Channel-noise-induced stochastic facilitation in an auditory brainstem neuron model

Brett A. Schmerl, Mark D. McDonnell

Abstract

Neuronal membrane potentials fluctuate stochastically due to conductance changes caused by random transitions between the open and closed states of ion channels. Although it has previously been shown that channel noise can nontrivially affect neuronal dynamics, it is unknown whether ion-channel noise is strong enough to act as a noise source for hypothesized noise-enhanced information processing in real neuronal systems, i.e., "stochastic facilitation". Here we demonstrate that biophysical models of channel noise can give rise to two kinds of recently discovered stochastic facilitation effects in a Hodgkin-Huxley-like model of auditory brainstem neurons. The first, known as slope-based stochastic resonance (SBSR), enables phasic neurons to emit action potentials that can encode the slope of inputs that vary slowly relative to key time constants in the model. The second, known as inverse stochastic resonance (ISR), occurs in tonically firing neurons when small levels of noise inhibit tonic firing and replace it with burstlike dynamics. Consistent with previous work, we conclude that channel noise can provide significant variability in firing dynamics, even for large numbers of channels. Moreover, our results show that possible associated computational benefits may occur due to channel noise in neurons of the auditory brainstem. This holds whether the firing dynamics in the model are phasic (SBSR can occur due to channel noise) or tonic (ISR can occur due to channel noise).

Introduction

In classical computational neuroscience, neurons are frequently modeled using deterministic equations, such as the Hodgkin-Huxley model, where ion conductances are treated as smooth, averaged values. However, biological membranes are populated by discrete ion channels that transition randomly between open and closed states. This stochasticity creates "channel noise."

While researchers have long recognized that extrinsic noise (such as random synaptic input) can assist neurons in detecting weak signals—a phenomenon known as stochastic facilitation—it has remained an open question whether the intrinsic noise generated by the channels themselves is potent enough to drive these effects. This paper investigates whether biologically realistic levels of channel noise are sufficient to trigger two specific types of facilitation within the auditory brainstem, a region where temporal precision is critical for tasks like sound localization.

Methodology

The study utilizes the Rothman-Manis model, a biophysically grounded Hodgkin-Huxley-type model specifically designed to represent neurons in the ventral cochlear nucleus (VCN). The researchers examined three distinct physiological configurations:

  1. Phasic (Type II): Neurons that typically fire only at the onset of a stimulus.
  2. Tonic (Type I-II): Neurons capable of sustained firing.
  3. Integrator (Type I-c): A simpler firing model.

To simulate the stochastic nature of these cells, the authors compared two mathematical approaches. The first, Exact Markov Chain Simulations, tracks the state of every individual ion channel, providing a high-fidelity but computationally expensive baseline. The second, System Size Expansion (SSE), uses Stochastic Differential Equations (SDEs) to approximate the noise. This allowed the team to simulate realistic populations of up to millions of channels, matching the estimated densities found in real VCN neurons (approximately 25,000 potassium and 83,000 sodium channels).

Key Phenomena: SBSR and ISR

The research highlights two primary ways in which channel noise facilitates neuronal function:

Slope-Based Stochastic Resonance (SBSR)

Phasic neurons are naturally "slope-deaf"; they respond to the rapid onset of a signal but fail to fire when presented with slowly rising inputs. The authors discovered that intrinsic channel noise enables SBSR, allowing these neurons to fire in response to slow slopes they would otherwise ignore. This effectively transforms the neuron into a probabilistic slope detector, where the probability of a spike correlates with the steepness of the input ramp.

Inverse Stochastic Resonance (ISR)

In tonic neurons, the study observed a counter-intuitive effect: adding a small amount of noise can actually inhibit firing. This is known as Inverse Stochastic Resonance. In models with Class-II excitability (Type I-II), noise pushes the neuron from a steady firing state into a "resting" state or induces burst-like dynamics. This phenomenon relies on the mathematical "bistability" of the neuron—a state where the system can coexist in both a resting and a firing regime—which noise can exploit to switch the neuron's behavior.

Results and Discussion

The findings challenge the common assumption that intrinsic noise "averages out" as the number of channels increases. The authors demonstrate that even with $10^5$ channels, the noise remains strong enough to fundamentally alter dynamical behavior.

Key results include:

  • Validation of SSE: The System Size Expansion approximation was found to be highly accurate when compared to exact Markov simulations, validating it as a tool for large-scale biophysical modeling.
  • Optimal Noise Levels: For tonic models, the strongest inhibition of firing (the ISR effect) occurred at channel counts between $10^4$ and $10^5$, which aligns closely with biological estimates for auditory neurons.
  • Biological Realism: Because these effects emerge at realistic channel densities, it is highly probable that SBSR and ISR occur in vivo, suggesting that the auditory system may utilize intrinsic noise to enhance signal processing.

Conclusion and Future Directions

This study provides evidence that channel noise is a significant factor in neuronal dynamics, even in large neurons. By enabling slope detection in phasic cells and modulating firing rates in tonic cells, channel noise contributes to the computational repertoire of the auditory brainstem.

Future research may expand these findings by moving beyond "point neuron" models to multi-compartmental models that account for complex geometries like dendrites and axons. Additionally, further investigation is needed to quantify the exact information-theoretic gains—measuring the "bits per second"—that these noise-induced states provide to the auditory pathway.

Summary updated

Abstract

Neuronal membrane potentials fluctuate stochastically due to conductance changes caused by random transitions between the open and closed states of ion channels. Although it has previously been shown that channel noise can nontrivially affect neuronal dynamics, it is unknown whether ion-channel noise is strong enough to act as a noise source for hypothesized noise-enhanced information processing in real neuronal systems, i.e., "stochastic facilitation". Here we demonstrate that biophysical models of channel noise can give rise to two kinds of recently discovered stochastic facilitation effects in a Hodgkin-Huxley-like model of auditory brainstem neurons. The first, known as slope-based stochastic resonance (SBSR), enables phasic neurons to emit action potentials that can encode the slope of inputs that vary slowly relative to key time constants in the model. The second, known as inverse stochastic resonance (ISR), occurs in tonically firing neurons when small levels of noise inhibit tonic firing and replace it with burstlike dynamics. Consistent with previous work, we conclude that channel noise can provide significant variability in firing dynamics, even for large numbers of channels. Moreover, our results show that possible associated computational benefits may occur due to channel noise in neurons of the auditory brainstem. This holds whether the firing dynamics in the model are phasic (SBSR can occur due to channel noise) or tonic (ISR can occur due to channel noise).
BibTeX
@inproceedings{schmerl2013channel,
  author = {Brett A. Schmerl and Mark D. McDonnell},
  title = {{Channel-noise-induced stochastic facilitation in an auditory brainstem neuron model}},
  booktitle = {Physical},
  year = {2013},
  doi = {10.1103/physreve.88.052722},
}
Figure