Le Bao, Bjoring Margot C, Meliza C Daniel
Department of Psychology, University of Virginia, Charlottesville, VA, USA.
Program in Fundamental Neuroscience, University of Virginia, Charlottesville, VA, USA.
Nat Commun. 2025 Sep 26;16(1):8452. doi: 10.1038/s41467-025-63182-y.
In the perceptual illusion known as phonemic (or auditory) restoration, when part of a word is deleted and replaced by noise, listeners still hear the complete word. The neural mechanisms that create this illusion by generating predictions of missing information remain poorly understood. Zebra finches (Taeniopygia castanotis) use song, a sequence of complex vocal elements, to communicate in noisy social environments. Here, we found that in anesthetized finches, populations of single units in the homolog of auditory cortex respond to songs in which the original audio of an element is completely replaced by noise as if the missing element were present alongside the noise, despite only noise being physically present. This occurs even for songs birds have never heard, but not if the context is masked or lacks species-typical syntax. These results suggest that local neural dynamics instantiate a general model of conspecific song that biases auditory responses to restore missing information.
在被称为音素(或听觉)恢复的感知错觉中,当单词的一部分被删除并用噪声替代时,听众仍然能听到完整的单词。通过生成缺失信息的预测来产生这种错觉的神经机制仍未得到充分理解。斑胸草雀(Taeniopygia castanotis)利用歌声,即一系列复杂的发声元素,在嘈杂的社交环境中进行交流。在这里,我们发现,在麻醉的雀类中,听觉皮层同源区域的单个神经元群体对其中某个元素的原始音频完全被噪声替代的歌声做出反应,就好像缺失的元素与噪声同时存在一样,尽管实际仅存在噪声。即使对于鸟类从未听过的歌声,这种情况也会发生,但如果语境被掩盖或缺乏物种典型的句法结构,则不会出现这种情况。这些结果表明,局部神经动力学实例化了一个同种歌声的通用模型,该模型会使听觉反应产生偏差以恢复缺失的信息。