Neural and Cognitive Mechanisms of Flexible Speech Perception

Abstract

Speech perception requires listeners to derive stable linguistic representations from acoustically variable and often ambiguous input. A central question in cognitive neuroscience concerns how the brain achieves this stability while remaining flexible to context and task demands. This dissertation investigates the neurophysiological correlates of speech processing, focusing on how listeners form and apply phonetic categories across behavioral and neural levels of analysis. Across three experimental studies, participants perceived speech sounds across different tasks and contexts while behavioral responses and electrophysiological measures were recorded. The findings reveal reliable individual differences in phonetic categorization that generalize across tasks and phonetic contexts, suggesting stable perceptual strategies rather than purely task-driven effects. At the neural level, auditory-evoked potentials demonstrate that categorical structure emerges early in processing and is dynamically modulated by attention-invariant lexical cues and listener-specific perceptual biases. Together, these results support a model in which speech perception reflects a specialized yet flexible system that organizes acoustic input around probabilistic category structure modulated through top-down connectivity. Rather than serving as fixed boundaries, categorical divisions represent dynamic neural activity that integrates sensory evidence with internal representations and experience. By linking behavioral variability to neural dynamics, this dissertation advances our understanding of how categorical perception is characterized and clarifies the mechanisms through which the brain balances stability and flexibility in spoken language comprehension.