Sparse balance: Excitatory-inhibitory networks with small bias currents and broadly distributed synaptic weights

Area of Science:

• Computational Neuroscience
• Neural Circuit Dynamics
• Systems Neuroscience

Background:

• Cortical circuits require precise excitatory-inhibitory (E-I) balance for stability.
• Standard models necessitate strong feedforward bias currents for spontaneous irregular activity, lacking experimental support.

Purpose of the Study:

• To investigate an alternative E-I balance regime that generates asynchronous activity without large feedforward inputs.
• To explore networks supported by sparse, dynamic neuronal populations.

Main Methods:

• Developed computational models of neural networks.
• Simulated network activity and analyzed neuronal population dynamics.
• Employed mean-field analysis to characterize network properties.

Main Results:

• Identified a 'sparse balance' regime supporting irregular activity via a dynamic, sparse neuronal set.
• Demonstrated that high-variance synaptic strength distributions are crucial for this regime.
• Showed that synaptic fluctuation speed, not magnitude, determines sparsity.
• Observed robust nonlinear responses to uniform and non-Gaussian inputs.

Conclusions:

• Sparse balance networks offer a more experimentally plausible mechanism for E-I balance and spontaneous activity in cortical circuits.
• These networks exhibit distinct computational properties, including enhanced responsiveness to stimuli.
• Synaptic dynamics play a critical role in shaping network states and sparsity.