A multi-layer mean-field model of the cerebellum embedding microstructure and population-specific dynamics

Area of Science:

• Computational neuroscience
• Neurobiology
• Systems neuroscience

Background:

• Mean-field (MF) models are essential for large-scale brain simulations, but lack representation for certain brain regions like the cerebellum.
• Existing MF models primarily focus on the isocortex, leaving a gap in understanding other crucial neural structures.

Purpose of the Study:

• To design and simulate a multi-layer mean-field model of the cerebellar microcircuit.
• To validate the cerebellar MF model against experimental data and a spiking neural network (SNN) model.
• To provide a computationally efficient tool for investigating cerebellar function.

Main Methods:

• Developed a multi-layer MF model incorporating Granule Cells, Golgi Cells, Molecular Layer Interneurons, and Purkinje Cells.
• Used a system of equations with inter-dependent transfer functions for neuronal populations and topology.
• Optimized the model's time constant using experimental local field potentials from mouse cerebellar slices.

Main Results:

• The MF model successfully reproduced average dynamics of cerebellar neuronal populations under various input patterns.
• The model predicted Purkinje Cell firing modulation based on cortical plasticity and feedforward inhibition.
• Validation against experimental data and SNN models confirmed the MF's accuracy.

Conclusions:

• The developed cerebellar MF model is a computationally efficient tool for neuroscience research.
• This model facilitates future investigations into the link between microscopic neuronal properties and macroscopic brain activity.
• It holds potential for studying both physiological and pathological conditions within virtual brain models.