A low-dimensional, generalizable encoding manifold for auditory cortex
- 1Oregon Health & Science University
Abstract
Neural populations in auditory cortex (AC) perform sensory computations that support stimulus category decoding and flexible behavior. The underlying geometry and generalizability of these computations for natural stimuli remain poorly understood, particularly at the single-neuron level, where neurons display wide-ranging tuning specificity and temporal acuity. To address this gap, we developed ACNet, a foundation model of cortical sound encoding, to predict the time-varying activity of >3000 neurons in AC of ferrets. Training data included 42 hours of natural sounds spanning over 100 categories and were collected from multiple recording sites across multiple animals. The model achieved state-of-the-art response prediction accuracy. Model activity was succinctly captured by a low-dimensional neural manifold, which generalized (>80% of variance) across animals. Analysis of ACNet activations revealed the emergence of rate-based, sparse sound coding across layers, a prominent feature of the auditory cortex. These transformations were concomitant with the emergence of more accurate and neurally aligned auditory category decoding, even though ACNet was not explicitly trained to categorize sounds. The model also revealed tuning differences between anatomically distinct cell types. Taken together, our results demonstrate that foundation models of sensory systems can reveal generalizable computations by large neural populations.
Related
- A new neural code for position: generalizable spatial maps in mouse medial frontal cortex
- The representation of order and randomness in the auditory cortex
- Altered stimulus-specific adaptation at the auditory cortex of a mouse model of Fragile X Syndrome
- Large cranial windows distort auditory cortical maps and degrade feature integration in secondary auditory cortex
- Visual-cortex-projecting neurons in mouse auditory cortex show distinct audiovisual responses independent of motor-related activity
BibTeX
@misc{parida2026lowdimensional,
title = {{A low-dimensional, generalizable encoding manifold for auditory cortex}},
author = {S. Parida and J. C. Wingert and J. D. Stickney and S. V. David},
howpublished = {bioRxiv},
year = {2026},
month = aug,
doi = {10.64898/2026.08.30.747962},
url = {https://www.biorxiv.org/content/10.64898/2026.08.30.747962v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - A low-dimensional, generalizable encoding manifold for auditory cortex
AU - S. Parida
AU - J. C. Wingert
AU - J. D. Stickney
AU - S. V. David
PY - 2026
DA - 2026/08/30
PB - bioRxiv
DO - 10.64898/2026.08.30.747962
UR - https://www.biorxiv.org/content/10.64898/2026.08.30.747962v1
AB - Neural populations in auditory cortex (AC) perform sensory computations that support stimulus category decoding and flexible behavior. The underlying geometry and generalizability of these computations for natural stimuli remain poorly understood, particularly at the single-neuron level, where neurons display wide-ranging tuning specificity and temporal acuity. To address this gap, we developed ACNet, a foundation model of cortical sound encoding, to predict the time-varying activity of >3000 neurons in AC of ferrets. Training data included 42 hours of natural sounds spanning over 100 categories and were collected from multiple recording sites across multiple animals. The model achieved state-of-the-art response prediction accuracy. Model activity was succinctly captured by a low-dimensional neural manifold, which generalized (>80% of variance) across animals. Analysis of ACNet activations revealed the emergence of rate-based, sparse sound coding across layers, a prominent feature of the auditory cortex. These transformations were concomitant with the emergence of more accurate and neurally aligned auditory category decoding, even though ACNet was not explicitly trained to categorize sounds. The model also revealed tuning differences between anatomically distinct cell types. Taken together, our results demonstrate that foundation models of sensory systems can reveal generalizable computations by large neural populations.
ER -