Naturalistic language comprehension engages a cascade of widespread brain networks in the one second following comprehension

Language comprehension (LC) is a cornerstone of human cognition, enabling the extraction of meaning from written and spoken communication with remarkable efficiency. Decades of neuroimaging research have identified the brain networks associated with LC, but limitations in spatial and temporal resolution have hindered a comprehensive characterization of the whole-brain (millimeter scale), real-time (millisecond precision) dynamics underlying this process. To overcome these constraints, we applied a fusion of multimodal brain imaging techniques (fMRI and EEG) in healthy adults (n = 30) to map the spatiotemporal progression of neural network engagement during LC in the one second following comprehension. Our findings reveal a cascade of brain network activations, beginning with the occipitotemporal perceptual word processing network (250 ms), followed by the temporoparietal semantic retrieval network (400 ms), the posterior default mode inferential network (500 ms), the frontotemporal semantic integration network (600 ms), and finally, a distributed goal-directed comprehension network (700 ms). Crucially, inferential processing emerged as a “hinge point,” linking early word processing to later higher-order networks. Efficient LC was associated with greater mediation by this inferential network and reduced reliance on top-down semantic integration. These findings provide evidence that naturalistic LC relies on rapid, dynamic interactions across widespread brain networks, with individual differences in LC reflecting specific subnetwork interactions. This work offers a framework for investigating the temporal evolution of distributed brain network dynamics in complex cognition across domains and clinical populations.