, presentation of certain visual stimuli such as long bars and large gratings produce a distinct “narrow-band gamma” – with a bandwidth of ~20 Hz and centre frequency between 30-70 Hz (Grayet al. , 1989) – which has a clear bump in the PSD and whose power and centre frequency depend on stimulus properties (Jia et al. , 2013; Murty et al. , 2018). This stimulus-induced narrowband gamma is also distinct from “broadband gamma”, which is typically associated with increases in firing rate (see, for example, (Ray & Maunsell, 2011)). While there are many studies where the effect of FC/power is compared in MCI vs healthy controls ((Berendse et al. , 2000; Rodinskaia et al. , 2022); see Discussion for more references), these were studied mainly during resting state, and therefore did not produce narrowband gamma. Recent studies have shown that such narrow-band gamma oscillations can be induced in human EEG/MEG by presentation of stationary cartesian or moving annular gratings (Orekhova et al. , 2015; Murty et al. , 2018; Pantaziset al. , 2018), whose power changes with age (Murty et al. , 2020) and mental disorders such as MCI/AD (Murty et al. , 2021). However, whether aging or mental disorders change stimulus-induced gamma FC, to our knowledge, is not well studied.
Further, while both aging and cognitive disorders generally reduce the power of brain oscillations, the effect of these two factors on FC could be different. For example, Sullivan and colleagues (2019) reported that healthy aging and MCI produce distinct changes in fMRI FC (measured using bivariate-correlation) across different networks in the brain. Another study (Finnigan & Robertson, 2011) reported altered relationship between alpha (8-12 Hz) and theta (4-8 Hz) oscillations with healthy aging and pathological aging. Finding robust connectivity measures that distinguish aging versus mental disorder could help in developing better biomarkers that are more sensitive to the onset of cognitive disorders compared to power-based biomarkers.
The main objective of this study is to test how stimulus-induced gamma FC varies with healthy aging and in MCI subjects compared to their healthy controls. We recorded EEG data from healthy (N=218) and MCI (N=11) subjects while they maintained fixation on large achromatic gratings presented on a monitor (same dataset as the one used in (Murtyet al. , 2020, 2021)). Large gratings were shown to induce two stimulus-induced gamma oscillations, termed “slow” (20-34 Hz) and “fast” (35-65 Hz) (Murty et al. , 2018, 2020, 2021). Further, these large stimuli produced a prominent suppression of the alpha power. Since the stimuli produced prominent changes in alpha, slow gamma and fast gamma bands, we estimated FC in these three bands using PPC, which is an unbiased estimator of the squared PLV (see (Vinck et al. , 2010) for details). We first analysed the effect of aging on visual stimulus-induced FC by splitting subjects into middle-aged (50-64 yrs) and elderly (>64 yrs) groups, as done in previous studies (Murty et al. , 2020, 2021). Since FC could potentially be dependent on power along with aging, we performed a power-matching procedure (see detailed below) to control for power dependence. We also tested FC dependency on power and age using a robust linear regression model. Finally, we compared the FC for MCI subjects versus their age and gender matched controls.