Nonequilibrium Statistical Mechanics of Continuous Attractors

Area of Science:

• Neuroscience
• Computational Neuroscience
• Theoretical Neuroscience

Background:

• Continuous attractor networks (CANs) model persistent neural activity encoding sensory information like head direction and spatial location.
• Understanding how external signals rapidly update these internal representations in CANs is crucial but not well understood.

Purpose of the Study:

• To determine the fundamental limits on the rate at which external signals can update internal representations in CANs.
• To apply these findings to place cell networks and derive a measure of memory capacity.

Main Methods:

• Theoretical analysis of continuous attractor network dynamics.
• Derivation of update speed limits based on network parameters.
• Application of derived limits to biological neural networks, specifically place cells.

Main Results:

• Identified fundamental constraints on the speed of updating internal representations within CANs.
• Derived a velocity-dependent nonequilibrium memory capacity for neural networks.
• Demonstrated how spatial and time-dependent external signals interact with attractor dynamics.

Conclusions:

• The study establishes theoretical limits for rapid representation updates in neural networks with continuous attractors.
• Introduced a novel metric, velocity-dependent nonequilibrium memory capacity, applicable to biological systems like place cell networks.
• Provides a framework for understanding the dynamic manipulation of neural representations by external inputs.