Eluding oblivion with smart stochastic selection of synaptic updates

Area of Science:

• Neuroscience
• Computational Neuroscience
• Memory Research

Background:

• Synaptic dynamics are characterized by bounded variables, leading to a 'forgetful' nature where new experiences overwrite old memories.
• The rate of forgetting is directly proportional to the number of synapses modified by new experiences, impacting learning speed and memory retention.

Purpose of the Study:

• To explore mechanisms that mitigate rapid forgetting in synaptic plasticity.
• To investigate strategies for enhancing memory retention by optimizing synaptic modification processes.

Main Methods:

• Review of theoretical 'tricks' to slow down synaptic learning.
• Analysis of stochastic synapse selection for memory consolidation.
• Examination of spike-driven synaptic dynamics with noisy neural activity.

Main Results:

• Selective modification of a small fraction of synapses allows for memory acquisition without compromising old memory retention.
• Reducing the number of modified synapses extends memory span at a potential cost of information density.
• Stochastic selection of synapses for modification can be triggered by specific neuronal responses, optimizing learning and retention.

Conclusions:

• Smart, stochastic selection of synapses is a key strategy to elude fast forgetting (oblivion).
• This approach balances learning speed and long-term memory retention, crucial for cognitive function.
• Naturally emerging spike-driven synaptic dynamics with chaotic neural activity support these memory-enhancing mechanisms.