In silico brain models for understanding pathologies

This talk will delve into the use of biologically inspired in silico models for studying neural pathologies. These models are virtual replicas of specific brain regions or multi-area circuits, made of realistic spiking neurons or more abstract mass models. Unlike traditional artificial neural networks, spiking neural networks offer a more detailed representation of real biological neuron behaviour. They encode spatiotemporal information through precise temporal patterns in spiking activity, they can be equipped with short- and long-term plasticity mechanisms and can be integrated with other more abstract models. By leveraging the peculiarities of these models, we can simulate, and therefore better understand, neural pathologies accurately and efficiently. Different case studies demonstrate how in silico brain models can be manipulated to provoke changes at the neuronal or synaptic level, and how these manipulations impact network behaviour and influence neural dynamics, making it possible to infer circuit alterations occurring in neuropathologies.