Optimal plasticity for memory maintenance during ongoing synaptic change.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Synaptic connections in brain circuits undergo significant turnover and remodeling over short timescales.
• This synaptic flux persists even without learning or known plasticity signals, posing a challenge for information retention.

Purpose of the Study:

• To investigate how neural circuits preserve learned information amidst substantial, learning-independent synaptic changes.
• To determine the optimal level of compensatory plasticity required to counteract ongoing synaptic fluctuations.

Main Methods:

• Theoretical analysis of neural circuit dynamics.
• Modeling compensatory plasticity in response to synaptic fluctuations.
• Mathematical analysis of error gradient computation in plasticity mechanisms.

Main Results:

• The required compensatory plasticity is largely independent of specific plasticity mechanisms and circuit architectures.
• Optimal compensatory plasticity is at most equal in magnitude to synaptic fluctuations, and often less.
• A high degree of learning-independent synaptic change is compatible with plasticity mechanisms that accurately compute error gradients.

Conclusions:

• Neural circuits can maintain learned information despite significant synaptic turnover.
• Minimal compensatory plasticity is sufficient to counteract disruptive synaptic changes, aligning with experimental observations.
• Error-driven plasticity mechanisms can support stable information storage in dynamic neural circuits.