The long-term vision of the group is to understand the role of nonlinear dendrites in hippocampal model-based predictions.

Thanks to the recent advances in in vitro experimental techniques, we have reasonably good understanding of how temporally synchronous and spatially clustered inputs elicit dendritic spikes in hippocampal pyramidal neurons at the biophysical level.

On the behavioral level, we know, that the hippocampus is essential for spatial navigation, and, more generally, in model based planning in both humans and rodents. In collaboration with Judit Makara, we use virtual reality experiments to uncover how neural networks represent relevant task variables.

The population activity in the hippocampus is characterised by internally generated sequential activity patterns: place cells show sequential activity in individual theta cycles during navigation and they replay the trajectories in either forward or reverse directions at rest. We apply state of the art machine learning and data analysis tools to discover similar sequential activity patterns in non-spatial data.

We are building computational models to bridge the gap between these levels of investigations and understand the contribution of nonlinear dendrites in sequence generation, navigation and planning.