gestalt theory

Psychological point of view that says it is necessary to consider the whole of something, since the whole has a meaning apart from its individual elements. In advertising, the implication of this theory is that a particular brand must be considered as an organized whole and not just as a total of its attributes. Therefore, if an attribute is altered in any way or if a new attribute is added, consumer perception of the brand may change radically, and consumers who have previously purchased the product may no longer continue to do so.

The idea of the organized whole implicit in the gestalt theory also applies to the advertising context for the product. If the context is changed, it is likely that the market for the product will change as well.

A theory developed in opposition to the classical theory of psychology best represented by J. S. Mill and H. von Helmholtz. In the classical account of perception, our sensory receptors analyse the energies provided by the physical world into independent, simple, but unnoticeable sensations, and the world teaches us to perceive those objects and events that would, under normal conditions, most probably have produced any given set of sensations. Many perceptual phenomena, however, seem at first to defy analysis in terms of such 'atomistic' independent sensations. As an object moves laterally or sagittally in the field of view, its apparent shape and size remain unchanged (i.e. they display perceptual constancy) even though its projected image stimulates a changing set of visual receptors; a melody sounds the same although transposed to a new key and therefore to different auditory receptors; and the form depicted by any pattern can be completely changed by changing its context.

The most important example of this last point is the 'figure–ground phenomenon', i.e. that the same outline can be perceived as different alternative figures, with very different shapes. As Edgar Rubin, a phenomenologist, noted in 1921, the region that is perceived as 'figure' appears to have a hard surface, with a recognizable shape and definite boundaries; the 'ground' is usually less surface-like, and without definite boundary, extending indefinitely behind the figure's contour. In Fig. 1a, either the vase (Fig. 1b) or the pair of faces (Fig. 1c) can readily be perceived as figure, with the other possibility then relegated to ground. For a shape to be perceived or recognized, it must be figure. The perception of objects, of depth, and of scenes rests, in each case, on shape perception (for instance, on the shapes that comprise the depth cues of linear perspective). Yet the classical theory seemed unable to account for the figure–ground phenomenon and, therefore, for shape perception.

Such phenomena implicate a Gestalt (configurational) quality in addition to, or in place of, the individual elements composing the pattern of stimulating energies.

To Gestalt psychologists, notably Max Wertheimer, Kurt Koffka, and Wolfgang Köhler, form is the primitive unit of perception, taking its properties from underlying configured brain processes ('fields'). These brain processes were thought to be direct responses to the patterned energies acting on the sensory nervous system. The latter was variously conceived as an electrical network or a colloidal bioelectric medium. Brain fields might be studied indirectly by their effects on perceptual organization, an attempt pursued most concertedly by Köhler and H. Wallach (1944) through the study of the 'figural after-effect.' After staring at the fixation point, x, in an 'inspection figure' (Fig. 2a), a 'test figure' (Fig. 2b) appears distorted to subsequent viewing, as exaggerated in Fig. 2c; presumably the latter's contours are 'repelled' from regions previously 'satiated' by the brain fields (direct currents) involved in the perception of the inspection figure.

In point of fact, however, although there is ample evidence of organization in the nervous system (in that nerve fibres are arranged in patterns that constrain their function), there is no evidence for the direct-current, steady-state, whole-figure process models of the Gestalt theories. Indeed, such steady-state brain fields seem inappropriate even as metaphors, because in order to perceive any extended object or scene, the eye must direct the very narrow region of detailed vision (the fovea) at different places, so that different parts of the object are seen in succession by the same part of the eye at a rate of about four discrete glances per second. This basic fact would result in a superimposition of different fragments of the stimulus configuration in rapid succession on any hypothesized brain field. It is hard to see how this process could be represented by the holistic steady-state models of the Gestalt psychologists.

With no viable physiological model, the Gestalt theory consists primarily of the so-called 'laws of organization'. The figure–ground method for studying these 'laws' uses ambiguous patterns of either dots (Fig. 3a) or lines (Figs. 3b and 4c) which can be perceived as forming one shape or another, reversible pictures in which the contour that separates two regions gives only one of them recognizable shape as figure at any moment in time (Figs. 1, 4a and 4b), or reversible-perspective outline drawings (such as 'wire' cubes) which can be perceived as either flat or three-dimensional objects (Fig. 5). Configuration is varied to discover what factors lead one or another figure–ground organization to predominate.

One such factor is the 'law of good continuation', i.e. we perceive the organization that interrupts the fewest lines (for example, Figs. 3a and 3b are perceived as a square wave and a sine wave, not as the set of truncated squares in Fig. 3c). Another is the 'law of enclosedness', i.e. the enclosed region tends to be figure (for example, Fig. 4a is perceived as a set of convex shapes, Fig. 4b as concave ones). It is not merely that we perceive familiar as opposed to unfamiliar shapes: letters and numbers (surely familiar) can be concealed by embedding them in a completely unfamiliar set of squiggles, if the latter provide good continuation, as they do in Fig. 4c but not in Fig. 4e.

Many such 'laws' were proposed. They are potentially of practical importance because they seem to determine whether any shape that is presented (by artist, photographer, cartographer, architect, or computer) will in fact be perceived. They seemed potentially of the greatest theoretical importance, because they were taken to show that we perceive objects and events not by learning to interpret our sensations, but because evolution has provided nervous systems that yield three-dimensional perceptual organizations under appropriate conditions. In Fig. 5a the good continuation of the line i–ii would have to be broken in order to perceive the pattern as a three-dimensional cube; in Fig. 5b, the edge i–ii must be broken to see the figure as flat.

