Neural Adaptation Dynamics

Modeling Perceptual Bistability:
Impact of Neuronal Adaptation and Noise on Alternation Dynamics

Project Report | Code

This project was conducted as part of the [NEU 466M] Quantitative Methods in Neuroscience course, taught by Professor Thibaud Taillefumier during the Spring 2024 semester at The University of Texas at Austin.

Abstract

Perceptual bistability, characterized by stochastic alternations between two neuronal population activities, reflects the brain's processing of information that can be perceived in two distinct ways. This study investigates the dynamics of a proposed attractor network model, which identifies noise as the primary driver of these alternations. We reconstructed the rate-based version of this model using differential equations and the Euler method to systematically explore the nuanced impacts of noise, adaptation, and excitatory inputs on the model’s behavior. Our results reveal that adaptation and noise critically affect the duration and distribution of temporal intervals between state transitions, profoundly influencing the model's ability to mimic experimental observations. Furthermore, we found that careful parameter selection is crucial in constructing network models that accurately capture perceptual bistability. This research not only underscores the crucial roles of noise and adaptation in the dynamics of alternation but also highlights the challenges inherent in simulating complex biological phenomena through computational models.