New cerebellum simulation captures brain activity with millisecond precision

A new study supported by EBRAINS 2.0 and the Virtual Brain Twin Project has presented a biologically grounded computer model of the cerebellar cortex – the part of the brain involved in coordinating movement and supporting learning – that simulates brain activity with remarkable realism.

The spiking neural network incorporates key features of real neurons, including synaptic dynamics, receptor-specific signalling, and the behaviour of NMDA receptors, which play an important role in learning and plasticity. The model was calibrated using detailed cellular simulations to ensure that its behaviour closely matched that of biological neurons.

The study also drew on two tools from the EBRAINS ecosystem. The researchers first reconstructed a highly detailed model of the cerebellar cortex using Brain Scaffold Builder. This biologically rich reference model was then used to systematically calibrate a more computationally efficient spiking neural network, with the NEST simulator, enabling the final framework to capture realistic brain dynamics while remaining suitable for large-scale simulations.

The researchers showed that the new framework, available via EBRAINS, accurately reproduced the timing and patterns of brain activity observed in experiments on murine cerebellar tissue. By selectively “switching off” different receptor types in the simulations, they also revealed how NMDA, AMPA and GABA receptors shape the way cerebellar circuits respond to inputs arriving at different frequencies.

The authors, from the University of Pavia, suggest the model could be used for large-scale simulations to explore how cerebellar disorders emerge and how the cerebellum supports learning.