Neural oscillations in the temporal pole for a temporally congruent audio-visual speech detection task.

Takefumi Ohki,Naruhito Hironaga,Takashi Hanakawa,Yuu Kaneko,Hidetoshi Takahashi,Atsuko Gunji,Kazuo Hiraki,Shozo Tobimatsu,Masumi Inagaki,Yuichi Takei,Yoko Kamio,Yosuke Kita

doi:10.1038/srep37973

Abstract

Though recent studies have elucidated the earliest mechanisms of processing in multisensory integration, our understanding of how multisensory integration of more sustained and complicated stimuli is implemented in higher-level association cortices is lacking. In this study, we used magnetoencephalography (MEG) to determine how neural oscillations alter local and global connectivity during multisensory integration processing. We acquired MEG data from 15 healthy volunteers performing an audio-visual speech matching task. We selected regions of interest (ROIs) using whole brain time-frequency analyses (power spectrum density and wavelet transform), then applied phase amplitude coupling (PAC) and imaginary coherence measurements to them. We identified prominent delta band power in the temporal pole (TP), and a remarkable PAC between delta band phase and beta band amplitude. Furthermore, imaginary coherence analysis demonstrated that the temporal pole and well-known multisensory areas (e.g., posterior parietal cortex and post-central areas) are coordinated through delta-phase coherence. Thus, our results suggest that modulation of connectivity within the local network, and of that between the local and global network, is important for audio-visual speech integration. In short, these neural oscillatory mechanisms within and between higher-level association cortices provide new insights into the brain mechanism underlying audio-visual integration.

Highlights

The detection of the stimuli[9]
We found that accuracy in the temporal congruency (TC) condition was significantly higher than in the temporal incongruency (TI) condition, which indicated that the audio-visual binding process was conducted more successfully in the TC condition (z-value 4.215, p < 0.001)
The likelihood ratio test demonstrated that reaction time (RT) followed different ex-Gauss distributions across temporal conditions, which suggests that RT between the TC and TI conditions was significantly different (p < 0.001)

Summary

Introduction

The detection of the stimuli[9]. These studies demonstrated the early influence of multisensory processing in primary cortices[1,10], but the mechanisms through which multisensory integration in higher-order brain regions takes place are still unclear. The TP is assumed to play a more significant role than the STS in identifying a speaker as a representative higher integrative process[17,18] To elucidate such functional aspects of the TP, we focused on “the cocktail party effect”, in which a listener isolates and attends a specific speaker despite external noises with the aid of visual cues such as the speaker’s mouth movements[19,20,21]. For whole-brain network encoding, interregional oscillatory coordination (coherence) is thought to mediate information transfer across neural networks[33,34,35] It appears to be crucial for perceptual and cognitive processes[36,37,38], and for cross-modal sensory processing[39,40]

Methods

Results

Conclusion