Biological organisms continuously make behavioral decisions based on sensory data gathered from the natural world. The quality of these decisions depends on computing consistent and accurate estimates of biologically important features from these natural signals. Thus to understand how sensory data streams are processed, we must first understand the statistical relationship between sensory inputs and features in the real world. In my candidacy seminar I will describe how we take a two part approach to studying this question in the context of motion estimation based on visual input in the blowfly (Calliphora vicina). First, the joint statistical properties of visual signals and motion are evaluated using a unique "fly-camera". Second, the performance of biological motion estimation in blowflies is characterized using extracellular recording from visual wide field motion sensitive neurons in response to the natural scenery captured with the fly-camera. This two part approach will allow us to determine how the statistics of these dynamic natural signals affect optimal motion estimation in biological systems.