Search

Marr proposed a computational paradigm for studying the visual system, wherein aspects of vision would be amenable to study with what might be regarded a computational-reductionist approach. First, vision would be cleaved into separable 'computational theories', in which the visual system is characterized in terms of its computational goals and the strategies by which they are carried out. Each such computational theory could then be investigated in increasingly concrete terms, from symbols and measurements, to representations and algorithms, to processes and neural implementations. This paradigm rests on some general expectations of a symbolic information processing system, including his stated principles of explicit naming, modular design, least commitment, and graceful degradation. In retrospect, the computational framework also tacitly rests on additional assumptions about the nature of biological information processing: (1) separability of computational strategies, (2) separability of representations, (3) a pipeline nature of information processing, and that (4) the representations employ primitives of low dimensionality. These assumptions are discussed in this retrospective.

When the white disks in a scintillating grid are reduced in size, and outlined in black, they tend to disappear. One sees only a few of them at a time, in clusters which move erratically on the page. Where they are not seen, the grey alleys seem to be continuous, generating grey crossings that are not actually present. Some black sparkling can be seen at those crossings where no disk is seen. The illusion also works in reverse contrast.

Surface shape is visually derived from multiple sources of three-dimensional (3-D) information. The apparent unity of the percept creates an assumption that there are perceptual processes that resolve and combine the different channels of information within some common 3-D representation. However, integration by the explicit conversion of the information from different sources into a common representation is computationally difficult and unnecessary. An alternative, integration by association, is discussed and related to a variety of experimental results on 3-D-cue conflict.

A study is reported of the perception of visual surfaces in wire-frame stimuli generated by combinations of monocular surface contours and binocular disparity that provide differing information about 3-D relief. Observers vary considerably in the relative contribution made by the binocular and monocular cues to the perception of overall 3-D form. Without training, many observers may entirely fail to perceive surface curvature from the binocular disparity patterns, interpreting the form of the surface only according to the monocular information. For other observers, both cues contribute to the end percept, with the monocular interpretation dominating where the disparity information indicates planarity and with disparity dominating where disparity information suggests curvature and the monocular interpretation suggests planarity. Where stereo and monocular interpretations indicate inconsistent surface curvature features at a common location, more complex resolution strategies are suggested.

Subjective contours, according to one theory, outline surfaces that are apparently interposed between the viewer and background (because of the disruption of background figures, sudden termination of lines, and other interposition 'cues') but are not explicitly outlined by intensity discontinuities. This theory predicts that if the cues are not interpreted as evidence of interposition, no intervening surface need be postulated, hence no subjective contours would be seen. This prediction, however, is difficult to test because observers normally interpret the cues as interposition evidence and normally see the subjective contours. Tests are reported on a patient with visual agnosia who is unable to make the usual interposition interpretations and unable to see subjective contours, but has normal ability to interpret standard visual illusions, stereograms, and in particular, stereogram versions of the standard subjective contour figures, which elicit to him strong subjective edges in depth (corresponding to the subjective contours viewed in the monocular versions of the figures.

The tendency to interpret as figure, relative to background, those regions that are lighter, smaller, and, especially, more convex is well known. Wherever convex opaque objects abut or partially occlude one another in an image, the points of contact between the silhouettes form concave cusps, each indicating the local assignment of figure versus ground across the contour segments. It is proposed that this local geometric feature is a preattentive determiner of figure-ground perception and that it contributes to the previously observed tendency for convexity preference. Evidence is presented that figure-ground assignment can be determined solely on the basis of the concave cusp feature, and that the salience of the cusp derives from local geometry and not from adjacent contour convexity.

Apparent depth in stereograms exhibits various simultaneous-contrast and induction effects analogous to those reported in the luminance domain. This behavior suggests that stereo depth, like brightness, is reconstructed, ie recovered from higher-order spatial derivatives or differences of the original signal. The extent to which depth is analogous to brightness is examined. There are similarities in terms of contrast effects but dissimilarities in terms of the lateral inhibition effects traditionally attributed to underlying spatial-differentiation operators.