During evolution, motion perception was probably shaped by selective pressures that were stronger and more direct than those shaping other aspects of vision. . . As a result of such selective pressures, our visual systems contain neural mechanisms specialised for the analysis of motion (p385).
In their comprehensive review of aspects of visual motion analysis from a computational perspective, Hildreth and Koch (1987) point out that:
The pattern of movement in a changing image is not given to the visual system directly, but must be inferred from the changing intensities that reach the eye. The 3-D shape of object surfaces, the locations of object boundaries, and the movement of the observer relative to the scene can in turn be inferred from the pattern of image motion. Typically, the overall analysis of motion is divided into two stages: first, the measurement of movement in the changing 2-D image, and second, the use of motion measurements, for example to recover the 3-D layout of the environment. It is not clear whether motion analysis in biological systems is necessarily performed in two distinct stages, but this division has served to facilitate theoretical studies of motion analysis and to focus empirical questions for perceptual and physical studies (p480).
Warren (1995) goes further and stresses the importance of movement and action to perception as a whole:
Traditionally, the problems of perception and action have been treated as logically independent. It has been assumed that the goal of perception is to recover objective quantities such as size, distance, shape, colour and motion, yielding a general-purpose description of the scene that can provide the basis for any subsequent behaviour. Considering vision in the context of action has important implications for this view of perception (p264).
He goes on to show that some of the anomalies of perception that are apparent when we consider it in a still snapshot way unrelated to action may not matter so much when we see perception as something that unfolds over time. Viewed thus:
. . . judgments at any instant may be qualitative or even nonveridical, and yet over the course of the act adaptive behaviour emerges from the animal-environment interaction (p264).
All this points to the fact that the detection of motion is highly significant to us. It is all the more surprising therefore that research into this aspect of our visual perception mechanisms seems to lag behind work on other aspects. While colour vision and stereopsis have received the greatest attention by researchers, as Nakayama (1985) says,
. . . it is clear that colour processing is not present in all species and that binocular vision is restricted in animals with laterally placed eyes. As such numerous animals either lack colour vision or significant binocular vision or both. No animals have been found that lack mechanisms for motion processing (p627).
From the earliest ages, motion attracts attention, and infants orient toward moving stimuli by using head and eye movements . . . The causal direction of the connection between motion and information is not known. Infants might be hard-wired to attend to moving things a useful adaptation for learning about objects and events. Alternatively, it may be information, not motion, that guides attention. Infants may preferentially attend to events more than static scenes because more or better information about spaces and objects is available to them from kinematic sources than from static ones (p346).
It appears from studies by Kaufmann (1995) that a child s ability to detect very rapid motion is fully developed soon after birth whereas the ability to detect very slow motion seems to improve gradually with age. As adults we can detect 3-6 mm of movement per second of an object when it is 1 m away. However, it appears that we substantially underestimate the velocity of moving objects (Caelli 1981 pp145-171).
Ullman (1987) suggests that the importance of motion detection has led to particular physical results:
In view of the central role of motion perception, it is not surprising that the analysis of visual motion is wired into the system from the earliest processing stages. In some species, including the pigeon and the rabbit, rudimentary motion analysis is performed as early as the retinal level (p1280).
We know too that, after recovery from damage to the human visual processing mechanisms, it is our perception of movement that returns first. Furthermore, unlike for some other aspects of vision, most of the retina is available for motion detection and the thresholds for discovery are more or less the same over the whole area (Krumhansl, 1984).
Graphics Multimedia Virtual Environments Visualisation Contents