Theoretical neuroscience: Math for life

Among the natural sciences, biology has the reputation of being less precise compared to disciplines like physics. The chaos of living systems just will not fit into mathematical formulas. Professor Wiktor Młynarski begs to differ. He wants to show that life can indeed be calculated and simulated to a certain degree. “How can we combine the seemingly messy, unpredictable world of biology with the exact, formal world of mathematics?” This question inspires the specialist in computational neuroscience, who was appointed to LMU’s Faculty of Biology in 2023.

With his research group, he develops models that simulate the processing of sensory stimuli in the brain. The models are designed to help us better understand what exactly happens in our web of neurons when we see, hear, or smell. Unlike his experimentalist colleagues, Młynarski does not do experiments in the laboratory. His work is purely mathematical and theoretical, and his tools are pen and paper, a flipchart – and the computer on which he generates complex simulations. Even though he studied computer science in Poland at the Jagiellonian University, he calls himself a bio-scientist: “Biology is the science of life; neuroscience studies the brain – and that’s exactly what we do. It’s just that we do it with the methods of mathematics and computer-based tools.”

Rooting models in reality

He already followed this pathway while undertaking his doctorate at the Max Planck Institute for Mathematics in the Sciences in Leipzig, and he further pursued this avenue of research at the interface of computer and life science while working in the Department of Brain and Cognitive Sciences at MIT and subsequently in a research group for theoretical biophysics and neurosciences at the Institute of Science and Technology Austria (ISTA).

With his models, Młynarski wants to investigate which problems the brain can solve, how it solves them, and how good it is at solving them. He wants to ascertain to what extent theoretical simulations can describe actual neurobiological phenomena. “Building a model for its own sake is not an interesting endeavor,” he explains. “You have to root it in reality somehow.” These roots can take two forms: practice and theory. On the one hand, you can repeatedly compare the models against real phenomena, as described by experimental neuroscientists. On the other, you can probe certain theoretical boundaries – universal laws of nature that cannot be exceeded from a purely mathematical standpoint.

Animals are not robots  

Młynarski is currently applying this method to understand how sensory stimuli are further processed in the brain and how this is related to locomotion. Recent experiments in behaving animals demonstrated that processing of stimuli in the brain violates our basic intuitions. That is to say, neuronal activity is not just controlled by visual signals, but also by the locomotor system. “This is not how we humans would build a system like that,” says Młynarski. A robot has a camera to perceive its environment and motors to power its motion, but these two flows of information run separately into a central processing unit. In animals, by contrast, the ‘camera’ is somehow directly connected to the ‘motor.’ “We know this happens in lots of species, but we have no idea which problem it solves.” Finding out is one of the goals of Młynarski’s research.

Could his work help robotics learn from nature in the future? Yes, that is certainly a possibility, reckons the neuroscientist. As soon as we understand, for example, why evolution coupled our sensory organs and locomotor system in this way, we could transfer this problem-solving approach to technical fields of application. “At the moment, we don’t even know what problems are solved by this coupling.” As part of research into artificial intelligence, neuro-AI and neurosciences are currently the focus of intense debate. However, this cross-fertilization is often misunderstood: “We frequently hear that computer science was inspired by biology. But many people are not aware that this can mean many things.” Even if computer science adopted some basic theoretical principles from biology, computers and brains work very differently. “Inspiration does not imply identity. Artificial intelligence systems inspired by biological findings may still work very differently than systems they were inspired by.”