Dendro-plexing of Single Input Spikes via Multiple Synaptic Contacts Can Enhance Cortical Neuron Computation and...

Area of Science:

• Computational neuroscience
• Neuronal modeling
• Synaptic plasticity

Background:

• Cortical neurons exhibit multiple synaptic contacts on postsynaptic targets, a phenomenon lacking clear functional explanation.
• Dendritic cable filtering influences postsynaptic potentials (PSPs), with proximal synapses yielding brief PSPs and distal synapses yielding broader PSPs.

Purpose of the Study:

• To investigate the functional implications of multiple synaptic contacts and dendritic filtering in cortical neurons.
• To develop and evaluate a novel neuron model incorporating these features for enhanced computational capabilities.

Main Methods:

• Development of a "Filter-and-Fire" (F&F) neuron model that integrates multiple synaptic contacts and dendritic cable filtering.
• Comparison of the F&F model's memory capacity and pattern recognition abilities against a standard leaky Integrate-and-Fire (I&F) neuron model.
• Training the models to emit precisely timed spikes for specific input patterns and recognize spatio-temporal data (e.g., MNIST digits).

Main Results:

• The F&F model demonstrated a threefold increase in memory capacity compared to the I&F model for specific input patterns.
• The F&F neuron successfully learned to recognize spatio-temporal input patterns, a task where the I&F model failed.
• The model suggests that "dendro-plexing" enriches neuronal computation and offers potential for reduced axonal wiring.

Conclusions:

• Multiple synaptic contacts, combined with dendritic filtering, significantly enhance the computational power of cortical neurons.
• The proposed F&F model provides a plausible explanation for the functional significance of multi-synapse connectivity.
• This neuronal architecture offers benefits in both computational capacity and potential "hardware savings" in neural circuits.