Characterizing Variability of Audiovisual Speech Perception Based on Periodic and Aperiodic Features of Prestimulus Brain Activity

Abstract

AbstractBurgeoning pieces of evidence support the idea that differences in the properties of large-scale cortical networks underlie perceptual variability. Here, employing psychophysical experiments in tandem with electroencephalogram (EEG) recordings, we explored the inter-individual and inter-trial differences in McGurk percept to probe the underlying brain dynamics before (prestimulus) and after (poststimulus) the stimulus presentation. Initially, based on our participants’ (n = 18) overall susceptibility to the McGurk effect, we categorized them into two groups - ’rare’ and ’frequent’ groups of perceivers. Subsequently, we investigated the differences in their neuronal dynamics manifested in the oscillatory properties of the prestimulus and compared it with the poststimulus EEG signals during the McGurk perception (audio- /pa/, video- /ka/) and, also during the perception of the congruent audiovisual (AV) stimuli (syllables /pa/, /ta/, and /ka/). We primarily focused on the power spectrum components (periodic oscillations and aperiodic components) and global coherence. At the inter-individual and inter-trial level, we observed that for the prestimulus duration, the ’rare’ perceivers’ aperiodic (or “background” 1/f) component elicited a lower beta band (15-30 Hz) and gamma band (31-45 Hz) activity than the ’frequent’ group of perceivers. Moreover, the periodic oscillations for the ’rare’ group showed significance in the alpha (8-12 Hz) and beta bands during the prestimulus time window. ’Rare’ perceivers also showed an increased global coherence in the theta (4-7 Hz), alpha, and beta bands but a lower gamma band activity than the ’frequent’ perceivers. Finally, ’rare’ perceivers elicited higher cross-hemispheric connectivity (imaginary coherence) than the ’frequent’ group at the source across different frequency bands, supporting the notion that the fluctuations in the distributed large-scale network underlie the perceptual variability in speech across individuals.