Auditory perceptual maps in humans and mice share common structures and predict perceptual decisions in discrimination learning

Abstract

To efficiently perceive sensory information and guide behavior, the brain organizes incoming sensory stimuli into internal maps that capture perceptual relatedness between stimuli. Whether these maps, typically assessed in scaling paradigms without feedback, also shape perceptual decisions during reinforcement-based conditions remains unclear. Here, we assess task-naïve perceptual maps from similarity judgments of pulsed sound stimuli, and compare them to perceptual maps obtained from multiple discrimination tasks in humans (n = 152) and mice (n = 11). We find that task-naïve maps predict how well humans discriminate sounds, how quickly they learn, and which stimulus features guide their choices. Moreover, naïve and task-based maps share key structures, suggesting a stable perceptual map architecture throughout learning. Remarkably, human task-based and task-naïve perceptual maps share key features and structures with the task-based perceptual maps of mice, indicating congruent structures of auditory perception across species. Together, our results indicate that perception relies on robust internal maps that provide a common framework to flexibly guide behavior in changing environments.