It has been recognised that the higher-level mechanisms of information processing and storage in the brain are implemented through the temporal dynamics and spatial distribution of neuronal activity. Since spatiotemporal evolution of activity patterns is a function primarily of the architecture of interneuronal connectivity, this has prompted extensive theoretical research on neural networks. In the perspective of theoretical neuroscience, a network is a collection of coupled dynamical systems, each describing an individual processing unit, a neuron. The latter typically are lower-dimensional approximations of some biologically realistic model of the single neuron that capture only the features thought to be most significant within the network context. However, in contrast to artificial neural networks studied in computer science, neuroscientists try to preserve as much biological relevance as possible. A limiting factor is the mathematical or computational tractability as well as the gaps in our knowledge on the finer details of the real system.