From early mathematical models of neuronal activity to today’s data-rich and computationally intensive neuroscience, theoretical and computational approaches have played a central role in shaping our understanding of the brain. As experimental, analytical and AI-based methods continue to evolve, major scientific advances increasingly emerge at their intersections—where models confront data, and where biology, physics, mathematics and computation jointly inform our understanding of cognition and brain (dys)function.
Manifesto
Computational neuroscience aims to determine the principles governing neuronal activity and brain functions. It has a long history with seminal works dating back to the early 20th century. A notable example in France is Louis Lapicque’s pioneering mathematical modeling of neuronal excitability in 1907 that laid the ground for the “Integrate and Fire” neuron model. Over time, computational neuroscience has become an essential tool for the study of the brain, in particular to overcome the technological limitations that constrain experimental discoveries and to collectively interpret diverse types of data by integrating them in biophysical models. Interdisciplinary by nature, it has always combined approaches from physics, applied mathematics and numerical science in addition to biological neuroscience. In parallel, the availability of increasingly complex data due to progress in experimental neuroscience –such as massive parallel recording of distributed neuronal activity– underlies the need for elaborate analysis methods, like high-dimensional statistical models. Last, recent advancements in neighboring fields like artificial intelligence and neuroinformatics have questioned the boundaries that were previously established between computational neuroscience and these disciplines. For instance, artificial neural networks were inspired by biological neuronal networks to analyze data and have become models to study the brain cognition, with the bidirectional interactions between the fields spreading over more than 50 years.
This situation calls for a reexamination of definitions and themes within computational neuroscience. The Computational Neuroscience Research Network (“RT NeuroComp” in French) aims to improve the integration between emerging and traditional fields related to computational neuroscience, including physics, applied mathematics, computer science, biology, cognitive science, and clinical research. The RT operates through a decentralized structure centered on Working Groups (WGs), which members can establish flexibly to address new needs or opportunities. Our view is that boundaries with neighboring fields cannot be rigidly fixed, as very active research directions recently emerged at the intersection of several disciplines. In this way, WGs are meant to reflect the diversity of current research directions and perspectives, including those at the fringes of the traditional definition of computational neuroscience. The RT steering committee will coordinate the WGs to facilitate interactions between them and with related scientific communities, in particular with mathematics, physics, computer science via the corresponding CNRS, INSERM, INRIA and university structures.
By uniting our research community, the RT NeuroComp thus aims to clarify and foster internal interactions, as well as with experimental neuroscience and neighboring fields. This is key to the visibility of the French community in computational neuroscience, both nationally and internationally. We thus welcome members from researcher and engineer to student levels to create a lively environment that promotes new national collaborations and transdisciplinary partnerships.