Characterising brain connectivity in terms of contagion: a multiple time series perspective

In understanding the neurobiological underpinnings of alcoholism, we explore EEG data viewed as interrelated multiple time series parallel to the contagion effects observed in financial markets. This study proposed a new methodological approach to compare control and treatment (alcoholism) groups based on their EEG signals. The concept of volatility clustering is used to identify possible grouping of brain signals from EEG to differentiate the characteristics of the control and the treatment groups, following the test developed by (Barrios and Redondo, 2024). Contagion representing clustered volatility is equated to synchronised patterns of activities across different brain regions manifesting signals triggered by particular stimuli, e.g., alcohol intake. The method is illustrated using EEG data taken from The UCI KDD Archive (Begleiter, 1999; Zhang et al., 1995) with 10 alcoholic and 10 control subjects, with 64 electrodes sampling at 256 Hz, under conditions involving either a single stimulus or dual stimuli (matched and non-matched), derived from the Snodgrass and Vanderwart picture set. Each subject completed 10 runs per paradigm, and the data were averaged over 10 trials for each stimulus condition. Volatility clustering in multiple EEG time series can separate the two groups significantly.