Multilevel Auditory Processing of Continuous Speech, from Acoustics to Language

点击翻译按钮获取中文摘要

PROJECT SUMMARY For continuous speech to be perceived as intelligible, the brain employs an extended series of representations along the auditory pathway, emphasizing different temporal features of speech at each stage. From brainstem to midbrain, thalamus, and through multiple cortical areas, these stages support representations of fast pitch- relevant periodicity, slower representations corresponding to vocal-tract/envelope modulations, and ultimately, linguistic and semantic representations. Furthermore, the strength of these representations is not determined solely in a feedforward way, but also by top-down processes, modulated by attention and listening effort. It is well known that degradation of a speech signal will interfere with these neural representations in different ways: some losses are irreparable, but others can be compensated for at a later stage. A gap in knowledge arises because the different neural stages, and the corresponding continuous speech representations, are rarely studied as a longitudinal chain. Partial failure at one processing level might be compensated for at a subsequent level, and the level of compensation might be associated with the level of listening effort. The specific objectives of this application are to determine how continuous speech is progressively represented along and beyond the auditory pathway, from midbrain to language areas, in young normal- hearing listeners. Both electroencephalography (EEG) and magnetoencephalography (MEG) will be employed for their high time resolution, simultaneously recording from subcortical and multiple cortical areas respectively, along with measures of behavior, and listening effort. The speech representations of different neural stages will be quantified by specified measures of time-locked neural processing and of neural connectivity. Pupillometry measures will be used as a physiological proxy for listening effort, in addition to self- reported effort measures. Our central hypothesis is that a grounded understanding of the progression of representations of continuous speech processing, viewed as a network with both feedforward and feedback connectivity (and including task-based and effort-linked connectivity changes), will elucidate the acoustic and neural conditions under which continuous speech is ultimately perceived as intelligible. Aim 1 investigates a broad set of neural measures of continuous speech processing, simultaneously obtained along and beyond the auditory pathway, and the extent to which they can predict intelligibility. Aim 2 investigates the extent to which also incorporating measures of sustained listening effort, in concert with the above neural measures of continuous speech processing, may allow better predictions of intelligibility than the neural measures alone. Aim 3 investigates the functional role of neural connectivity in supporting continuous speech processing along the auditory pathway and beyond.