A robust balancing mechanism for spiking neural networks

Area of Science:

• Computational Neuroscience
• Neural Dynamics
• Synaptic Plasticity

Background:

• The irregular, low firing rates in the cortex are typically explained by a balance between excitation and inhibition.
• Understanding the mechanisms maintaining this balance is crucial for comprehending neural computation.

Purpose of the Study:

• To propose and validate a novel nonlinear mechanism for achieving a balanced state in random spiking neural networks.
• To investigate the role of synaptic plasticity, specifically short-term depression, in this balancing mechanism.
• To demonstrate the robustness of this mechanism, even without significant external currents.

Main Methods:

• Developed a self-consistent mathematical analysis to model the neural network dynamics.
• Conducted extensive simulations on networks of increasing sizes to verify the theoretical findings.
• Focused on the nonlinear response of synaptic activity due to short-term depression.

Main Results:

• Identified a robust nonlinear balancing mechanism driven by synapse plasticity (short-term depression).
• Demonstrated the emergence of a stable balanced regime in spiking neural networks.
• Confirmed that the mechanism operates effectively even without strong external currents.
• Observed that the balanced state is fluctuation-driven, leading to highly irregular spiking in all neurons.

Conclusions:

• Short-term depression in excitatory-excitatory synapses provides a robust nonlinear mechanism for balancing excitation and inhibition in neural networks.
• This mechanism can explain the irregular spiking dynamics observed in the cortex.
• The findings highlight the importance of intrinsic network properties and synaptic plasticity in maintaining cortical activity patterns.