On the Structure of Cortical Microcircuits Inferred from Small Sample Sizes

Area of Science:

• Neuroscience
• Computational Neuroscience
• Network Science

Background:

• Cortical microcircuits display nonrandom connectivity, often attributed to clustering.
• Distinguishing between different network topologies is challenging due to small sample sizes in experimental recordings.

Purpose of the Study:

• To investigate network topologies that explain nonrandom features of cortical microcircuits.
• To develop a reliable method for inferring network structure from limited experimental data.

Main Methods:

• Simulated distinct network topologies (clustered, distance-dependent, broad degree distributions).
• Evaluated network properties using small sample sizes.
• Developed and applied the sample degree correlation (SDC) statistic.

Main Results:

• Multiple distinct network topologies appeared similar when analyzed with small sample sizes.
• The SDC statistic reliably distinguished between network topologies regardless of sample size.
• Rat visual and somatosensory cortex data did not fit simple topological classes but were consistent with a combined spatial and hierarchical clustering model.

Conclusions:

• Inferring global network structure from local connectivity data is problematic due to small sample limitations.
• The SDC is a robust measure for classifying network topology from limited neuronal recordings.
• Cortical connectivity likely arises from a combination of spatial proximity and nonspatial, asymmetric clustering.