Representational drift without synaptic plasticity.

Area of Science:

• Neuroscience
• Computational Neuroscience

Background:

• Neural computations enable stable behavior despite dynamic biological changes.
• Representational drift (RD) describes neuronal response changes over time, distinct from behavioral alterations.
• RD is traditionally attributed to synaptic weight modifications, necessitating downstream adaptation.

Purpose of the Study:

• To propose and validate a simpler mechanism for representational drift.
• To investigate if changes in neuronal excitability, rather than synaptic plasticity, can explain RD.
• To assess the impact of excitability changes on neural tuning and population readouts.

Main Methods:

• Utilized spike coding networks (SCN) to model neural dynamics.
• Simulated changes in neuronal excitability without altering synaptic weights.
• Developed and tested decoders to assess population readout stability across different excitability states.

Main Results:

• Excitability changes alone can alter neuronal tuning, mimicking observed RD.
• These changes do not necessitate costly adaptation in downstream neural areas.
• Spike coding networks demonstrated that excitability shifts can explain the extent of experimentally observed tuning changes.
• Decoders trained on specific excitability settings showed performance degradation on others, but a general decoder remained effective.

Conclusions:

• Neuronal excitability changes offer a parsimonious explanation for representational drift.
• This mechanism preserves stable downstream decoding and behavior without synaptic plasticity.
• The findings challenge the necessity of synaptic plasticity for explaining RD and stable neural function.