Synaptic self-organization of spatio-temporal pattern selectivity.

Area of Science:

• Computational Neuroscience
• Neural Networks
• Synaptic Plasticity

Background:

• Spiking model neurons can be optimized for specific spatio-temporal spike pattern recognition.
• The capability of biological synaptic plasticity mechanisms to achieve this temporal coding remains unclear.

Purpose of the Study:

• To investigate if existing synaptic plasticity mechanisms can enable neurons to become sensitive to specific input spike patterns.
• To explore the self-organization of temporal pattern detection and memory robustness in neural networks.

Main Methods:

• Simulations of spiking model neurons incorporating Hebbian mechanisms, hetero-synaptic plasticity, and synaptic scaling.
• Analysis of how these plasticity rules influence synaptic efficacies and neuronal responses to spatio-temporal input patterns.

Main Results:

• A combination of Hebbian mechanisms, hetero-synaptic plasticity, and synaptic scaling is sufficient for self-organizing sensitivity to repeating spatio-temporal spike patterns.
• Hetero-synaptic plasticity promotes specialization and faithful representation of pattern sequences in neuronal populations.
• Pattern detection demonstrates robustness against noise and distortions, with memories protected from overwriting during periods of pattern absence.

Conclusions:

• The proposed combination of plasticity mechanisms supports precise temporal coding and computation in spiking neural networks.
• This provides a potential explanation for the long-term robustness of memory traces in the brain despite continuous synaptic plasticity.
• The findings enhance the plausibility of temporal coding as a fundamental mechanism in neural computation.