Cognition and Brain Plasticity Unit, Department of Cognition, Development and Educational Psychology, University of Barcelona, Barcelona, Spain; Institut D'Investigació Biomèdica de Bellvitge (IDIBELL), Hospitalet de Llobregat, Barcelona, Spain.
Radboud University, Donders Centre for Cognition, Nijmegen, the Netherlands; Radboudumc, Donders Centre for Medical Neuroscience, Department of Medical Psychology, Nijmegen, the Netherlands.
Neuropsychologia. 2022 Aug 13;173:108286. doi: 10.1016/j.neuropsychologia.2022.108286. Epub 2022 Jun 6.
Alpha and beta power decreases have been associated with prediction in a variety of cognitive domains. Recent studies in sentence comprehension have also reported alpha and/or beta power decreases preceding contextually predictable words, albeit with remarkable spatiotemporal variability across reports. To contribute to the understanding of the mechanisms underlying this phenomenon, and the sources of variability, the present study explored to what extent these prediction-related alpha and beta power decreases might be common across different modalities of comprehension. To address this, we re-analysed the data of two EEG experiments that employed the same materials in written and in spoken comprehension. Sentence contexts were weakly or strongly constraining about a sentence-final word, which was presented after a 1 s delay, either matching or mismatching the expectation. In written comprehension, alpha power (8-12 Hz) decreased before final words appearing in strongly (relative to weakly) constraining contexts, in line with previous reports. Furthermore, a similar oscillatory phenomenon was evidenced in spoken comprehension, although with relevant spatiotemporal differences. Altogether, the findings agree with the involvement of both modality-specific and general-domain mechanisms in the elicitation of prediction-related alpha power decreases in sentence comprehension. Specifically, we propose that this phenomenon might partly reflect richer and more precise information representation when linguistic contexts afford prediction.
α 和 β 频段的功率降低与各种认知领域的预测有关。最近在句子理解方面的研究也报告了在语境可预测的单词之前出现的 α 和/或 β 频段的功率降低,尽管在不同的报告中存在显著的时空可变性。为了深入了解这种现象背后的机制和可变性的来源,本研究探讨了这些与预测相关的 α 和 β 频段功率降低在多大程度上可能在不同的理解模态中是普遍存在的。为了解决这个问题,我们重新分析了两项 EEG 实验的数据,这些实验在书面和口语理解中使用了相同的材料。句子上下文对句子结尾词的约束较弱或较强,在 1 秒延迟后呈现,与预期相符或不相符。在书面理解中,与之前的报告一致,在强(相对于弱)约束语境中出现的最后一个词之前,α 频段(8-12 Hz)的功率降低。此外,在口语理解中也出现了类似的振荡现象,尽管存在相关的时空差异。总之,这些发现与在句子理解中诱发与预测相关的 α 频段功率降低的特定模态和一般领域机制的参与一致。具体来说,我们提出这种现象可能部分反映了当语言语境提供预测时更丰富和更精确的信息表示。