Multiple timescales account for adaptive responses across sensory cortices

Kenneth W. Latimer, Dylan Barbera, Michael Sokoletsky, Bshara Awwad, Yonatan Katz, Israel Nelken, Ilan Lampl, Adrienne L. Fairhall, Nicholas J. PriebebioRxiv 2019 · DOI

Abstract

Sensory systems encounter remarkably diverse stimuli in the external environment. Natural stimuli exhibit timescales and amplitudes of variation that span a wide range. Mechanisms of adaptation, a ubiquitous feature of sensory systems, allow for the accommodation of this range of scales. This study investigates whether common rules of adaptation exist across different sensory modalities. By measuring membrane potential responses in the visual (V1), somatosensory (S1), and auditory (A1) cortices of mice to discrete stimuli at various frequencies, we find that adaptive profiles are largely preserved across these areas. These profiles include response attenuation and specific responses to the cessation of stimulation. We demonstrate that these complex adaptive behaviors can emerge from a simple model based on the integration of fixed filters operating over multiple timescales.

Introduction

Our sensory systems are constantly bombarded by stimuli that vary wildly in intensity and timing. To prevent sensory overload and maintain sensitivity, the brain "adapts," adjusting its response based on recent history. While adaptation is a well-documented phenomenon, two fundamental questions remain: Is adaptation universal across different sensory modalities, and is it a "dynamic" process requiring state changes (like synaptic weight adjustments) or an "emergent" property of fixed neural properties?

Historically, comparing adaptation across V1, S1, and A1 has been difficult due to disparate experimental paradigms. This research standardizes the approach to determine if these cortices share a common underlying "algorithm" for adaptation.

Methods

Researchers performed in vivo whole-cell recordings in adult C57BL/6 mice, measuring the membrane potential of individual neurons in the primary visual (V1, $n=11$), somatosensory (S1, $n=14$), and auditory (A1, $n=9$) cortices.

Stimuli were standardized as "punctate" pulses—brief light flashes, whisker deflections, or auditory clicks—delivered at fixed frequencies (1–20 Hz) and randomized (Poisson) intervals. To analyze these responses, the authors employed a Linear-Nonlinear (LN) subunit model. Each neuron was modeled as a sum of four independent "subunits," each characterized by a fixed linear filter and a nonlinearity. The goal was to see if a time-invariant model, trained on random noise, could predict frequency-dependent adaptation, thereby proving adaptation is an emergent property of static filters.

Results: Universal Adaptive Motifs

The study identified three specific "adaptive signatures" common to all three sensory modalities:

  1. Attenuation: Individual response amplitudes decrease as the stimulus frequency increases.
  2. Simplification: Low-frequency responses are complex and multiphasic, while high-frequency responses become simpler and monophasic.
  3. Termination Responses: A distinct "off-response" occurs following the cessation of a high-frequency stimulus train.
Membrane potential responses across V1, S1, and A1 show similar patterns of attenuation and termination responses as stimulus frequency increases.

The 4-subunit model demonstrated high predictive power across all modalities, explaining a significant portion of the variance (Corrected $R^2$: V1 = 0.56, S1 = 0.61, A1 = 0.68). Notably, the model successfully predicted the presence and long latency of termination responses using only fixed filters. This suggests that the brain utilizes a "bank" of filters spanning multiple timescales (from 60ms to over 2 seconds) to process natural stimuli.

While the motifs were universal, some modality-specific differences were observed. For instance, S1 was found to be more resilient to attenuation at 10 Hz compared to V1 or A1. Additionally, approximately 59% of neurons followed a logarithmic (Weber-like) response pattern when stimulus frequency changed abruptly, responding to relative rather than absolute changes.

Discussion

The finding that a fixed-filter model can account for complex adaptation suggests that the "state" of the cortex may not need to physically change to adapt. Instead, the integration of different temporal filters naturally creates the appearance of a shifting state.

However, several questions remain. It is unclear to what extent these adaptive effects are "inherited" from subcortical structures like the thalamus or peripheral receptors. Furthermore, while the model uses nonlinearities to mimic synaptic rectification, it does not explicitly account for the specific roles of inhibitory interneuron types (e.g., SST or PV cells). Future work should also consider calibrating stimulus intensities to be perceptually equal across senses to better compare attenuation levels.

Summary updated

Abstract

Abstract Sensory systems encounter remarkably diverse stimuli in the external environment. Natural stimuli exhibit timescales and amplitudes of variation that span a wide range. Mechanisms of adaptation, ubiquitous feature of sensory systems, allow for the accommodation of this range of scales. Are there common rules of adaptation across different sensory modalities? We measured the membrane potential responses of individual neurons in the visual, somatosensory and auditory cortices to discrete, punctate stimuli delivered at a wide range of fixed and nonfixed frequencies. We find that the adaptive profile of the response is largely preserved across these three areas, exhibiting attenuation and responses to the cessation of stimulation which are signatures of response to changes in stimulus statistics. We demonstrate that these adaptive responses can emerge from a simple model based on the integration of fixed filters operating over multiple time scales.
BibTeX
@misc{latimer2019multiple,
  author = {Kenneth W. Latimer and Dylan Barbera and Michael Sokoletsky and Bshara Awwad and Yonatan Katz and Israel Nelken and Ilan Lampl and Adrienne L. Fairhall and Nicholas J. Priebe},
  title = {{Multiple timescales account for adaptive responses across sensory cortices}},
  year = {2019},
  doi = {10.1101/700062},
  howpublished = {bioRxiv},
}
Figure