Perception, 1979, volume 8, pages 589-593

The nonequivalence of abrupt and diffuse illusory contours

Barry L Richardson Division of Life Sciences, Scarborough College, University of Toronto, West Hill, Ontario M1C 1A4, Canada Received 20 November 1978, in revised form 24 April 1979

Abstract. A distinction is drawn between diffuse regions of apparent brightness and illusory, but abrupt, brightness gradients. Differences between the two phenomena are sufficient to disqualify conclusions about one based on observations about the other. It is suggested that diffuse contours may be attributable to peripheral mechanisms while sharp contours may be central or cognitive in origin.

1 Introduction Figure la has in its interior part a circular region of apparent brightness with "a somewhat indefinite border" (Kennedy 1978, p 187). Figure lb appears to have a bright triangle in its centre but the perimeter of this triangle (where the 'illusory contours' are seen) usually appears much sharper than the perimeter of the circular bright region in figure la. Illusory contours of the type exemplified in figure 1 b have been attributed to a perceptual 'completion' process (Kanizsa 1974) in which the bright triangle, seen as overlying the background (stratification) has a perimeter which effectively completes the figure suggested by the inducing elements. Kennedy (1978) asks: "Is completion necessary? Can figures entail no completion, but still result in illusory contour effects'!" (emphasis added). He acknowledges two

# »

* w * (b) Figure 1. The central region of figure la usually appears brighter when viewed directly but darker when viewed peripherally. The boundary of the central region is typically seen to be diffuse or fuzzy (adapted from Kennedy 1978). The central region of figure lb (after Kanizsa 1974) appears brighter when viewed directly. In this figure the boundary is typically reported to be sharp or abrupt. (To view figure lb peripherally, successively fixate the black dots around la, and for figure la successively fixate the black inducing elements of figure lb.)

590

B L Richardson

phenomenological classes of illusory contour by saying: "[illusory] contours can be abrupt and sharp, with a precise location, or can be more gradual and diffuse" (Kennedy 1978, p 187). With reference to figure la he writes: "Illusory brightness effects, it should be noted, are not accompanied by completions of the inducing elements—neither the triangles nor the zigzag lines call for completion to 'explain' them as visible evidence for more complete shapes. Consequently, there is no impression of an illusory figure overlapping the inducing elements", and concludes: "In sum, the figures here constitute clear cases indicating that neither stratification nor completion are necessary for illusory brightness" (Kennedy 1978, p 188; emphasis added). The descriptive shifts from the title of the paper "Illusory contours not due to completion" through "contour effects" and "brightness effects" to "illusory brightness", indicate that Kennedy has included all these effects as examples of illusory contours. However, from the fact that diffuse illusory contours do not appear to rely on stratification or completion processes, it does not follow that abrupt illusory contours are also independent of these processes, unless the two classes of illusory contours can be regarded as equivalent. The phenomenal difference between the sharp and diffuse contours exemplified in figures la and lb is not sufficient ground to assume separate underlying mechanisms but there are other differences suggesting that the visual experiences generated by the two figures are attributable to different processes. For example, Kennedy (1976) noted that some observers reported a dark central region in figure 2 and that this effect was enhanced by looking at the side of the figure, i.e. beyond the radiating lines. (In the same paper, Kennedy reported that many subjects could "see" certain areas of a figure, in this case one used by Schumann in 1904, as either dark or light depending only upon suggestions from the experimenter.) Though figures la and 2 differ in possibly important respects, they do both generate diffuse contours. If, in general, regions of brightness change which are bounded by fuzzy edges are prone to reversals of brightness difference such that the interior regions are sometimes dark

Figure 2. The central region is usually seen as brighter though some observers report it to be darker. The latter effect may be enhanced by looking at the sides of the figure (after Kennedy 1976).

