Computational Systems Neuroscience

How do neurons in the brain concert their activity in order to rapidly process sensory information and to form correct decisions? How is information stored and retrieved in a neural network? Could principles of neural computation be useful for technological solutions? In the Computational Systems Neuroscience group we investigate information processing in nervous systems of different animal models. In an interdisciplinary team we combine theoretical and experimental approaches with the ultimate goal to formulate valid theories and testable model hypotheses.

A major focus of our research is placed on the neuronal processes underlying sensation, memory formation, and decisions in insects. Insects are equipped with highly evolved sensory systems, they express complex context-dependent behavior, they show fundamental learning abilities, and they share many fundamental neuronal processing mechanisms with higher animals. Yet, insects have very limited neuronal resources in the order of 100k (flies) to 1Mio (bees and ants) neurons. Thus, insects are of particular interest for studying efficient coding strategies and they are ideally suited for the simulation of biologically realistic brain models.

The insect repertoire of goal-directed behavior is still beyond the capabilities of today’s artificial systems. In the field of Neurorobotics we test insect-inspired neural network models for the control of autonomous robots. The robotic platform provides an embodiment for the simulated nervous system, which in turn makes the robot sense, learn, and interact with its environment.

One step further we use our models to explore the potential of Neuromorphic Computing, an emerging technology where ‘artificial minibrains’ can be realized on dedicated microchips that physically implement a network of neurons and synapses. Parallel processing, distributed memory, and high energy efficiency are promising features of neuromorphic computing that could become useful for intelligent systems and autonomous robots.