Neural and behavioral variability in auditory perceptual learning

K A Martin, C Bredenberg, E P Simoncelli, C Savin and R C Froemke

Published in Annual Meeting, Neuroscience, Nov 2021.
Perceptual learning is associated with altered cortical representations of task-relevant stimuli in trained animals relative to naive ones. Despite their association with perceptual improvement, such changes in neural tuning are typically not measured throughout learning. Additionally, there is a large degree of variability in perceptual learning rates across animals, which is largely ignored when recording and interpreting neural activity. To address these limitations, we developed an experimental and computational framework for describing how sensory representations change during auditory perceptual learning.

We recorded from a population of neurons throughout auditory conditioning using two-photon imaging of layer 2/3 excitatory neurons in the auditory cortex of mice. To obtain water reward, mice learned to to distinguish a single, center frequency (11-16 kHz) from non-center frequencies by licking left and right, respectively. Discrimination performance improved over 10-45 days through multiple phases of learning (N=52 animals).

Across trained animals, despite similar trained behavioral performance, we observed two distinct tuning profiles of excitatory neurons. Specifically, individuals exhibited either a relative enhancement or suppression of the frequencies reported as center. Despite the variability in neural responses, there was a consistent increase in number of neurons with a maximum response a quarter octave away from the behavior discrimination threshold (N=10 animals). Additionally, we found these neurons had the steepest slope at the behavior discrimination threshold (N=10 animals). Thus, we found a relative increase of neurons that are most sensitive to detecting minor differences in frequency around the behavior discrimination threshold, potentially facilitating improvement in the discrimination between frequencies that are similar but are classified differently behaviorally.

To make sense of the across animal variability in tuning, we trained a network model by reward-modulated Hebbian synaptic learning (Williams, 1992) to learn the same task. We found that the simulated network learns at similar rates as real animals. The model captures the across-animal variability in tuning, specifically capturing the suppression or enhancement of center frequency representations. Overall, both data and model reveal nontrivial learning dynamics associated with perceptual learning in auditory cortex, with initial tuning driving across-animal variability in the emerging representations.

  • Listing of all publications