Combining Sampling Methods with Attractor Dynamics in Spiking Models of Head-Direction Systems.

Area of Science:

• Computational Neuroscience
• Systems Neuroscience
• Neural Dynamics

Background:

• The brain must infer and integrate noisy environmental signals to guide behavior effectively, a process complicated by inherent uncertainty.
• Sampling-based inference is a proposed mechanism for handling uncertainty, particularly in early sensory processing.
• Reconciling sampling-based methods with higher-order brain area dynamics, like attractor dynamics, remains a challenge.

Purpose of the Study:

• To present a spiking neural network model for the head-direction (HD) system that unifies sampling-based inference with attractor dynamics.
• To derive the necessary spiking neural network dynamics for sampling from diverse probability distributions, including those with Poisson noise.
• To propose a method for updating head direction estimates by integrating angular velocity samples with attractor dynamics.

Main Methods:

• Developed a spiking neural network model for the head-direction system.
• Derived spiking neural network dynamics to perform sampling from probability distributions with Poisson noise.
• Integrated angular velocity samples with a circular manifold to maintain attractor dynamics for head direction estimation.

Main Results:

• The model successfully combines sampling-based inference with attractor dynamics in the HD system.
• It generates specific, testable predictions for neurophysiological experiments, including correlated voltage fluctuations and firing patterns.
• Predicted characteristic statistics for the movement of the neural activity 'bump' representing head direction.

Conclusions:

• The study extends theories of probabilistic sampling with spiking neurons.
• It offers a new perspective on neural computations for orientation and navigation.
• Supports the integration of sampling-based methods and attractor dynamics as a framework for neural dynamics research.