Linguistic input drives brain network configuration during language comprehension

Abstract

Assessing the synchrony and interplay between distributed neural regions is critical to understanding how language is processed. Here, we investigated possible neuro-functional links between form and meaning during sentence comprehension combining a classical wholebrain approach to characterize patterns of brain activation resulting from our experimental manipulation with a novel multivariate network-based approach where the combination of graph-theory measures allow us to unravel the architectonic configuration of the language system. Capitalizing on the Spanish gender agreement system, we experimentally manipulated formal and conceptual factors: whether the noun-adjective grammatical gender relationship was congruent or not and whether the noun gender type was conceptual or strictly formal. Left inferior and middle frontal gyri, as well as left MTG/STG emerged as critical areas for the computation of grammatical relations. However, critically, we demonstrate how the interface between formal and conceptual features depends on the synergic articulation of brain areas divided in three subnetworks and extends beyond this classical left-lateralized perisylvian language circuit. Critically, we isolated a subregion of the left angular gyrus showing a significant interaction between gender congruency and gender type. The functional interplay between the angular gyrus and left perisylvian language-specific circuit was identified as crucial for constructing coherent and meaningful messages. Importantly, using graph theory we show the functional malleability of this complex system, so that the role each node play within the network changes depending on the available linguistic cues. Significance Statement:Neural networks can be described as graphs comprising distributed and interconnected nodes. These nodes share functional properties but also differ in terms of functional specialization and the number of interconnections mediating the efficient transfer of information. Previous work has shown functional connectivity differences based on graph-theory properties between typical and atypical samples. However, here graph-theory properties have been used to characterize the language connectivity signature using task-related fMRI. This approach allowed us to demonstrate how linguistic input drives brain network configuration during language comprehension. This is the first evidence for functional flexibility in brain networks: the communication capacity of each hub changes depending on whether the linguistic input grants access to meaning or whether is purely formal.