Efficient codes and balanced networks.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Growing interest in inhibitory interneurons and their circuits.
• Cortical inhibition exhibits tight balance with excitation on a millisecond timescale.
• This balance tracks both external stimuli and spontaneous neural fluctuations.

Purpose of the Study:

• Review recent theoretical approaches on the advantages of tight excitatory/inhibitory balance.
• Investigate the role of this balance in neural coding and computation.
• Challenge the dominant view of neural information representation.

Main Methods:

• Review of experimental evidence.
• Analysis of theoretical models investigating excitatory/inhibitory balance.
• Exploration of computational advantages of tight balance.

Main Results:

• Tight excitatory/inhibitory balance suggests a highly cooperative neural code.
• This code is orders of magnitude more precise than a Poisson rate code.
• Tight balance may facilitate high-dimensional population coding and learning.

Conclusions:

• Revises the dominant view of neural firing rates corrupted by Poisson noise.
• Tight balance is a signature of precise and cooperative neural coding.
• Provides a framework for understanding how cortical neurons learn complex functions.