Convergent hierarchical dynamics within the language network for speech listening and silent reading

Introduction: Spoken and written language are crucial to communication and likely recruit shared networks to map phonology and orthography to meaning. Models of comprehension for both domains suggest overlapping processes. The cortical instantiation of these models is critical to our understanding of comprehension, but the convergent neural mechanisms are unclear. We investigate the spatiotemporal dynamics of shared networks for listening and silent reading using ECoG.

Methods: 63 patients underwent ECoG and completed cued naming to matched spoken and written descriptions. We analyzed broadband gamma activity (BGA;70-150Hz) and integrated group responses with mixed-effects multilevel analyses. We then analyzed BGA from depths along STG (n=37). Lastly, we applied chronometric stimulation of STG during listening and silent reading (n=2).

Results: For each spoken word, sustained activation of STG (128% BGA) was followed by activation of posterior middle temporal gyrus (pMTG;41% BGA). For each written word, visual cortex activity was followed by activation of lateral occipitotemporal cortex, fusiform gyrus (Fus) intraparietal sulcus (IPS), STG (20%BGA), and pMTG (29%BGA). For both modalities, the last word triggered an identical lexical processing network (pMTG,Fus,IPS,pars triangularis). Activity from depths along STG (n=37) revealed a consistent anteroposterior gradient. Electrodes in Heschl’s gyrus (HG) showed sustained activation during listening consistent with acoustic processing and were quiescent during reading. Electrodes in planum temporale (PT) showed transient activity at speech onset and recurrent activity for written words. Stimulation of HG disrupted listening only, while stimulation of PT disrupted both listening and reading. Our prior work implicates PT in predictive encoding during production - this work suggests the same occurs in silent reading.

Conclusion : Using large-scale ECoG, we derived new insights into the neural basis of spoken and written language. Ultimately, we believe that this work would provide important insights to optimize the design of neural prosthetics for the treatment of reading-related disorders.