BCBL, Basque center on Cognition, Brain and Language, Donostia/San Sebastian, San Sebastian, Spain.
UPV/EHU, Universidad del Pais Vasco, San Sebastian, Spain.
Sci Rep. 2018 Jun 22;8(1):9503. doi: 10.1038/s41598-018-27898-w.
Prediction has been proposed to be a fundamental neurocognitive mechanism. However, its role in language comprehension is currently under debate. In this magnetoencephalography study we aimed to find evidence of word-form phonological pre-activation and to characterize the oscillatory mechanisms supporting this. Participants were presented firstly with a picture of an object, and then, after a delay (fixed or variable), they heard the corresponding word. Target words could contain a phoneme substitution, and participants' task was to detect mispronunciations. Word-initial phonemes were either fricatives or plosives, generating two experimental conditions (expect-fricative and expect-plosive). In the pre-word interval, significant differences (α = 0.05) emerged between conditions both for fixed and variable delays. Source reconstruction of this effect showed a brain-wide network involving several frequency bands, including bilateral superior temporal areas commonly associated with phonological processing, in a theta range. These results show that phonological representations supported by the theta band may be active before word onset, even under temporal uncertainty. However, in the evoked response just prior to the word, differences between conditions were apparent under variable- but not fixed-delays. This suggests that additional top-down mechanisms sensitive to phonological form may be recruited when there is uncertainty in the signal.
预测被认为是一种基本的神经认知机制。然而,其在语言理解中的作用目前仍存在争议。在这项脑磁图研究中,我们旨在寻找词形语音预先激活的证据,并描述支持这种激活的振荡机制。参与者首先看到一个物体的图片,然后在固定或变化的延迟后,他们听到相应的单词。目标单词可以包含一个音位替换,参与者的任务是检测发音错误。单词开头的音位是摩擦音或爆破音,产生了两种实验条件(预期摩擦音和预期爆破音)。在单词前的间隔内,无论是固定延迟还是变化延迟,两种条件之间都出现了显著差异(α=0.05)。对这种效应的源重建显示,一个涉及多个频率带的全脑网络,包括与语音处理相关的双侧颞上区,在 theta 范围内。这些结果表明,即使在时间不确定的情况下,由 theta 带支持的语音表示也可能在单词出现之前就已经活跃。然而,在单词出现前的诱发反应中,只有在变化延迟而不是固定延迟的情况下,两种条件之间才会出现差异。这表明,当信号存在不确定性时,可能会招募到对语音形式敏感的额外的自上而下的机制。