Excerpt from
L.Kleine-Horst: Empiristic theory of visual gestalt perception. Hierarchy and interactions of visual functions. (ETVG), Part 9, I

I. The edge type illusion

1. Introduction

A "subjective" (or "illusory", "anomalous", "cognitive") contour is a contour for which there exists no immediate sensory basis in the form of a line, consisting of luminance gradients. It was first described by Schumann (1900), and termed "subjective line". There are primarily two types of subjective contours: the "edge type contour" (Figure 9-1) and the "line-end type contour" (Figure 9-9), as termed by Lesher and Mingolla (1993). The first type represents a continuation of an "objective", "real" contour separating two fields, usually with the same orientation as the "objective" contour. In contrast to this, the line-end type contour joins the ends of thin lines together, and usually runs perpendicular to the orientation of these lines. This is roughly the manner used to describe subjective contours. There have been many attempts to provide an explanation for subjective contours; none of them has met with general acceptance, see Halpern and Salzman (1983), and Spillmann and Dresp (1995), for review.

For example, Kanizsa (1955, 1974) proposed amodal completion as the most important factor in perceiving subjective contours, which Kennedy (1978a) and Rock and Anson (1979) rejected. Coren (1972) and Gregory (1972) assumed that monocular depth cues are responsible for subjective contours. Ware and Kennedy (1977) demonstrated that depth perception is not a necessary condition for the perception of illusory contours. Day and Jory (1978, 1980) assumed that brightness contrast and the spreading of enhanced brightness are essential for the perception of subjective contours, but Coren and Theodor (1975) raised important objections to the brightness contrast explanation.  

According to previous attempts at an explanation that postulate particular conditions as necessary or unnecessary for the emergence of subjective contours, there are two general models that seem to be fundamentally different. According to the first model, to which neurophysiologists in particular are inclined, subjective contours can be accounted for purely physiologically by "bottom-up" processes. According to the second model, subjective contours are brought about by mental processes that, along with the sensory data, determine perception in a "top-down" manner, as "object hypotheses" (Gregory 1972). This assumption is referred to as a "cognitive contour". Rock and Anson (1979), too, proposed the cognitive explanation.


Figure 9-1. The Kanizsa square representing edge type "subjective" contours              

I shall attempt to account for "subjective contours" within the framework of the ETVG, by using the system of gestalt laws (Part 8) The goal of Chapter I is to fully determine the primary conditions that give rise to "subjective contours" of the edge type, as well as the accompanying "subjective fields". In Chapter II, subjective contours of the line-end type will be accounted for. The relevant ETVG principles are recalled only briefly.

The edge of an object's retinal projection forms a line of luminance gradients. We experience this line ("contour") as a "line of inhomogeneities". In scientific literature, the contours extending between the four "pacmen" in Figure 9-1, for which there is no immediate sensory stimulus basis, are termed "subjective contours", in contrast to the "objective contours", the lines of luminance gradients. The accuracy of this terminology is dubious, as it overlooks that the contour bordering a pacman off (as a "figure") from its surrounding field (as the figure's "outfield") is likewise an experienced, and thus "subjective", contour. Every perceptual experience is "subjective", and thus essentially different from the objective cause of the experience: the retinal image, or even the object being optically projected onto the retina. There is no reason to construct an essential phenomenal difference between the subjective "pacman-contours" and the subjective "inter-pacmen contours", with respect to their subjectivity and their contour property. The subjective contours between the pacmen are not continuations of essentially different "objective" contours. Instead, they are continuations of the perceived subjective contours, corresponding to these objective contours. A recognizable difference between these two types of subjective contours is that the locations of the subjective pacmen contours match the locations of objective pacmen contours, but the locations of the subjective interpacmen contours do not match the locations of any objective contour. If we replace the expression "subjective" (whose  usual meaning as "illusory" is too narrow for our present purposes) with "phenomenal", we consequently have to make a distinction, not only between "objective" and "phenomenal" contours, but also between "matching phenomenal contours" (MPC) and "non-matching phenomenal contours" (NPC).

Just as we replaced the term "subjective contour" with "phenomenal contour", the term "material contour" can be used instead of "objective contour", the former being more suitable in the context of the ETVG. These changes in terminology serve to remind one of the fundamental differences between the ETVG and other visual theories - differences which are important in the interpretation process. Previous theories are, at best, ontologically dualistic, i.e. two "spheres of being" are posited: a "phenomenal sphere" (conscious- ness, mind) and a "material sphere" (matter), thus leading to the notorious "mind/matter" problem. In contrast, the ETVG is the first scientific theory to have been developed within the framework of a "trialistic" world view, the "Three-Sphere Four-Level Model of Reality" (Kleine-Horst 1992d), that transcends the traditional monistic, or dualistic, scientific world view.  According to the trialistic view, there are three symmetrically structured "spheres of being" in which the totality of evolution takes place: the "material", the "phenomenal" and the "functional" sphere. For the ETVG, the trialistic viewpoint means that, among other things, we have to postulate a third class of contour, a "functional" contour that exists "between" the "material" and the "phenomenal" contour. This functional contour possesses a certain inner hierarchical structure and is the immediate condition of the holistic perceptual experience of a "phenomenal contour", an MPC as well as an NPC.

Perception (psychic consciousness, PC) can be influenced or even produced in different ways: first, directly by psychic functions (PF) at the same level; second, indirectly "from the bottom up" by the sensory stimuli belonging to the Universal Cosmic Matter (UCM), via the vital matter (VM), the physical (VF) and psychical (PF) functions; this can be seen in Fig. 9-2. A third way, "from the top down", uses activity of Mental Consciousness (MC) above PC (as seen in Fig. 10-7). These are valid for the experiencing of "subjective" contours and fields as well.

It shall be realized that the NPCs can be described on the basis of veridical perception. Figure 9-2 shows a part of the "real" factor hierarchy located in the domain of visual perception. Involved in visual perception are above all body and psyche: the body as a hierarchy of matters that can be functionalized, the psyche as a hierarchy of functions that can be phenomenalized. The figure shows the structure of the phenomenal sphere, where the gestalt qualities, i.e. the dependent correlates of the gestalt factors, form a holistic percept indicated by hierarchically enclosed ellipses.


Figure 9-2. Part of the ten-level hierarchy of 25 psychical visual factors: the six antagonistically actualized gestalt factors necessary for the holistic  static,  two-dimensional,  polar  "figure/outfield" percept
founded on two physical visual factors (Zm, Zl).

The gestalt factors are responsible for the "veridical" perception (their "Informative effect") as well as the "illusory" perception (their "formative" effect). Every gestalt factor can, by its specific gestalt stimulus, be actualized twice, once stronger, once weaker, both via adjacent retinal areas. There is a functional antagonism between the two actualizations, i.e. the stronger the actualization via the one area, the weaker the actualization via the complementary area. A phenom- enal polarity corresponds to this functional antagonism. The complementary areas via which the double actualization of a factor is relayed, form the factor`s "receptive field". Since all 17 factors at Levels 1 to 5 are memory contents, implicitly acquired in early infancy, all these factors are associatively connected with one another, with themselves, and with the factor "attention", so that, with the actualization of one factor, all 18 factors are "illusorily" actualized, to a certain degree. By this, a system of 18x18 (324) influence relationships ("gestalt laws") is established.

back to contents