PROJECT SUMMARY (See instructions): Learning is a fundamental component of neural systems. It is essential for gaining new knowledge or skills, and impaired learning faculties can have major deleterious effects. Although various forms of associative and perceptual learning have been studied our current understanding of how neural circuits learn remains incomplete. Humans and animals leverage category learning––grouping stimuli together based on shared and often higher order features––to deal with the world’s dazzling complexity. Yet, the neural mechanisms underlying category learning are not known. In this proposal, we focus on auditory category learning leveraging the rich nature and dynamic structure of natural and synthetic soundscapes. Auditory category learning is ubiquitous across the animal kingdom, with vivid examples from invertebrates to humans, and is intimately related to the transformation of sound to percepts. However, mechanistic studies of category learning in audition are not yet mature. In mice, where powerful experimental toolkits exist and new tools are continuously developed, studies of auditory category learning are scarce. Here we will study the multi-regional basis of reshaping of population activity following learning. We will analyze a unique dataset of simultaneous population recordings across brain areas in mice learning an auditory categorization task. We will apply novel computational methods to study how cortical auditory circuits process sounds and how this processing changes following learning. We will explore multiple types of trained categorization as well as responses to categories of natural sounds. We will characterize changes both at the level of single neurons and at the level of neuronal populations. Through simultaneous recordings across brain areas we will delineate changes both at the level of individual brain areas and at the level of interactions between brain areas. Finally, we will use perturbations to improve our causal understanding of the circuitry. These experiments and analyses will allow us to explore specific hypotheses regarding the plasticity and circuit computations of cortical circuits in the mammalian brain, and how these support category learning.