Barbero Francesca M, Calce Roberta P, Talwar Siddharth, Rossion Bruno, Collignon Olivier
Institute of Research in Psychology and Institute of Neuroscience, Louvain Bionics Center, University of Louvain, Louvain-la-Neuve 1348, Belgium
Institute of Research in Psychology and Institute of Neuroscience, Louvain Bionics Center, University of Louvain, Louvain-la-Neuve 1348, Belgium.
eNeuro. 2021 Jun 24;8(3). doi: 10.1523/ENEURO.0471-20.2021. Print 2021 May-Jun.
Voices are arguably among the most relevant sounds in humans' everyday life, and several studies have suggested the existence of voice-selective regions in the human brain. Despite two decades of research, defining the human brain regions supporting voice recognition remains challenging. Moreover, whether neural selectivity to voices is merely driven by acoustic properties specific to human voices (e.g., spectrogram, harmonicity), or whether it also reflects a higher-level categorization response is still under debate. Here, we objectively measured rapid automatic categorization responses to human voices with fast periodic auditory stimulation (FPAS) combined with electroencephalography (EEG). Participants were tested with stimulation sequences containing heterogeneous non-vocal sounds from different categories presented at 4 Hz (i.e., four stimuli/s), with vocal sounds appearing every three stimuli (1.333 Hz). A few minutes of stimulation are sufficient to elicit robust 1.333 Hz voice-selective focal brain responses over superior temporal regions of individual participants. This response is virtually absent for sequences using frequency-scrambled sounds, but is clearly observed when voices are presented among sounds from musical instruments matched for pitch and harmonicity-to-noise ratio (HNR). Overall, our FPAS paradigm demonstrates that the human brain seamlessly categorizes human voices when compared with other sounds including musical instruments' sounds matched for low level acoustic features and that voice-selective responses are at least partially independent from low-level acoustic features, making it a powerful and versatile tool to understand human auditory categorization in general.
可以说，声音是人类日常生活中最相关的声音之一，多项研究表明人类大脑中存在声音选择性区域。尽管经过了二十年的研究，但确定支持语音识别的人类大脑区域仍然具有挑战性。此外，对声音的神经选择性是仅仅由人类声音特有的声学特性（例如，频谱图、谐波性）驱动，还是也反映了更高层次的分类反应，仍在争论之中。在这里，我们通过快速周期性听觉刺激（FPAS）结合脑电图（EEG）客观地测量了对人类声音的快速自动分类反应。参与者接受包含以4赫兹（即每秒四个刺激）呈现的来自不同类别的异质非语音声音的刺激序列测试，语音声音每三个刺激出现一次（1.333赫兹）。几分钟的刺激足以在个体参与者的颞上区域引发强大的1.333赫兹声音选择性局灶性脑反应。对于使用频率扰乱声音的序列，这种反应几乎不存在，但当声音出现在音高和谐波噪声比（HNR）匹配的乐器声音中时，可以清楚地观察到这种反应。总体而言，我们的FPAS范式表明，与包括低水平声学特征匹配的乐器声音在内的其他声音相比，人类大脑能够无缝地对人类声音进行分类，并且声音选择性反应至少部分独立于低水平声学特征，这使其成为理解一般人类听觉分类的强大而通用的工具。
