A crucial experiment to decide between these views is to make the eye movements control the movements of a target so that its image remains on the same part of the retina even when the eye moves, i.e. to produce a stabilized retinal image. A way to do this is to attach the target (and a lens which focuses it) onto a tightly fitting contact lens (Fig. 2). The whole system then moves with the eye and the retinal image is fixed on the retina. A more elaborate apparatus (Fig. 3) enables a wider range of targets to be used. The target is in the projection system P. The beam from P is reflected from the mirror M, which is attached to a contact lens worn by the subject. It enters the eye after passing through the telescope T. When the eye rotates through an angle θ, the beam from the mirror M rotates through an angle 2θ. If the telescope T has an angular magnification of ½, the beam which enters the eye rotates through θ and its image falls on the same part of the retina even when the eye moves.
When a person views a stabilized image, the structure of the target fades out in two or three seconds and the field appears dark grey or black. If the target has no sharp boundaries between areas of high contrast, the field remains black so long as stabilization is maintained. Thus the retinal-image movements are essential to normal vision, as was suggested by Marshall and Talbot.
If the target does contain sharp boundaries between areas of strong contrast, it reappears intermittently. These reappearances are usually hazy and sometimes fragmentary. There has been some controversy whether these reappearances are due to imperfect stabilization or to a weak visual signal which remains even when the image is stationary on the retina. Sharp after-images may be imprinted when a target with sharp boundaries is illuminated with a brief, but strong, flash of light. These after-images, which are certainly stationary on the retina, exhibit the same hazy, fragmentary, and intermittent reappearances. This strengthens the view that a weak secondary signal remains even when the retinal image is accurately stabilized.
When a target consists of a line pattern (Fig. 4a), fragmentation of the stabilized image is observed (Fig. 4b, c, d). Fragmentation is not purely random, but the factors which determine what part of a pattern is seen at a given moment are not understood. Pattern units which are seen — or not seen — as a whole can be identified. The circle is one such unit and a complete circle may be seen even when the target is an incomplete circle. Fragmentation supports the ideas of those who have postulated the existence of pattern units (including the Gestalt school) but does not support any previously proposed scheme in detail. It seems probable that the human visual cortex contains cells which respond to particular pattern elements. The signals which reach these cells when the image is stabilized are very weak and difficult to distinguish from a background discharge of 'noise'. From time to time the signal in one cell is recognized as greater than the noise, and the corresponding pattern element is 'seen'. Only very rarely do sufficient of these cells have signals above threshold simultaneously so that the whole target is seen.
Suppose that a target consists of a red centre surrounded by a green annulus (Fig. 5), with the outer boundary unstabilized and the inner boundary stabilized. Then the whole field appears green. The signals from the outer boundary give information that immediately within this boundary the field is green. Little or no information is received from the inner boundary, so the logical deduction is that the whole field is green. The target is 'completed', as happens in normal vision with a target which extends over the blind spot.
A more subtle example of the same process is obtained when a subject views a stabilized image of a large circle — 38 cm (15 in) or more — divided by a diameter into two parts, light and dark. The boundary may fade first in the outer region while it is still seen in the central region. The available information cannot then logically be reconciled, because it is possible to go from a light to a dark region without crossing any perceived boundary. In this situation, the brain struggles to reconcile the irreconcilable. Various shapes of hazy boundaries appear briefly. Finally the whole field goes grey — and a 'logical', but useless, picture is obtained.
If eye movements operate in the way indicated by the experiments we have described, then the main visual information accepted by the visual system comes from receptors near to boundaries between strong contrasts of illumination or of colour — if there are any such boundaries. This may account for the extent to which an artist can convey both form and sensuous feeling by means of a drawing in which just a few lines indicate the boundaries (see art and visual abstraction).
An animal, in order to survive in a natural situation, needs to keep in mind prominent features of the area in which it is placed and to give instant attention to any movement in the visual field which may reveal predator or prey. The visual receptors that respond mainly to changes of illumination filter the visual information so as to retain the permanent background but to give great prominence to any change, even in the periphery of the field. There is thus an inbuilt bias in favour of the information that is most important for survival.
Fig. 1. a. Light–dark boundary superimposed on a schematic regular array of retinal receptors. b. Receptors that receive fluctuating signals when boundary is given a small oscillation.
Fig. 2. Direct attachment apparatus with external contact lens mounted on a stalk.
Fig. 3. Telescopic system involving a mirror that is caused to rotate by the eye.
Fig. 4. a. Target. b, c, d. Fragments seen at different times when the retinal image is stabilized.
Fig. 5. Target with inner boundary stabilized, outer boundary unstabilized.
(Published 1987)
See visual system: organization
— Robert William Ditchburn
- Bibliography
- Ditchburn, R. W. (1973). Eye-Movements and Visual Perception.
- Yarbus, A. L. (1967). Eye-Movements and Vision (trans.).
