Collaborative Research: Neural-cognitive analysis of spatial scenes with competing, dynamic sound sources

本项目研究从混合声源中提取重要信息的神经认知机制。想象一下，有一天，你再也无法区分迎面而来的汽车喇叭声和其他街道上的声音。这种专注于一个声源而忽略其他声源的认知能力对听力障碍者来说是一个日常挑战。虽然检测和定位单个声音的基本神经机制是已知的，但我们不知道大脑如何完成对多个声源的听觉场景分析。到目前为止，研究主要集中在啮齿动物和食肉动物的较低脑中心，而源分离的神经机制预计将在更高的水平，在听觉皮层。这项研究将记录单个皮层神经元的反应，并对相同的声学场景进行人体实验。 基于这些结果的整合，一个功能性的听觉模型将被开发。 这将提供新的科学见解，并为助听器，社交机器人和监控系统提供智能算法。该项目将为研究生和本科生提供研究机会，并为高中和大学生提供外展活动和在线学习资源，以提高公众对神经科学的认识。研究成果和模型将与学术界分享。本计画将以跨学科的方法，整合心理声学实验与单一单元电生理学，以了解听觉场景分析的中枢机制。本研究将探讨听觉系统如何定位一个目标声音在时间上嵌入在一个空间上分离的掩蔽。单单位录音将针对听觉皮层的尾侧区域，这是复杂声音分析的假定“何处”通路。我们假设皮层活动代表新旧声音，因此可以从整体混合物中减去“旧”掩蔽源的内部表示。这有助于更清晰地感知“新”目标元素，展示了听觉场景分析中的基本心理物理现象。为了验证这一假设，我们将分别和组合识别单个声源的神经信号。然后，我们将解释这些信号的感知数据的基础上获得的声音定位测试与多个移动和固定的声源。发现听觉场景分析的基本脑机制将为完善的心理物理学领域提供新的神经生理学见解，并为声源分离提供潜在的技术解决方案。

This project investigates neurocognitive mechanisms that extract important information from a mixture of sound sources. Imagine a day where you could no longer distinguish the honking horn of a car coming right at you from other street sounds. This cognitive ability to attend to one sound source while ignoring others presents an everyday challenge for people with hearing impairments. While the basic neural mechanisms for detecting and localizing single sounds are known, we do not know how the brain accomplishes auditory scene analysis with multiple sound sources. So far, studies have focused on lower brain centers in rodents and carnivores, while the neural mechanisms for source segregation are expected to be at higher levels, in the auditory cortex. This study will record the responses of single cortical neurons and conduct human-subject experiments for the same acoustic scenarios. Based on the integration of these results, a functional auditory model will be developed. This will provide new scientific insights and enable intelligent algorithms for hearing aids, social robotics, and surveillance systems. The project will provide research opportunities for graduate and undergraduate students and include outreach activities and online learning resources for high-school and college students to increase the public awareness of neuroscience. The research results and the model will be shared with the academic community. This proposal will use an interdisciplinary approach to gain understanding of the central mechanisms of auditory scene analysis by integrating psychoacoustical experiments with single-unit electrophysiology. The study will investigate how the auditory system localizes a target sound temporally embedded in a spatially separated masker. Single-unit recording will target the caudal region of the auditory cortex, the putative "where" pathway for complex sound analysis. We hypothesize that cortical activity represents both the old and new sounds, so that the internal representation of the "old" masking source can be subtracted from the overall mixture. This facilitates a clearer perception of the "new" target element, demonstrating a fundamental psychophysical phenomenon within auditory scene analysis. To test this hypothesis, we will identify the neural signals for individual sound sources separately and in combination. We will then interpret these signals based on the perceptual data gained from sound localization tests with multiple moving and stationary sound sources. Discovering the fundamental brain mechanisms for auditory scene analysis will provide new neurophysiological insight into a well-established psychophysical field and offer potential technical solutions for sound-source segregation.