This paper explores the importance of cognitively impenetrable processing in biological visual systems, in particular low level processing of motion information. Three visual tasks involving motion processing that are critical to living creatures have been identified, namely the detection of direction of heading, estimation of time to contact and image segmentation. The basis for motion processing in biological systems appears to be local motion detectors. The performance of a variety of biologically inspired motion detection models are therefore examined and compared. Some important issues for further processing of local motion information to solve the three visual tasks are also discussed.