Event-driven simulation of spiking neurons with stochastic dynamics

Area of Science:

• Computational Neuroscience
• Computational Biology
• Systems Neuroscience

Background:

• Traditional neural network simulations often struggle with large-scale networks and inherent neuronal noise.
• Event-driven strategies efficiently simulate neuronal interactions by focusing on discrete events like action potentials.
• Existing methods primarily handle deterministic dynamics between events.

Purpose of the Study:

• To extend the event-driven simulation strategy to incorporate stochastic dynamics between events.
• To enable efficient simulation of large neural networks with inherent noise and massive external inputs.
• To provide a method for studying neuronal dynamics under random current injection from large background populations.

Main Methods:

• Extension of the event-driven simulation strategy to handle stochastic state variable dynamics.
• Numerical evaluation of statistical properties of single neurons under random current injection.
• Incorporation of the impact of large background neuronal populations into simulations.

Main Results:

• The developed strategy effectively simulates neural networks with stochastic inter-event dynamics.
• It accurately captures the effects of inherent neuronal noise and large-scale synaptic input.
• The method allows for the study of networks with arbitrary external afferents and stochastic properties.

Conclusions:

• The enhanced event-driven simulation strategy offers a powerful tool for computational neuroscience.
• It facilitates more realistic and efficient simulations of complex neural systems.
• This approach advances the study of neural network dynamics, noise, and large population effects.