Humans as well as animals react to environmental changes by adjusting their behaviour. Thereby the nervous system which processes sensory information about environmental changes and controls behavioural responses has to rely on its sensory cells and neurons that respond with limited reliability. Even if exactly the same stimulus is presented repeatedly, considerable response variability can be found in various cell types in different animals. Neurons in the visual cortex of monkeys, for example, show an extremely high variability with spike count variances in the range of the mean response amplitude. Somewhat smaller, nevertheless formidable, variability is found in the visual system of insects. These observations lead to the questions where all this variability originates and how the animal deals with it. In my thesis I address these questions investigating peripheral as well as central processing stages of the blowfly visual system which served for long as a model system for studies on the reliability of neural coding.

Using the fly visual motion pathway as a model system I have investigated how the nervous system deals with inevitable noise sources at peripheral as well as central processing stages. Using a novel measure for the reliability of the responses I was able to show that peripheral synaptic filtering is not only advantageous to optimise the neural code in an information theoretical sense but also is a prerequisite to enhance reliable over unreliable response components. The filtering and the convergence on different processing levels contribute to the performance of a central neuron not being limited by photon noise. The tangential cells could also be shown to be robust against motion noise which is unavoidable as it results from the complex visual motion patterns experienced during behaviour.