Proceedings of the Annual Meeting of the Cognitive Science Society

Multiple acoustic dimensions contribute to speechcategorization. Yet highly diagnostic dimensions contributegreater ‘perceptual weight’ in influencing speechcategorization than less diagnostic dimensions. Recentresearch demonstrates that perturbations in short-term inputregularities lead to rapid dynamic re-weighting of auditorydimensions. Here, we test the hypothesis that phonetic-category-level activation via a highly diagnostic acousticdimension is critical in driving this rapid tuning of how inputmaps to phonetic categories. To do so, we manipulate theinherent relative effectiveness, the perceptual weight, of twoacoustic dimensions in signaling English vowelcategorization using noise-vocoded versus clear speech. Weobserve that rapid tuning across statistical regularities isaffected by dimensions’ effectiveness in signaling the vowelcategories. These findings indicate that category activationvia a highly diagnostic dimension drives adaptive tuning inspeech perception, consistent with error-driven supervisedlearning.