Whole-Brain Millisecond-Scale Effective Connectivity Atlas of Auditory and Visual Naming

Abstract

Neurobiological models suggest that speech relies on interactions among distant cortical regions interconnected by white matter pathways. Prior studies have shown that speech-related functional coactivation--simultaneous high-gamma augmentation across distributed regions--reflects underlying neural interactions, as validated by electrical stimulation mapping. However, it remains unclear at the whole-brain level when, for how long, and in which directions cortical regions transmit information, and how dynamic information flows contribute to speech. Here, we investigated the causal roles of directional neural information flow during auditory and visual naming using intracranial EEG from 9,526 artifact-free nonepileptic sites across 127 patients. Information flow, estimated by transfer entropy-based effective connectivity, was classified as excitatory when high-gamma acceleration in one region predicted subsequent acceleration in another, and inhibitory when deceleration predicted downstream deceleration. Following stimulus onset, excitatory flows emerged from modality-specific sensory cortices and propagated to higher-order regions, later becoming bidirectional as functional coactivation developed. Each excitatory flow event was transient (<500 ms), typically followed by inhibitory flows and decreased coactivation, during which excitatory flows often emerged along new pathways. During auditory naming, faster response times were associated with stronger excitatory flows in left perisylvian regions; during visual naming, faster responses were linked to stronger flows in bilateral basal temporal cortices. Stronger excitatory flows at specific time points predicted a higher probability of stimulation-induced symptoms (Spearmans rho: 0.54-0.81; p<0.00001), whereas associations with inhibitory flows peaked later. Excitatory flows near visual naming response onset were particularly associated with stimulation-induced speech arrest and face sensorimotor symptoms, whereas functional coactivation alone failed to reveal comparable associations. These findings demonstrate that transient acceleration of directional neural interactions through white matter supports successive stages of speech processing. As activity within one pathway decelerates, excitatory flow accelerates along another, enabling functional transitions critical for naming. By establishing the causal contribution of temporally precise, anatomically specific white matter pathways, this study substantiates and extends existing neurobiological models of speech. To facilitate replication and dynamic whole-brain visualization, we provide open access to the full dataset (62.15 GB) and analysis code.