Adaptive Stochastic Resonance in Tinnitus Generation

Computational framework linking hearing loss, neural plasticity, and tinnitus with University of Erlangen and ENT Hospital Erlangen

Duration: October 2015 - November 2016
Type: Research Project
Institution: University of Erlangen
Partner: ENT Hospital Erlangen

Project Overview

Investigated the role of adaptive stochastic resonance as a biologically plausible mechanism underlying subjective tinnitus and auditory signal processing. The project developed and analyzed neural network models capable of estimating signal quality via autocorrelation, generating adaptive internal noise, and detecting performance gradients to form a complete feedback loop for stochastic resonance. Results demonstrated that simple neural circuits can dynamically optimize noise levels to enhance information transmission, supporting the hypothesis that increased neural activity observed after hearing loss may arise from an adaptive compensation mechanism. The work provides a computational framework linking hearing loss, neural plasticity, and tinnitus generation, while also highlighting potential applications in tinnitus treatment and next-generation hearing aids based on stochastic resonance principles.

Methodology

Neural Network Models

Developed neural network architectures capable of estimating signal quality via autocorrelation analysis, providing the foundation for adaptive noise generation.

Adaptive Noise Generation

Implemented mechanisms for generating adaptive internal noise that responds to signal quality and performance metrics.

Performance Gradient Detection

Created algorithms to detect performance gradients, enabling the system to form a complete feedback loop for stochastic resonance optimization.

Key Contributions

  • Adaptive Stochastic Resonance Framework: Investigated adaptive stochastic resonance as a biologically plausible mechanism underlying subjective tinnitus and auditory signal processing.
  • Neural Network Model Development: Developed and analyzed neural network models capable of estimating signal quality via autocorrelation.
  • Complete Feedback Loop: Created a system that generates adaptive internal noise and detects performance gradients to form a complete feedback loop for stochastic resonance.
  • Dynamic Noise Optimization: Demonstrated that simple neural circuits can dynamically optimize noise levels to enhance information transmission.
  • Tinnitus Hypothesis Support: Provided computational evidence supporting the hypothesis that increased neural activity observed after hearing loss may arise from an adaptive compensation mechanism.
  • Computational Framework: Developed a computational framework linking hearing loss, neural plasticity, and tinnitus generation.
  • Clinical Applications: Highlighted potential applications in tinnitus treatment and next-generation hearing aids based on stochastic resonance principles.

Technical Stack

Python Primary Language
C++ Programming Language
Neural Networks Machine Learning
Stochastic Processes Mathematical Modeling
Signal Processing Computational

Key Findings

Biological Plausibility

Established adaptive stochastic resonance as a biologically plausible mechanism for subjective tinnitus

Dynamic Optimization

Demonstrated that simple neural circuits can dynamically optimize noise levels to enhance information transmission

Hearing Loss Connection

Supported hypothesis that increased neural activity after hearing loss arises from adaptive compensation

Clinical Applications

Highlighted potential for tinnitus treatment and next-generation hearing aids

Related Skills

Computational Neuroscience Neural Network Modeling Stochastic Resonance Auditory Signal Processing Python Programming C++ Programming Biological Systems Scientific Computing Autocorrelation Analysis Feedback Loop Systems

Research Impact

This work provides a novel computational framework that bridges the gap between hearing loss, neural plasticity, and tinnitus generation. By demonstrating that adaptive stochastic resonance can emerge from simple neural circuits optimizing information transmission, the research offers both theoretical insights into tinnitus mechanisms and practical pathways for developing new treatment approaches and hearing aid technologies.

Related Publication

Patrick Krauss, Claus Metzner, Achim Schilling, Christian Schütz, Konstantin Tziridis, Ben Fabry, Holger Schulze (2017)

Adaptive Stochastic Resonance for Unknown and Variable Input Signals

Scientific Reports, 7(1), 2450

All sensors have a threshold, defined by the smallest signal amplitude that can be detected. The detection of sub-threshold signals, however, is possible by using the principle of stochastic resonance, where noise is added to the input signal so that it randomly exceeds the sensor threshold. The choice of an optimal noise level that maximizes the mutual information between sensor input and output, however, requires knowledge of the input signal, which is not available in most practical applications. Here we demonstrate that the autocorrelation of the sensor output alone is sufficient to find this optimal noise level. Furthermore, we demonstrate numerically and analytically the equivalence of the traditional mutual information approach and our autocorrelation approach for a range of model systems. We furthermore show how the level of added noise can be continuously adapted even to highly variable, unknown input signals via a feedback loop. Finally, we present evidence that adaptive stochastic resonance based on the autocorrelation of the sensor output may be a fundamental principle in neuronal systems.

PDF DOI Cite
Back to All Projects