Transient boosting of action potential backpropagation for few-shot temporal pattern learning.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Synaptic Plasticity

Background:

• Cortical neurons detect spike patterns in noisy data.
• Rapid development and robustness of pattern-selective neuronal responses are not fully understood.

Purpose of the Study:

• Propose a biologically plausible synaptic plasticity rule for rapid learning of patterned synaptic inputs.
• Investigate the role of somatodendritic coupling in this learning process.
• Explore network-level learning in recurrent networks.

Main Methods:

• Development of a novel synaptic plasticity rule.
• Modeling intracellular self-supervised learning.
• Simulation of recurrent neural networks.
• Analysis of spike-triggered somatodendritic coupling effects.

Main Results:

• The proposed rule enables rapid learning of intermittently co-activated presynaptic-neuron communities.
• A spike-triggered increase in somatodendritic coupling significantly boosts synaptic crediting for learned responses.
• This mechanism is essential for high signal-to-noise ratio pattern learning in single neurons.
• Recurrent networks utilizing this rule demonstrate faster, few-shot learning of multiple patterns.

Conclusions:

• The study presents a novel mechanism for rapid, self-supervised learning of neural patterns.
• Backpropagating action potentials play a key role in facilitating this rapid pattern learning.
• The findings offer insights into how neural circuits efficiently process complex information.