A Hebbian learning rule mediates asymmetric plasticity in aligning sensory representations

Area of Science:

• Neuroscience
• Computational Neuroscience
• Synaptic Plasticity

Background:

• Mutual spatial alignment of sensory representations is crucial for brain function.
• Synaptic plasticity is a key mechanism for shaping and maintaining these representations.

Purpose of the Study:

• To model synaptic plasticity in response to spatial displacement between two sensory input channels.
• To investigate how Hebbian learning and receptive field properties influence plasticity asymmetry.

Main Methods:

• Developed a Hebbian model with separately normalized synaptic weights for each input channel.
• Simulated spatial displacement between two sensory representations with varying receptive field strengths and widths.

Main Results:

• Observed highly asymmetric plasticity between the two channels, with weaker/broader receptive fields exhibiting more plasticity.
• Identified three plasticity dynamics regimes: winner-take-all, mixed-shift, and no-shift, dependent on displacement and channel correlation.
• Found that sequential small displacements enhance plasticity, aligning with experimental observations.

Conclusions:

• Hebbian learning with separate channel normalization leads to asymmetric plasticity, driven by receptive field characteristics.
• The model explains the maintenance of spatial registry and the emergence of aligned representations in developing circuits.
• Findings have implications for understanding sensory integration and experience-dependent neural development.