Multi-site plasticity to support parallel memories in a recurrent network model of the mushroom body

Despite their relatively small brains insects show complex and adaptive behaviors. The limited neuronal resources require particularly efficient coding strategies and thus make insects favorable model systems for identifying fundamental properties of nervous system function. The collaborative aim of our consortium is to unravel the mushroom body function and underlying mechanisms in the model organism Drosophila melanogaster with respect to the implementation and recall of parallel and distributed memories that persist on different time scales. The ultimate goal of this project (as part of the consortium) is the generation of a biologically realistic neural network model of the mushroom body. This model should incorporate known mechanisms such as sparse coding and mechanisms of functional and structural plasticity as investigated in the collaborative approach of this consortium. Our model shall reproduce the observed neurophysiology and correctly predict the fly behavior in tested and untested learning experiments. We will address our goals in a complementary approach, combining modeling at different levels of abstraction with data analysis specifically tailored to the quantitative estimation of model parameters in close collaboration with our experimental partners.