Helen Wills Neuroscience Institute, University of California Berkeley, Berkeley, CA, USA.
Center for Music in the Brain, Aarhus University & The Royal Academy of Music, Aarhus, Denmark.
Eur J Neurosci. 2022 Sep;56(5):4583-4599. doi: 10.1111/ejn.15769. Epub 2022 Jul 26.
Many natural sounds have frequency spectra composed of integer multiples of a fundamental frequency. This property, known as harmonicity, plays an important role in auditory information processing. However, the extent to which harmonicity influences the processing of sound features beyond pitch is still unclear. This is interesting because harmonic sounds have lower information entropy than inharmonic sounds. According to predictive processing accounts of perception, this property could produce more salient neural responses due to the brain's weighting of sensory signals according to their uncertainty. In the present study, we used electroencephalography to investigate brain responses to harmonic and inharmonic sounds commonly occurring in music: Piano tones and hi-hat cymbal sounds. In a multifeature oddball paradigm, we measured mismatch negativity (MMN) and P3a responses to timbre, intensity, and location deviants in listeners with and without congenital amusia-an impairment of pitch processing. As hypothesized, we observed larger amplitudes and earlier latencies (for both MMN and P3a) in harmonic compared with inharmonic sounds. These harmonicity effects were modulated by sound feature. Moreover, the difference in P3a latency between harmonic and inharmonic sounds was larger for controls than amusics. We propose an explanation of these results based on predictive coding and discuss the relationship between harmonicity, information entropy, and precision weighting of prediction errors.
许多自然声音的频谱由基频的整数倍组成。这种特性,即谐和性,在听觉信息处理中起着重要作用。然而,谐和性在多大程度上影响了音高以外的声音特征的处理尚不清楚。这很有趣,因为谐和声音的信息熵低于非谐和声音。根据感知的预测处理理论,由于大脑根据其不确定性对感觉信号进行加权,这种特性可能会产生更显著的神经反应。在本研究中,我们使用脑电图来研究在音乐中常见的谐和和非谐和声音(钢琴音和高帽钹声)对大脑的反应。在多特征Oddball 范式中,我们测量了具有和不具有先天性失歌症(一种音高处理障碍)的听众对音色、强度和位置偏差的错配负波(MMN)和 P3a 反应。正如假设的那样,我们观察到与非谐和声音相比,谐和声音的振幅更大,潜伏期更早(MMN 和 P3a 均如此)。这些谐和性效应受声音特征的调节。此外,控制组的 P3a 潜伏期与谐和声音和非谐和声音之间的差异大于失歌症患者。我们基于预测编码提出了对这些结果的解释,并讨论了谐和性、信息熵和预测误差的精度加权之间的关系。