Abstract
Accents are ubiquitous in spoken communication, and while listeners can rapidly adapt to accented speech, the neural mechanisms supporting this flexibility remain poorly understood. Successful adaptation requires developing new sound representations without compromising the stability of long-term speech norms. This delicate balance between plasticity and stability illustrates a fundamental challenge faced by all cognitive systems. To investigate how the brain manages this trade-off, we recorded electroencephalographic activity from 23 native English speakers as they categorized words produced in either canonical American English or an unfamiliar accent. We contrasted two potential mechanisms: one in which listeners fully restructure their sound-to-category mappings to reflect accent-specific pronunciations, and another in which they downweight the functional relevance of sounds that deviate from long-term expectations. Listeners relied on short-term speech regularities to reduce perceptual weighting of acoustic dimensions that did not conform to the canonical norm. Consistent with this perceptual shift, we observed less robust neural encoding of sound differences along the downweighted dimensions. Notably, these adaptive neural adjustments emerged as early as 100 milliseconds, at latencies associated with subphonemic auditory processing, and persisted through later stages linked to phonological and post-phonological processing. These findings indicate that rapid adaptation to unfamiliar accents involves downweighting the functional relevance of sound cues based on short-term input statistics, rather than fully restructuring native sound-to-category mappings. This mechanism enables flexible adjustment to novel speech inputs while preserving long-term linguistic representations, illustrating how the auditory system negotiates the trade-off between plasticity and representational stability.