Dynamic Mechanisms in Neural Networks

^{Copyright: Relation between recurrency of connectivity and dimensionality of activity in neural networks (copyright: Dahmen, Recanatesi et al., 2022 (CC-BY license))}

The focus of this group is the investigation of mechanisms that shape the activity in neuronal networks on multiple spatial scales. Research builds on theoretical descriptions from the microscopic level of single cells and synapses to effective equations capturing interacting brain areas. This allows the discovery of the mechanisms that underlie and shape the observed phenomena of correlated and oscillatory neuronal activity.

The group uses mean-field methods to identify the local circuits that generate oscillations, to  explain multi-stability in networks with numerous areas of visual cortex, to investigate the onset of spatio-temporal activity patterns in spiking networks, and to develop quantitative theories of correlations in the activity of populations of cells. Such theories, for example, show how inhibitory feedback causes a weakly correlated balanced state. More elaborate methods include the theory of disordered systems. This way variability of local synaptic connections can be included, for example to explain the emergence of low-dimensional, slow modes of collective activity observed in motor cortex. By adapting renormalization group methods to neuronal systems, states of non-equilibrium criticality are investigated, finding for example marginal criticality in spatially organized networks.