such as the motions of objects and observers. Our findings suggest an early capacity for perceiving a coherent environment, in which stationary objects remain perceptually at rest when the observer moves, and moving objects may be detected. The basis of these early abilities seems to be a combination of optic flow information with nonflow distance information. This combination successfully grapples with the geometry of object-observer motion, in which an optical change alone may be produced in more than one way.
Taken at face value, the present results constitute another example of linkages in perception among dependent variables. Perceived motion is not strictly a function of some optical change, but results from a computation involving distance as well ( Gogel, 1982). This characterization, however accurate, may have limited generality. Recall that we have restricted our test conditions intentionally to cases in which detection of motion required distance information. Although these cases form an important subset of naturally occurring cases, in many, perhaps most, ordinary circumstances additional information is available. Perception of the moving and stationary parts of the environment may often be accomplished directly from optic flow variables, such as optical shearing or relations between optical velocity and occlusion ( Lee, 1974). Thus, it would be incorrect to claim from our results to date that moving infants' motion detection generally requires distance information. What can be said is that infants display such dependent variable coupling in cases that require it.
The role of kinematic information in motion detection fits our general view of ecological validity and risk aversion. Information in spatiotemporal change is crucial, but its specific status depends on its ecological roots. In some cases when objects and observers move, the geometry of the situation requires information in addition to optic flow; when information sources are combined, the movements and positions of parts of the environment are well-specified. Perhaps because of its ecological validity, perceptual systems capable of utilizing this spacetime geometry are present early in life.
Perception of persisting properties, such as object unity and form, and events, such as object or observer motion, comprise some of the most fundamental tasks of development. I have argued that spatiotemporal information, because of its ecological status and the evolution of perceptual systems, plays a preeminent role in the achievement of these tasks. As a closing example, we might contrast this perspective with a different one on the initial perception of motion. The notion that optical changes due to observer and object motion are not distinguishable early in life has a long history ( Helmholtz, 1909/ 1962; James, 1890; cf., Piaget, 1954). On such a view, the world would appear to move whenever objects or observers do. The resulting chaos would surely contribute greatly