Listeners form average-based representations of individual voice identities

Abstract

People can learn to identify a person based on their voice, despite variation in their voice. Here, the authors show that this ability relies on a statistical abstraction mechanism during which people form average-based representations of voices, even without prior exposure to the average.