The nonequivalence of abrupt and diffuse illusory contours

591

and sometimes bright (in relation to the figure's surround) and if regions of brightness change bounded by abrupt edges are not so susceptible to reversals of dark versus light impressions, we have other grounds to doubt the equality of diffuse and abrupt illusory contours. The experiment reported here investigated a number of such differences associated with figures 1 a and 1 b as exemplars of the diffuse versus abrupt contour dichotomy. 2 Method 2.1 Subjects The subjects were eight males and eight females, whose ages ranged from nineteen to thirty-five years, attending Scarborough College, University of Toronto. 2.2 Apparatus and procedure Subjects were shown figure 1, which they viewed at a normal reading distance (about 40 cm) and in normal room illumination. The following questions were read aloud by the experimenter. (i) "Some people say that they see a brighter or whiter interior in one or both of these figures. Do you see one or both of these figures as having an interior that appears lighter or brighter than the surround?" (The experimenter pointed out the interior of each figure and the surround of each figure.) According to the subject's response, the experimenter noted "la", " l b " , "neither", or "both". (ii) Question 2 was identical to question 1 except that the words "darker or grayer" were substituted for "brighter or lighter". (iii) Do you see one of the interior areas of brightness change as having a sharp or abrupt edge and the other as having a fuzzy or more diffuse edge? Specifically, does one or do both of them look sharp in outline" (The experimenter recorded one of the three responses, viz, la, lb, or neither.) "And does one or do both of them look fuzzy or more diffuse in outline?" (The experimenter recorded one of the three responses as above.) (iv) "Please look at one of the partial black discs in b, for about one second, then switch your gaze to one of the other black discs for about one second and then to the third and so on. While you are doing this, try to judge whether or not the interior of figure a looks brighter, darker or no different from its surround. Do not actually look at figure a but make the judgement while flicking your eyes from one black disc to the other." The experimenter recorded one of the responses, "brighter", "darker", "no different". (v) While looking at the black dots in figure a make the same judgement for the interior of figure b. (The experimenter again recorded "brighter", "darker", or "no different".) 3 Results and discussion It can be seen in figure 3 that only figure lb (the Kanizsa triangle) was reliably reported as having a brighter interior {p < 0-01, binomial probability test). Of the fifteen subjects who reported a brighter interior to figure lb, six also reported a brighter interior to figure la (the star) but an equal number reported seeing this figure with a darker interior. This result supports Kennedy's (1976) comments to the effect that the central region of figure la readily reverses from light to dark. The tendency sometimes to see the interior of figure 1 a as dark may well have been facilitated, in this study, by the small separation between these two figures such that the star remained in peripheral vision while the eyes (and attention) were directed primarily to the triangle. That the star's interior does look darker when viewed peripherally is supported by the data shown in figure 3 and this effect is significant

592

B L Richardson

(p < 0-01; binomial probability test p = 0-33, q = 0-67). Had figures la and lb been viewed independently, there may have been fewer subjects opting for the "darker" response to the star. However, the total absence of "darker" responses to the triangle (figure lb) when viewed directly and the small number of "darker" responses when it was viewed peripherally, are a strong indication of some basic differences in the perceptual processes involved. Furthermore, figure lb when viewed peripherally was most often reported to no longer appear brighter or darker in its interior. The data indicating that the interior region of the Kanizsa triangle has a sharp or abrupt outline and that the star interior has a diffuse outline are consistent with previous reports. In summary, the results of this experiment suggest that there are important differences between the two classes of illusory contours exemplified by figures la and lb. The diffuse contour bounds a region that may appear dark or light, with a "dark" response favoured when the figure is viewed peripherally. The sharp contour is reliably seen as bounding a brighter interior when the image occupies the foveal region but when viewed peripherally, the central triangle is not reliably seen as either dark or light. Thus, explanations for the subjective experience generated by figure lb

Figure 3. Subjects' responses to the various questions about the apparent relative darkness or brightness of the interior regions of the two figures la and lb. The comparison, for each figure, was made between its interior region and its surround, and responses were recorded for direct and peripheral viewing (see procedure section for details of questions). The first four columns (from the left) show data recorded when figures were viewed directly. The last three columns show data recorded when the figures were viewed peripherally.

The nonequivalence of abrupt and diffuse illusory contours

593

cannot necessarily be expected to account for the phenomenological impressions of figure la. Therefore, Kennedy's (1978) offering of a subjective contour not due to completion, is not so much a refutation of the completion hypothesis (Kanizsa 1974) as it is an indication that there are at least two classes of illusory contour, one diffuse and one sharp. The fact that they behave differently under varying conditions suggests that the effects have different origins and hypotheses about such origins are currently being investigated. One promising hypothesis is that diffuse contours are primarily associated with retinal mechanisms while those manifesting abrupt edges are more cognitive (Gregory 1972) in character. To illustrate a possible case of a contour that is totally cognitive, figures 4a and 4b have been included. If figure 4a appears as a cube rather than a hexagon, and if figure 4b appears as a pyramid rather than a kite, the 'contours' within these figures must have some perceptual existence and yet no brightness gradient need be reported to accompany them. Perhaps the contours are better regarded as 'imagined' instead of 'seen' but such a distinction seems gratuitous since the same status (or lack of it) could be suggested for all 'illusory' contours. For the present, the suggestion is that sharp and illusory contours are different phenomena and that sharp contours are part of a class attributable to cognitive rather than peripheral mechanisms.

fl

Figure 4. Figures 4a and 4b may appear to be a cube and pyramid respectively, rather than a hexagon and kite. If the three-dimensional experience is strongest, and the contours within the figures are not accompanied by a brightness gradient, the status of the contours would seem to be totally cognitive. References Gregory R L, 1972 "Cognitive contours" Nature (London) 238 51-52 Kanizsa G, 1974 "Contours without gradients or cognitive contours?" Italian Journal of Psychology 1 93-113 Kennedy J M, 1976 "Attention, brightness, and the constructive eye" in Vision and Artifact Ed. M Henle (New York: Springer) Kennedy J M, 1978 "Illusory contours not due to completion" Perception 1 187-189 Schumann F, 1904 "Einige Beobachtungen liber die Zusammenfassung von Gesichtseindrucken zu Einheiten" Psychologische Studien 1 1-32

The nonequivalence of abrupt and diffuse illusory contours.

Perception, 1979, volume 8, pages 589-593 The nonequivalence of abrupt and diffuse illusory contours Barry L Richardson Division of Life Sciences, S...
521KB Sizes 0 Downloads 0 Views