Strategies Underlying Auditory Categorization
Degree type
Graduate group
Discipline
Psychiatry and Psychology
Subject
Behavior
Categorization
Cortex
Electrophysiology
Learning
Funder
Grant number
License
Copyright date
Distributor
Related resources
Author
Contributor
Abstract
Categorization allows us to flexibly and efficiently respond to a complex and variable sensory world by grouping continuous stimuli into discrete, behaviorally meaningful categories. While this process has traditionally been studied in trained subjects and high-order brain areas, less is known about how categorization unfolds across learning and how it is implemented in early sensory regions such as the auditory cortex (ACx). In this thesis, we combine psychophysical modeling, high-throughput behavioral paradigms, and large-scale neuronal recordings to investigate how behavioral strategies and neuronal activity shape auditory categorization in mice. In Chapter 2, we show that subjects integrate multiple sources of uncertainty during categorization, including sensory reliability and stimulus relevance, and that a Bayesian model can capture both the structure and variability of this integration. Despite being trained on the same task, individual subjects exhibited distinct strategies for weighing uncertainty sources, highlighting the need to understand decision-making at the level of individual behavior. In Chapter 3, we investigate how categorization strategies emerge and evolve during learning. Using a wheel-based auditory task, we find that patterns in early training predict generalization behavior and that subjects vary in how they acquire category boundaries. Inactivation of the ACx during behavior impairs performance, supporting an active role for the ACx in auditory categorization. In Chapter 4, we introduce a novel experimental platform that combines a lick-based behavioral task with chronically implanted Neuropixels 2.0 probes, enabling simultaneous, long-term recordings from large populations of neurons across the depth of the ACx. We show that population activity in ACx tracks behaviorally defined category boundaries, and that putative inhibitory interneurons are more likely to encode category-related information than excitatory neurons. Together, these findings highlight the importance of individual learning strategies in shaping perceptual categorization and suggest that the ACx not only encodes low-level sensory features, but also participates in the representation of categories. This work establishes a new framework for investigating how internal models and neuronal circuits jointly contribute to perceptual decision-making.