Hierarchical Encoding of Millisecond-Scale Temporal Configuration of Sound in the Mammalian Auditory Pathway

Yuying Zhai, Xinyu Du, Peirun Song, Hai-liang Xu, Hangting Ye, Xuehui Bao, Irfan Mehmood, Hisashi Tanigawa, Wanqiu Niu, Zhiyi Tu, Pei Chen, Tingting Zhang, Lingling Zhang, Xuan Zhao, Xiongjie Yu, David Pérez-González, Manuel S. MalmiercabioRxiv 2024 · DOI

TL;DR

This paper reveals that the precise millisecond-scale timing of sounds, not just their pitch or loudness, is a fundamental auditory feature. This "temporal configuration" is processed hierarchically in the brain, with higher-order areas like the auditory cortex playing a crucial role, and is conserved across humans and rats, suggesting its evolutionary importance.

What Problem Does This Paper Address?

While we understand how the brain processes basic sound features like pitch and loudness, how it deciphers the subtle, millisecond-level timing patterns within sounds – their "temporal configuration" – remains less clear. This is crucial because natural sounds, from speech to music to environmental noises, rely heavily on these fine temporal structures for their identity and meaning. The authors investigate whether this temporal configuration is treated as a distinct auditory feature, how it's represented across the auditory pathway, and if this processing is conserved across species.

Methods

The researchers employed a multi-pronged approach:

  • Behavioral Discrimination Tasks: Humans and rats were tested on their ability to distinguish between click trains that had identical average inter-click intervals (ICIs) but different temporal patterns. This was done using a delayed match-to-sample task.
  • Electrophysiology (EEG and ECoG): Electroencephalography (EEG) in humans and electrocorticography (ECoG) in rats were used to record brain activity in response to these click trains, specifically looking for mismatch negativity (MMN) responses, an indicator of automatic deviance detection.
  • Neuropixels Recordings: High-density Neuropixels probes were used to record neural activity from individual neurons across different stages of the auditory pathway in rats (Inferior Colliculus (IC), Medial Geniculate Body (MGB), and Primary Auditory Cortex (A1)). This allowed them to map where sensitivity to temporal configuration emerges.
  • Auditory Cortex Cooling: To investigate the role of top-down feedback, the auditory cortex was reversibly cooled to temporarily deactivate it and observe the effect on MGB activity.
  • Laminar Analysis: Within the auditory cortex, recordings were analyzed layer-by-layer to determine which cortical layers are most involved in processing temporal configuration.

Datasets

Not explicitly discussed as "datasets" in the traditional machine learning sense. The study generated its own auditory stimuli (click trains with varying temporal configurations) and recorded neural and behavioral responses from human and rat subjects.

Key Insights and Ideas

  • Temporal Configuration is a Distinct Auditory Feature: The study strongly suggests that the precise millisecond-scale ordering of sound events is not just a byproduct of other features but a fundamental perceptual dimension, akin to pitch or loudness.
  • Hierarchical Processing: Sensitivity to temporal configuration isn't present from the start. It emerges gradually along the auditory pathway, with minimal sensitivity in the IC, intermediate in the MGB, and strong processing in the auditory cortex (A1).
  • Corticothalamic Feedback is Crucial: The auditory cortex actively shapes how the thalamus (MGB) represents temporal information, highlighting the importance of top-down feedback loops.
  • Cross-Species Conservation: The fact that both humans and rats show similar MMN responses to changes in temporal configuration suggests this is a fundamental, evolutionarily conserved auditory processing mechanism.
  • Superficial Cortical Layers Lead: Within the auditory cortex, the superficial layers (supragranular) seem to play a more significant role in encoding temporal configuration compared to deeper layers.

Key Results

  • Humans: Could reliably discriminate click trains differing only in temporal configuration, even when average ICIs and overall energy were matched. These differences also elicited robust MMN responses.
  • Rats: Showed analogous MMN-like ECoG responses to changes in temporal configuration, demonstrating cross-species generality.
  • Neuronal Recordings (Rats):
    • IC: Minimal configuration sensitivity.
  • MGB: Intermediate configuration sensitivity.
  • A1: Strong stimulus-specific adaptation to temporal configuration.
  • Auditory Cortex Cooling: Reduced configuration sensitivity in the MGB, confirming the role of corticothalamic feedback.
  • Laminar Analysis: Supragranular layers of A1 showed stronger adaptation for temporal configuration than infragranular layers.

Limitations and Future Work

  • Limitations: The paper primarily uses click trains, which are simplified stimuli. While useful for isolating temporal configuration, they don't fully represent the complexity of natural sounds. The study acknowledges that the low CSI in the IC for click trains might reflect a specific difficulty in discriminating temporal configurations rather than a general lack of stimulus-specific adaptation (SSA) capacity.
  • Future Work: The authors suggest further research to explore the specific synaptic and circuit mechanisms underlying this temporal processing, such as recurrent connectivity and inhibitory microcircuits. They also emphasize the importance of relating neuronal sensitivity to perceptual thresholds and biases in humans to solidify temporal configuration as a feature-like dimension of auditory representation.

Summary updated

Abstract

Abstract Natural sounds are defined not only by their spectral content but also by their fine temporal structure. Here we show that millisecond-scale temporal configuration—defined by the ordering of inter-click intervals (ICIs) within a click train—behaves as a distinct auditory feature, conserved across species and emerging hierarchically along the auditory pathway. Human listeners reliably discriminated click trains that shared the same average ICI but differed in temporal configuration, and these differences elicited robust mismatch negativity (MMN) responses in an oddball paradigm, indicating automatic cortical deviance detection. Awake rats showed analogous MMN-like ECoG responses to configuration changes, demonstrating cross-species generality. Neuropixels recordings along the inferior colliculus–medial geniculate body–primary auditory cortex (IC–MGB–A1) axis revealed minimal configuration sensitivity in IC, intermediate sensitivity in MGB, and strong stimulus-specific adaptation to temporal configuration in A1. Reversible cooling of auditory cortex reduced configuration sensitivity in MGB, implicating corticothalamic feedback in shaping thalamic representations. Layer-resolved analyses further showed that supragranular A1 neurons carry stronger configuration-specific adaptation than infragranular neurons. These findings identify temporal configuration as a feature-like dimension of sound and delineate a hierarchical, feedback-dependent IC–MGB–A1 circuit architecture for encoding fine temporal structure in the mammalian brain.
BibTeX
@misc{zhai2024hierarchical,
  author = {Yuying Zhai and Xinyu Du and Peirun Song and Hai-liang Xu and Hangting Ye and Xuehui Bao and Irfan Mehmood and Hisashi Tanigawa and Wanqiu Niu and Zhiyi Tu and Pei Chen and Tingting Zhang and Lingling Zhang and Xuan Zhao and Xiongjie Yu and David Pérez-González and Manuel S. Malmierca and Xiongjie Yu},
  title = {{Hierarchical Encoding of Millisecond-Scale Temporal Configuration of Sound in the Mammalian Auditory Pathway}},
  year = {2024},
  doi = {10.1101/2024.12.02.626513},
  howpublished = {bioRxiv},
}
Figure