The domain-separation language network dynamics in resting state support its flexible functional segregation and...

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Linguistics

Background:

• Modern linguistic theories and network science suggest language processing relies on hierarchical, segregated large-scale subnetworks.
• These include dorsal (phonological) and ventral (semantic) streams, flexibly recruited for receptive and expressive language tasks.

Purpose of the Study:

• To investigate if the functional segregation of language streams is supported by underlying network segregation.
• To explore the dynamic reconfiguration of language networks during resting state.

Main Methods:

• A dynamic conditional correlation approach was used to create time-varying language networks.
• K-means clustering identified recurring temporal patterns within these dynamic functional connectivity (dFC) networks.
• Machine learning predicted linguistic performance from dFC states.

Main Results:

• Resting-state language network dynamics robustly clustered into four distinct states.
• Three states spatially resembled speech perception/phonological processing, speech production, and semantic processing.
• The fourth state represented a low-connectivity baseline, and these states dynamically reconfigured in a domain-separation manner, predicting individual linguistic performance.

Conclusions:

• Language network dynamics in resting state exhibit a domain-separation pattern.
• This dynamic "meta-network" framework supports flexible functional segregation and integration crucial for language and speech processing.