whole-bRaIn rodent SpikING neural NETworks (RisingNet)

Generating a spiking whole-brain model in rodents is critical for computational, theoretical and medical applications and represents a fundamental step toward human brain models (Einevoll et al., 2019; Amunts et al., 2019). Compared to humans, modelling the rodent brain gives unique advantages in exploiting experimental datasets with high definition of cellular processes and connectivity, as well as in dealing with smaller brain networks and simplified behaviours (D’Angelo, 2019). This allows to test in detail the principles of brain organization and computation before scaling up to humans. The RisingNet consortium gathers 3 new non-HBP partners (receiving 61% of the resources) supported by 3 core HBP partners strategically positioned for the success of the project. RisingNet will generate a model of the rodent brain by applying a modular approach, in which the main brain microcircuits are organized in modules integrated through the connectome. The modular approach of RisingNet is compatible with the architecture of The Virtual Brain as well as with the Brain Reference Atlas and is scalable for applications to robotic controllers. This makes RisingNet instrumental to WP1 as well as to WP2 and WP3. RisingNet simulations will run on EBRAINS and the results will be mapped against high-resolution brain signals for cross-validation. The constructive principle is to develop a “scaffold” brain model incorporating modules of different microcircuits including cerebral cortex, cerebellum, hippocampus and basal ganglia along with intercalated nuclei (thalamus, pontine nuclei and the inferior olive) wired through the connectome derived from the Mouse Allen Brain Atlas. The microcircuits will be made of point neurons endowed with realistic dynamic properties and will maintain the salient firing dynamics and synaptic integration properties of the realistic models from which they will derive. The RisingNet whole-brain models, thanks to its modular scaffold architecture, will allow to simulate specific functional states targeted toward core scientific issues. These will address behavioural tasks integrating sensori-motor control, spatio-temporal navigation, and action selection operated through the cortico-cerebellar, cortico-hippocampal and cortico-basal loops of the model (D'Angelo et al., 2019). The RisingNet whole-brain model, by extending and differentiating the modules and their connectivity, will eventually serve as a virtual brain simulator allowing to simulate experiments carried out in WP2. The RisingNet whole-brain model will also be a valuable tool for implementing robotic and neuromorphic controllers in WP3. Finally, the RisingNet microcircuit modules will be adapted and integrated into The Virtual Mouse Brain (TVMB) nodes allowing simulations of brain dynamics and activity maps. The RisingNet rodent whole-brain model covers a missing element that does not have any similar counterpart in the work plan and will substantially contribute to generate the HBP “common brain reference framework” of SGA3.