A general systems approach is taken to studying the emergent properties of the human perception/action system. Two task domains, the control of locomotion and the recognition of objects, are used to study human performance. The locomotion task involves the control of altitude. Experiments are described that will manipulate the type of texture, the speed of forward motion, and altitude. A general hypothesis is presented that performance in the altitude control task is a function of the signal-to-noise ratio within the flow field--where signal refers to optical activity resulting from change of altitude and noise refers to optical activity resulting from other sources. An analysis of the flow geometry is presented to illustrate how the motion of the observer and the position of texture elements combine to determine the optical information available to the observer. The object recognition task involves the discrimination of 3-dimensional wire-frame forms using the information available in dynamic occlusion. A key manipulation within this task was the mode of observation. Observers were either active (they could manipulate the object using a joystick to produce dynamic occlusion) or they were passive (they could observe the motions produced by the active observer, but they could not act on the display to produce dynamic occlusions). Three experiments are presented. The most important result was that no differences were found as a function of mode. In all three experiments passive observers performed at least as well as active observers.