Binary and analog variation of synapses between cortical pyramidal neurons

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Learning relies on changes in neuronal connections.
• Previous models assumed continuous synapse sizes, aligning with analog neural network models.

Purpose of the Study:

• To create the largest connectivity map of layer 2/3 pyramidal cells in the mouse visual cortex.
• To investigate the distribution of synapse sizes and its implications for cortical learning algorithms.

Main Methods:

• Automated analysis of serial section electron microscopy images.
• Construction of a large-scale neuronal connectivity map (250 × 140 × 90 μm³).
• Statistical modeling of synapse size distribution.

Main Results:

• Synapse size distribution between L2/3 pyramidal cells is better modeled as a sum of binary and log-normal analog variables, not a continuous log-normal distribution.
• Correlated binary variables suggest uniform activity-dependent plasticity, while uncorrelated analog variables indicate other influences.

Conclusions:

• Cortical synapse sizes exhibit both discrete (binary) and continuous (analog) components.
• This binary-analog model offers new constraints for understanding synaptic plasticity and learning rules in the brain.
• Findings challenge the purely analog view of synaptic strength and support models involving bistable synaptic states.