Tracking the Dynamic Functional Network Interactions During Goal-Directed Auditory Tasks by Brain State Clustering

Both perceiving and processing external sound stimuli as well as actively maintaining and updating relevant information (i.e., working memory) are critical for communication and problem solving in everyday acoustic environments. The translation of sensory information into perceptual decisions for goal-directed tasks hinges on dynamic changes in neural activity. However, the underlying brain network dynamics involved in this process are not well specified. In this study, we collected functional MRI data of participants engaging in auditory discrimination and auditory working memory tasks. Independent component analysis (ICA) was performed to extract the brain networks involved and the sliding-window functional connectivity (FC) among networks was calculated. Next, a temporal clustering technique was used to identify the brain states underlying auditory processing. Our results identified seven networks configured into four brain states. The number of brain state transitions was negatively correlated with auditory discrimination performance, and the fractional dwell time of State 2-which included connectivity among the triple high-order cognitive networks and the auditory network (AN)-was positively correlated with working memory performance. A comparison of the two tasks showed significant differences in the connectivity of the frontoparietal, default mode, and sensorimotor networks (SMNs), which is consistent with previous studies of the modulation of task load on brain network interaction. In summary, the dynamic network analysis employed in this study allowed us to isolate moment-to-moment fluctuations in inter-network synchrony, find network configuration in each state, and identify the specific state that enables fast, effective performance during auditory processing. This information is important for understanding the key neural mechanisms underlying goal-directed auditory tasks.

Abstract licence: CC BY

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.