Advanced 3D brain models created to study memory and perception

The hippocampus is one of the most fascinating brain regions. Associated with the formation of memories, it also helps us to navigate through the world without getting lost. Sensory cortices on the other hand play an important role in how we perceive our environment and make appropriate movements, and how our brains determine what to focus on and what to ignore. While both regions have been extensively studied and many of their secrets revealed, there is still a lot we do not understand about them due to the high complexity of interacting parts, from individual synapses and the zoo of different neuron types, to the detailed connectivity rules between them. To better our understanding, EPFL researchers have built detailed computational models of these regions. Putting together the neurons comprising these regions and describing the rules of their interactions through computer code, they are able to simulate the brain activity in these regions and study the roles of each part in the concert of brain activity.

Unlike previous models, these models were built with the exact three-dimensional geometry of their corresponding brain region. This opens the door for future refinement and testing of the models with any new experimental data. By focusing on building such general three-dimensional models, the models can be also used to explore a wide range of phenomena.

This is not an easy process. Describing the rules governing the regions and turning them into computer simulations required the input of the many experts that have found and know these rules. The researchers have therefore collaborated with over 80 colleagues from all over the world to develop the largest and most detailed models of these brain regions.

The integration of data from multiple sources and collaboration among scientists are the strengths of these models, though they also presented challenges. By addressing these obstacles, the models have become more robust, adaptable, and accessible to a broader scientific community.”

Dr. Armando Romani, group leader of the Circuits groups at Blue Brain

They have now openly released the models to the scientific community along with the tools to study and use them. The models are described in four extensive papers that each focus on different aspects.

In Modeling and Simulation of Neocortical Micro- and Mesocircuitry. Part I, published in the journal eLife, the focus lies on the anatomy of the somatosensory regions and its connectivity. Its main insight is that the shape of brain regions affects the structure of brain networks formed within and a description of how connectivity at different scales comes together to form highly complex patterns. “We are sometimes used to thinking about local and long-range connectivity as separate systems”, notes Dr. Michael Reimann, group leader of the Connectomics groups at Blue Brain. “It really surprised us to see how the systems interact to form these very structured types of network.”

Part II, published in eLife alongside the first paper, describes the physiology of the brain region and how it was modeled and validated at the synaptic, neuronal and network-level. “This allowed us to make predictions about how particular components of the brain, such as specific connectivity patterns, contribute to observations about cortical processing made by our experimental colleagues,” explains lead researcher Dr. James Isbister. “The model’s 3D geometry allows us to study communication between brain areas, and most interestingly, to recreate experiments combining complex laboratory methods such as optogenetics with approaches only possible in simulations, such as lesions between very specific populations”.

A third paper in eLife explains how the model was then improved further to include the process of synaptic plasticity, the fundamental mechanism that allows us to learn new information. Its insights pertain to the complex rules that govern the processes that emerge when millions of synapses undergo plasticity under in vivo conditions – like in the living brain. “For the longest time, simulations have focused on plasticity rules based on lab experiments, under artificial conditions”, lead researcher Dr. Andras Ecker points out. “We wanted to explore plasticity in detailed networks and in vivo.”

Finally, a fourth paper in PLOS Biology presents a comprehensive in silico model of the rat CA1 region, integrating diverse experimental data from synapse to network levels, including the Schaffer collaterals – key conduits for information transfer and synaptic plasticity in the hippocampal circuit – as well as the effects of the neurotransmitter acetylcholine . “Each component was rigorously tested and validated, and we made all the input data, assumptions, and methodologies fully transparent” adds Dr. Romani. “Now accessible on hippocampushub.eu, this model serves as a flexible tool for scientists, providing extensive analyses and an interface to support further hippocampal research.”

Three additional journal articles and three preprint manuscripts demonstrate the value of the models to the scientific community. In them, the models have been used to study inter-areal processing, the neural code, and the relation between neuron connectivity and activity. Results of plasticity simulations were compared to electron microscopy data and a predicted motif effect on synapse strength was confirmed. “We have long known that brain networks are complex and follow specific rules” explains lead researcher Dr. Egas Santander. “The model allows us to begin to explore the reasons for those rules.”