Adam Gosztolai receives prestigious ERC Starting Grant from the EU

During everyday tasks, such as reaching a bus on our morning commute, our brains coordinate the activity of neurons across multiple regions. Our hippocampus recalls the relevant objects for the task, such as bus and bus stops; our prefrontal cortex measures the time it takes to get to the bus stop; our visual cortex senses the visual surroundings; our motor cortex coordinates actions such as walking, and so on. Until now, studying such global brain processes has only been possible using neuroimaging, such as fMRI and EEG. However, these techniques cannot get to the single neurons; instead, they only see a ‘blurred’ picture of the activity of many neurons at once. While recent electrode-based and optical recording technologies enable recording the activity of many neurons, they are limited to surveying focal brain regions and which neurons they record depending on how the experimenter or surgeon inserts the electrode. While these focal measurements are sufficient to understand how neurons coordinate simple tasks, such as moving a muscle, studying complex behaviours requires understanding the collective activity of single neurons across multiple regions during complex tasks. Linking single-neuron activity to global brain states is also highly relevant for neurology and psychiatry, as drugs act on the function of single cells while the desired effect is on global brain activity.

This ERC project aims to develop a mathematical theory linking single-neuron recordings across the brain into a unified model that can predict global brain states. Adam Gosztolai and his team will do this by finding and mathematically characterising the dynamical motifs in neural population activity that are in some sense ‘invariant’, i.e., they do not depend on which neurons the researchers record. “If we succeed, we will be able to ’tile’ the activity of the whole brain by taking local recordings, one at a time, and reconstructing the remaining activity using our mathematical formalism”, explain Adam Gosztolai. “Our dream is that this methodology will establish a ‘foundation model’, a large mathematical model of many single neurons across the brain that can make neuron-specific predictions in different tasks. I expect this technology will help predict global brain function in complex animal tasks, allowing experimenters to conduct fewer and more well-informed animal experiments. I also hope that bridging the gap between single neuron recordings and neuroimaging will facilitate better clinical diagnostics and predicting the effect of pharmaceutical interventions.”

About the person
Adam Gosztolai is a research group leader of the “Dynamics of Neural Systems Laboratory” at the AI Institute of the Medical University of Vienna and a research affiliate at the Department of Cognitive Sciences at the Massachusetts Institute of Technology (MIT). He read engineering and mathematics at University College London and the University of Cambridge and obtained his PhD in mathematics from Imperial College London. Following his PhD, Dr Gosztolai has done postdoctoral research at the École Polytechnique Férédale de Lausanne (EPFL), in the fields of computational neuroscience and machine learning. For his postdoctoral research he was awarded a prestigious Human Frontiers Science Foundation Fellowship. In his research, Dr Gosztolai studies the dynamical processes encoded in the activity of a large number of neurons in the brain to distil fundamental principles of how these collective dynamics are linked to neural processes such as cognition and motor control. He aims at transferring these principles to design artificial neural systems that mimic how the brain works. In this context, he is fascinated by collaborations between neuroscience and artificial intelligence, particularly pertaining to the development of novel algorithms using generative machine learning for use in brain-machine interfaces in spinal cord injury rehabilitation.