For the most part, unfortunately, these 'laws' remain subjective demonstrations of unknown reliability. They can conflict: in Figs. 3a and 3b good continuation is in conflict with enclosedness. Because their relative strengths are unknown, they cannot be applied predictively. Many of them can, however, be subsumed under a single minimum principle, i.e. that we perceive the simplest or most homogeneous organization that will fit the pattern of sensory excitation. Fig. 5a, for example, is simpler (in terms of the number of different lines and angles that have to be specified) as a flat figure than is Fig. 5b, whereas they are equally simple as cubes (cf. Hochberg and Brooks 1960). This formulation offered a unifying thread to Gestalt theory that the 'laws' themselves did not. Because the Gestalt demonstrations seem visually compelling, as far as they go, and because of their potential importance, various objective and quantitative treatments of the minimum principle continued to be attempted (notably by F. Attneave, J. Hochberg, P. C. Vitz, and E. Leeuwenberg), all of them quite dissociated from the Gestalt theory and its brain fields.

What was new about Gestalt theory is almost surely not true: aside from the speculations about brain fields, the main Gestalt notion was that the whole of any perceptual organization determines the appearance of its parts. Although the limits of this principle were never fully spelt out, both the individual laws of organization and the minimum principle were clearly intended to be applied to an entire figure. Thus, whether any intersection in a reversible-perspective figure looks like a flat pattern or like a solid corner in space is presumably a function of the whole configuration. Such anti-elementarism surely goes too far. This point is suggested by the 'impossible pictures' devised by Penrose and Penrose in 1958, in which inconsistently oriented parts of objects combine in a single picture that is physically impossible as a three-dimensional organization yet nevertheless looks three dimensional. Fig. 6a is such a picture. The same point is proven by making a single intersection in the cube irreversible, and observing that the other intersection continues to reverse spontaneously (Hochberg 1978). Overall simplicity simply will not serve as even a vague and unquantitative explanatory principle.

What is true about Gestalt theory was not new. Not only were the Gestalt 'laws' never systematized or quantified: they were never explained by Gestalt theory, nor were the perceptual constancies, nor was the figure–ground phenomenon itself. In fact, the figure–ground phenomenon is most plausibly explained within the Helmholtzian framework, i.e. that we fit our perceptual expectations about objects and their edges to the relatively unusual situation of viewing lines on paper. If an outline is taken as an object's edge, even slight changes in viewpoint (such as head movements) from 2 to 1 in Fig. 7a would be expected to cause parts of the further surface (such as point i) to disappear behind the nearer surface; so that the definite shape of figure and the indefinite shape of ground, described in connection with Fig. 1, are characteristic of the expectations associated with the edges of objects. In this view, as Brunswik implied in 1956, the 'laws of organization' are merely assumptions about what parts of the visual field are most likely to belong to what object, under normal conditions. For example, because it is very unlikely that the viewer would be standing at that one precise point in space (point 1 in Fig. 7b) from which the edges of objects at different distances would be aligned (i) rather than offset (ii), the 'law of good continuation' simply reflects the probability that aligned edges belong to the same object.

The Helmholtzian theory that we perceive whatever most probably would produce the stimulation we receive, and that psychological structure therefore generally reflects physical structure not only still accounts best for the perceptual constancies and illusions (Gregory 1974) but it may account as well for Gestalt organization. The fact that we perceive impossible figures as three dimensional, however, and that figures which are made non-reversible at one corner are free to reverse elsewhere (Figs. 6a and 6b respectively) surely does not reflect the constraints of physical structure, and therefore is not accounted for by Helmholtzian principles without more specific amendments and extensions than have yet been undertaken.


Fig. 1. Figure and ground. The figure perceived in a can be either a vase (b) or a pair of faces (c).



Fig. 2. Figural after-effects. a. Inspection figure. b. Test figure. c. Exaggerated sketch of appearance of b after staring at point x in a.



Fig. 3. Laws of organization, I.



Fig. 4. Laws of organization, II.



Fig. 5. Organization and tridimensionality.



Fig. 6. Relative independence of wholes and parts. Object a looks three-dimensional even though the corners are inconsistently oriented and connected by an unbroken line (x). In cube b, the orientation is fixed at intersection 1, but perceived orientation reversals occur spontaneously at intersection 2 when the gaze is kept directed there.



Fig. 7. Objects' edges, figural properties, and 'good continuation'.


(Published 1987)

— Julian Hochberg

Bibliography
• Attneave, F. (1954). 'Some informational aspects of visual perception'. Psychological Review, 61.
• Brunswik, E. (1956). Perception and the Representative Design of Psychological Experiments.
• Ellis, W. D. (ed.) (1938). Source Book of Gestalt Psychology.
• Gregory, R. L. (1970). The Intelligent Eye.
• — —  (1974). 'Choosing a paradigm for perception'. In Carterette, E. C., and Friedman, M. P. (eds.), Handbook of Perception, vol. i.
• Hochberg, J. (1974). 'Organization and the Gestalt tradition'. In Carterette, E. C., and Friedman, M. P. (eds.), Handbook of Perception, vol. i.
• — —  (1978). Perception (2nd edn.).
• — —  and Brooks, V. (1960). 'The psychophysics of form: reversible-perspective drawings of spatial objects'. American Journal of Psychology, 73.
• Koffka, K. (1935). Principles of Gestalt Psychology.
• Köhler, W. (1929). Gestalt Psychology.

---