Andrej Warkentin: A neuronal model for visually evoked startle responses in schooling fish

BCCN Berlin / TU Berlin

Many aspects of fish school behavior can be explained qualitatively by self-propelled agent models with social interaction forces that are based on either metric or topological neighborhoods. Recently, startling of fish has been analyzed in its dependence of the network structure (Rosenthal et al., 2015) but a mechanistic model and its influence on the collective behavior is missing. Here we coupled a model for collective behavior with a neuronal model that receives looming visual stimulus input to initiate a startle response, inspired by the neurobiologically well-studied Mauthner cell system. First, we analyzed the basic properties of the startle behavior of a single fish as a reaction to a looming stimulus and built a neuronal model to reproduce the startle behavior. Next, we fitted the neuronal model to experimental, behavioral data from larval zebrafish. On the group level, we included the fitted neuronal model in the collective behavior model and looked at startling frequency as well as group cohesion and polarization depending on collective behavior parameters via simulations of the combined model. Our results indicate that the fitted neuronal model can lead to experimentally observed startling frequencies and that there are nontrivial relationships between the startling frequency and the group order so that more investigation is needed. In summary, we took first steps towards a biologically plausible model for startle response initiation in the context of collective behavior.  

Additional Information

MSc defense in the International Master Program "Computational Neuroscience"

Organized by

Pawel Romanczuk / Robert Martin