Network analysis of neuro-cognitive processes: studying mcgurk effect using EEG data

Abstract

We perceive the world around us through multiple senses. Perception of different senses is linked to different regions in the brain. Integration of the perception of objects through multiple senses enables the brain to make a coherent picture of the world around us. Understanding multisensory perception, therefore, is an important area of brain research as it relates to perception and behavior. In this project, the focus of our study is the interaction between auditory and visual perception. We study this interaction through McGurk effect, using two sets of EEG, data acquired from two different sets of subjects at National Brain Research Center (NBRC), Manesar. These datasets had been studied using the spectral-temporal approach as part of doctoral thesis research of Dr. G. Vinodh Kumar. In this project, we take a di erent approach and focus on studying the McGurk e ect through properties of the brain networks. The approach has been to look at global network measures, area-wise network measures, variation of modularity across time and hub identi cation using node-wise network measures. We nd that the hubs signi cantly di er between /ta/ and /pa/ networks (alpha band) for McGurk stimuli when looking at the 300-380 msecs time window. However, for congruent stimuli, the di erence in hubs is not signi cant. This correlates well with an event related peak found between 300-380 msecs in the earlier study mentioned above. We also nd a peak in modularity at 300msec in alpha band during /pa/ percept, re ecting a segregation of sensory speci c and multi-sensory areas. This segregation may be facilitated by enhanced alpha band coherence as shown in an earlier work. Modularity for Zpa_Per was found to be lower (or less signi cant) compared to Zta Per across frequencies, across thresholds and across time; suggesting that a modular network, i.e. more functional segregation, causes the McGurk e ect to be perceived. We also find that modularity in pre-stimulus signals could be predictive of the participants perception of the illusion { brain's intrinsic functional networks seem to adjust to the incoming stimuli to form percept. We also notice some other differences in the network properties. To understand signi cance of our networks/results, we compare the brain networks for some cases against random networks.