Spatial Stream Segregation in Auditory Cortex of the Behaving Monkey

声学环境是复杂的，因为它通常包含先验未知数量的未知声源。因此，从环境中提取特定的声源是一个令人生畏的神经计算问题，原则上是数学不适定的。然而，听觉系统在处理复杂的声音混合方面非常成功，使我们能够在拥挤的鸡尾酒会上跟随谈话，或者欣赏管弦乐队的表演。解析声学环境的过程被称为听觉场景分析（阿萨）。阿萨被认为依赖于场景的频谱-时间模式，以及听觉系统所采用的内部先验。系统可用的一个重要线索是空间位置。空间线索的处理在听觉皮层的水平上得到了相对较好的理解。相当少的是已知的神经元过程的作用，空间线索阿萨。此外，听力受损的听众对阿萨有实质性的问题，因为目前的助听器和耳蜗植入物仍然不能分离复杂的声学混合物。我的项目旨在揭示行为恒河猴听觉皮层中空间流的神经元基底，从而（i）进一步了解空间线索如何使阿萨在人类听众中发挥作用，（ii）将导致改进的人工语音噪声识别算法，从而直接改善听力受损者的阿萨。

The acoustic environment is complex, as it usually contains an a priori unknown number of unknown sound sources. To extract a particular sound source from the environment is therefore a daunting neuro-computational problem, which is in principle mathematically ill-posed. Nevertheless, the auditory system is very successful in dealing with complex sound mixtures, enabling us to follow a conversation at a crowded cocktail party, or to enjoy the performance of an orchestra. The process of parsing the acoustic environment has been termed auditory scene analysis (ASA). ASA is thought to rely on the spectro-temporal patterns of the scene, and on internal heuristics (priors) employed by the auditory system. An important cue available to the system is spatial location. The processing of spatial cues is relatively well understood up to the level of auditory cortex. Considerably less is known about the neuronal processes underlying the role of spatial cues for ASA. Moreover, hearing-impaired listeners have substantial problems with ASA, as current hearing aids and cochlear implants are still not able to segregate complex acoustic mixtures. My project aims at uncovering the neuronal substrates for spatial streaming in auditory cortex of behaving Rhesus macaques and will thus (i) further our understanding of how spatial cues enable ASA in human listeners, and (ii) will lead to improved artificial speech-in-noise recognition algorithms, thereby directly improving ASA in the hearing impaired.