Statistical learning in the auditory modality: Identities are all you need

Finding a behaviorally relevant target among non-targets is facilitated by prior knowledge about the statistical structure of the environment. While this contextual-cueing effect is well established in vision, it is less clear whether and how context-based facilitation is realized in the auditory modality. In five experiments with 136 participants, we employed a new auditory search task in which constant (vs. variable) human speech sounds (letters as distractors, numbers as targets) were presented sequentially from distinct spatial locations using five loudspeakers. We find that distractor identity is both necessary and sufficient for auditory contextual cueing. This finding stands in striking contrast to visual contextual cueing, where fixed spatial layouts are typically crucial for the effect. Thus, it highlights the complementary strengths of the two modalities: While the visual system prioritizes spatial context, the auditory system rapidly recodes input into categorical, identity-based representations, making these central to statistical contextual learning.