COMPUTATIONAL MODELING AND SIMULATION OF THE CEREBELLAR CIRCUITS INVOLVED IN EMOTIONAL CONTROL

Understanding the neural basis of emotional processing has become a central theme in contemporary neuroscience. Fear conditioning, an evolutionarily conserved form of associative learning, is a widely used paradigm for investigating the neural basis of emotion. Its usefulness stems from its clear behavioral outputs, well-established experimental structure, and the cross-species conservation of key underlying neural mechanisms. This makes it a highly translational model for investigating how the brain detects, processes, and learns from emotionally salient or threatening stimuli, ultimately contributing to adaptive behavioral responses. While historically overlooked in emotion research, the cerebellum—and specifically lobule VI—has recently gained attention for its potential integrative role in both motor and affective processes. In this thesis, we investigate the cerebellum’s contribution to emotional learning by combining computational modeling and experimental datasets. We developed a detailed, data-driven reconstruction of the mouse cerebellar lobule VI (Declive) using the Blue Brain Cell Atlas, the Brain Scaffold Builder framework, and we simulated its dynamics as single-point neural networks by using NEST. This Declive-specific model was compared against a canonical cerebellar microcircuit to identify structural and functional region-specificities. Building upon this, we challenged the model to explore Pavlovian conditioning in rodents, simulating neural responses to conditioned and unconditioned stimulation patterns. We therefore defined hypotheses for plasticity mechanisms and interactions between the olivocerebellar circuit and the periaqueductal gray region, involved in fear conditioning. Moreover, at a large scale, we identified the human fear network within the Virtual Brain framework by tuning the whole-brain system parameters on subject-specific neuroimaging datasets. The focus was on the cerebellar nodes involved in the brain network. We then correlated the parameter sets with fear-related response outcomes. Although these models operate at different scales and in different species, together they serve as complementary pioneer work aimed at investigating emotional processes and cerebellar involvement. This modeling strategy illustrates the potential of biologically grounded computational neuroscience to explore brain–behavior relationships across levels of organization.