Estimating Parallel Processing in a Language Task Using Single-Trial Intracerebral Electroencephalography
通过单次试验颅内脑电图数据，在图片命名任务中检验语言处理模型的并行处理假说，发现不同脑区活动在单次试验水平上重叠度低，挑战了传统并行处理观点。
We provide a quantitative assessment of the parallel-processing hypothesis included in various language-processing models. First, we highlight the importance of reasoning about cognitive processing at the level of single trials rather than using averages. Then, we report the results of an experiment in which the hypothesis was tested at an unprecedented level of granularity with intracerebral data recorded during a picture-naming task. We extracted patterns of significant high-gamma activity from multiple patients and combined them into a single analysis framework that identified consistent patterns. Average signals from different brain regions, presumably indexing distinct cognitive processes, revealed a large degree of concurrent activity. In comparison, at the level of single trials, the temporal overlap of detected significant activity was unexpectedly low, with the exception of activity in sensory cortices. Our novel methodology reveals some limits on the degree to which word production involves parallel processing.