Neural Dynamics Underlying the Emergence of Auditory Categorization and Learning

项目摘要 人类对世界的成功感知需要大脑将各种感觉信息组合在一起 这一过程被称为分类知觉（CP）。在其核心，CP是已知的 作为“不变性”或“多对一映射”问题：感官特征的无限集合必须是 被转换成有限的、不变的感知空间以由感知系统作用。分类 表现在人类认知和学习的几乎所有方面，包括对面部，颜色和 音乐.熟练的分类在口语和书面语中尤为重要， 通过其在阅读获得和基于记忆的学习障碍中的整体作用（例如，特殊阅读障碍 语言障碍）。尽管有大量的行为研究及其对理解接受性的重要性， 人类交流的核心能力，CP的神经机制仍然知之甚少。在 通过一系列的研究，这项工作将解决大脑在何时、何地以及如何 将连续的声学信号转换为感知系统所利用的离散的、有意义的类别。 高密度神经电脑记录（EEG/ERP）将在任务期间从人类听众获得 旨在挖掘分类处理的不同属性并调节其神经生物学。我们的中央 假设是听觉分类技能招募一个共同的，简约的额颞神经网络 它是动态的，有差异的，取决于注意力，刺激的熟悉程度， 上下文/复杂性、学习和先前的听力经验。新的多元分析技术将被使用 从听者的ERP中推导出“神经测量功能”，以“解码”听者的言语感知行为， 他们潜在的大脑活动这种常见的神经计算方法将用于研究几个 通过五个研究目标，调节听觉分类技能的因素：（目标1）时空 大脑中CP的出现;（目标2）线性与非线性信号动力学;（目标3）识别声音 不同的域（例如，（目标4）先前的听力经验和新的学习。目标5将 从EEG记录中测量功能连接，以确定内部信息的定向流动 CP脑网络随着先前目标的操纵而改变（例如，学习与处理成熟 类别）。提供了一个更完整的生物学描述的声学语音映射问题， CP最终将提供一个窗口，不仅正常的言语知觉，但可能揭示重要的神经 机制，以针对障碍，损害形成的听觉类别。

Project Summary Successful perception of the world requires that the human brain assemble diverse sensory information into common, well-formed groupings, a process known as categorical perception (CP). At its core, CP is known as the “invariance-” or “many-to-one mapping” problem: an infinite collection of sensory features must be converted into a finite, invariant perceptual space to be acted upon by the perceptual system. Categorization manifests in nearly all aspects of human cognition and learning including the perception of faces, colors, and music. Skilled categorization is particularly important in the context of spoken and written language as evident by its integral role in reading acquisition and auditory-based learning disorders (e.g., dyslexia, specific language impairment). Despite a wealth of behavioral studies and its importance to understanding receptive human communication, the neural mechanisms underlying the core ability of CP remain poorly understood. In a series of studies, the proposed work will address foundational questions of when, where, and how the brain converts continuous acoustic signals into discrete, meaningful categories exploited by the perceptual system. High-density neuroelectric brain recordings (EEG/ERP) will be obtained from human listeners during tasks designed to tap different attributes of categorical processing and modulate its neurobiology. Our central hypothesis is that auditory categorization skills recruit a common, parsimonious frontotemporal neural network that is both dynamically and differentially engaged depending on attention, familiarity of stimulus context/complexity, learning, and prior listening experience. Novel multivariate analytic techniques will be used to derive “neurometric functions” from listeners’ ERPs to “decode” listeners’ speech perception behaviors from their underlying brain activity. This common neurocomputational approach will be used to investigate several factors that modulate auditory categorization skills through five research aims: (Aim 1) the spatiotemporal emergence of CP in the brain; (Aim 2) linear vs. nonlinear signal dynamics; (Aim 3) identifying sounds from different domains (e.g., speech vs. music); (Aim 4) prior listening experience and novel learning. Aim 5 will measure functional connectivity from EEG recordings to determine how the directed flow of information within the CP brain network changes with the manipulations of prior aims (e.g., learning vs. processing mature categories). Providing a more complete biological description of the acoustic-to-phonetic mapping problem of CP will ultimately offer a window into not only normal speech perception but may reveal important neural mechanisms to target in disorders that impair the formation of auditory categories.