Abstract Many animals use visual motion cues to inform different behaviors. The basis for motion detection is the comparison of light signals over space and time. How a nervous system performs such spatiotemporal correlations has long been considered a paradigmatic neural computation. Here, we will first describe classical models of motion detection and introduce core motion detecting circuits in Drosophila. Direct measurements of the response properties of the first direction-selective cells in the Drosophila visual system have revealed new insights about the implementation of motion detection algorithms. Recent data suggest a combination of two mechanisms, a nonlinear enhancement of signals moving into the preferred direction, as well as a suppression of signals moving into the opposite direction. These findings as well as a functional analysis of the circuit components have shown that the microcircuits that process elementary motion are more complex than anticipated. Building on this, we have the opportunity to understand detailed properties of elementary, yet intricate microcircuits.