Neurophysiological Mechanisms of Speech Intelligibility in Noise - A Quantitative Framework

点击翻译按钮获取中文摘要

Project Summary/Abstract Communicating in noisy environments with many sound sources places enormous demands on the auditory system. Successfully extracting target speech information in such scenarios requires a combination of precise cochlear transduction and neural coding of sound information, and effective downstream cognitive processes that use the encoded information to segregate and selectively process the target speech of interest. Speech-in- noise problems (e.g., in old age or hearing loss) can thus arise from impaired “bottom-up” coding of information, or from declines in cognitive ability. Although the existence of these two components is well recognized, there is currently no integrative framework for quantifying the relative contributions of each to speech intelligibility in noise, and for identifying which aspect is deficient in an individual who is experiencing listening problems. The specific aims of this proposal are designed to address this gap by measuring the neurophysiological representation of envelope information and investigating how that varies with both “bottom- up” manipulations and “top-down” manipulations in the same individuals. First, envelope coding in the brainstem and cortex will be directly measured using electroencephalography (EEG). The EEG metrics will then be linked to perceptual intelligibility by examining how they covary as the speech is presented with different levels and types of background noise, and how they covary across individuals (Aim 1). Next, for the same individuals, the effect of attentional focus on the same envelope coding measures as Aim 1 will be examined by keeping the input speech mixture constant and manipulating which speech source is the focus of selective attention (Aim 2). This approach helps isolate the top-down component. Importantly, conducting all intelligibility and EEG measurements in the same individual subjects allows us to leverage individual differences and use regression techniques to characterize the relative contributions of bottom-up and top-down mechanisms to performance. Finally, for the speech-noise mixtures used in Aim 1, we will calculate the same envelope coding metrics at the output of a computational auditory-nerve model (Aim 3). Key mechanisms of cochlear dysfunction will be incorporated into the model to characterize their effects on envelope coding of speech in noise. By comparing “model” (Aim 3) and “neural” metrics (Aims 1 and 2), we will test whether cochlear mechanisms can account for the individual differences in bottom-up coding. The knowledge gained through the proposed set of experiments will be foundational in the development of objective diagnostics and interventions tailored for the specific nature of speech-in-noise problems that an individual patient is experiencing. Project completion will also provide the applicant with training in computational modeling, psychophysical and EEG experiment design, data collection, analysis, and interpretation, as well as scientific hypothesis testing. This complements her existing background in signal processing and statistics, and will set her on a solid path towards her long-term goal of an academic research career in auditory neuroscience.