Self-healing codes: How stable neural populations can track continually reconfiguring neural representations

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• The brain must balance stable representations with new information.
• Neural population activity shows significant changes over time, termed representational drift.
• This drift poses challenges for integrating dynamic codes with stable neural circuits.

Purpose of the Study:

• To investigate how neural plasticity mechanisms can stabilize evolving population codes.
• To understand how readout neurons can interpret changing neural activity without external feedback.
• To explain the integration of plastic neural codes with consolidated long-term memories.

Main Methods:

• Exploration of known plasticity mechanisms, including Hebbian learning and single-cell homeostasis.
• Modeling how these mechanisms exploit redundancy in distributed population codes.
• Analysis of recurrent feedback in readout circuits to correct drift-induced inconsistencies.

Main Results:

• Hebbian learning and homeostasis can compensate for gradual changes in neural tuning.
• Redundancy in population codes is key to stabilizing neural representations.
• Recurrent feedback aids in correcting inconsistencies arising from representational drift.

Conclusions:

• Simple, known plasticity mechanisms can stabilize neural tuning in the short term.
• These mechanisms offer a plausible explanation for maintaining coherence between plastic and stable neural representations.
• The study provides insights into how the brain integrates dynamic information with long-term memory.