Abstracts

Rationale: During relatively simple tasks, such as object naming (e.g. cup), complex interactions of dedicated cortical substrates must work in concert to accomplish perceptual, semantic, and articulatory processes. Yet, after decades of lesional analysis and fMRI studies, putative regions that may serve category-specific functions have just begun to be identified in ventral visual cortical regions. Debate continues over the exact location of these regions, their overlap or specificity, and the nature of information flow between them. We performed electrocorticography (ECoG) in 26 left hemisphere dominant patients to evaluate the global - whole brain - cortical network dynamics during the visual naming of nouns. Methods: Subjects were presented pictures of nouns or scrambled images and asked to overtly name them during ECoG recording. We used a surface-based mixed-effects multilevel grouped analysis (SB-MEMA) to identify significant regions of activation during noun naming, with respect to baseline (Fig. 1; p=0.05, corrected). Individual subdural electrodes (SDEs) overlying these regions were identified and selected for further analyses. Data were filtered into the gamma frequency (60-120Hz) and the percent change in power from baseline computed. A time series representation was used to visualize task-dependent cortical response (Fig. 2; t=0ms, stim onset). Functionally connectivity was assessed using amplitude envelope correlations between channels in each region. Results: Within 500ms after stimulus onset, global cortical activity was observed beginning earliest in posterior occipito-temporal, and then followed by frontal cortex. Of note are early power decreases in prefrontal cortex and ventral visual (e.g. ITG) cortex. Rapid changes in functional connectivity emerged within the same time window across the cortex. Conclusions: During language, highly parallel and bidirectional processes occur across the cortex. Articulatory response planning begins within 500ms after stimulus onset, providing serious constraints to current models of visual-linguistic processes.