Medicum, Faculty of Medicine, P.O. Box 63, FI-00014, University of Helsinki, Finland; Department of Neuroscience and Biomedical Engineering, P.O. Box 12200, FI-00076, Aalto University, Finland; Institute of Neuroscience and Psychology, University of Glasgow, 58 Hillhead Street, Glasgow, G12 8QB, UK; MEG Core, Aalto NeuroImaging, FI-00076, Aalto University, Finland.
Department of Signal Processing and Acoustics, P.O. Box 13000, FI-00076, Aalto University, Finland.
Neuroimage. 2020 Oct 1;219:116936. doi: 10.1016/j.neuroimage.2020.116936. Epub 2020 May 29.
Natural speech builds on contextual relations that can prompt predictions of upcoming utterances. To study the neural underpinnings of such predictive processing we asked 10 healthy adults to listen to a 1-h-long audiobook while their magnetoencephalographic (MEG) brain activity was recorded. We correlated the MEG signals with acoustic speech envelope, as well as with estimates of Bayesian word probability with and without the contextual word sequence (N-gram and Unigram, respectively), with a focus on time-lags. The MEG signals of auditory and sensorimotor cortices were strongly coupled to the speech envelope at the rates of syllables (4-8 Hz) and of prosody and intonation (0.5-2 Hz). The probability structure of word sequences, independently of the acoustical features, affected the ≤ 2-Hz signals extensively in auditory and rolandic regions, in precuneus, occipital cortices, and lateral and medial frontal regions. Fine-grained temporal progression patterns occurred across brain regions 100-1000 ms after word onsets. Although the acoustic effects were observed in both hemispheres, the contextual influences were statistically significantly lateralized to the left hemisphere. These results serve as a brain signature of the predictability of word sequences in listened continuous speech, confirming and extending previous results to demonstrate that deeply-learned knowledge and recent contextual information are employed dynamically and in a left-hemisphere-dominant manner in predicting the forthcoming words in natural speech.
自然语言建立在上下文关系的基础上，可以提示对即将到来的话语的预测。为了研究这种预测处理的神经基础，我们要求 10 名健康成年人在听一本 1 小时长的有声读物时记录他们的脑磁图 (MEG) 大脑活动。我们将 MEG 信号与语音包络相关联，以及与贝叶斯单词概率的估计相关联，分别带有和不带有上下文单词序列 (N 元组和 Unigram)，重点关注时间滞后。听觉和感觉运动皮层的 MEG 信号与语音包络以音节（4-8Hz）和韵律和语调（0.5-2Hz）的速率强烈耦合。词序列的概率结构，独立于声学特征，广泛影响听觉和罗兰德区域、后扣带回、枕叶皮质以及外侧和内侧额区域的≤2Hz 信号。在词起始后 100-1000 毫秒，在大脑区域中发生了精细的时间进展模式。虽然在两个半球都观察到了声学效应，但上下文影响在统计学上明显偏向于左半球。这些结果是连续语言中词序列可预测性的大脑特征，证实并扩展了先前的结果，表明深度学习的知识和最近的上下文信息在预测自然语言中即将到来的单词时被动态且以左半球为主导的方式使用。
