The Poggendorff Illusion: An Illusion of Linear Extent?

Abstract

In the standard version of the Poggendorff figure a transversal intersects two parallel verticals and the segment of the transversal between the two intersection points, A and B, is not shown. The two portions of the transversal outside the parallels then seem to be misaligned. Besides this illusion of direction, there is also an illusion of size, the distance AB being underestimated in the standard figure. The influence of configural components in determining this spatial distortion of the Poggendorff figure was examined by having subjects reproduce the inner oblique (at 45 degrees) extent AB in variations of the figure. This distance was found to be underestimated in Poggendorff variations which contained parallel (vertical) components that formed an acute angle with AB; and the underestimation increased as the number of these components present in the figure increased. The distance AB was found not be significantly distorted in figures which contained only those parallel (vertical) components that formed an obtuse angle with AB, yet their presence in the figure tended to counteract the underestimation. When the transversals were omitted, the underestimation was found to increase. The findings are interpreted in support of an explanation that reduces the Poggendorff effect to those factors which mediate the Müller-Lyer illusion.

Matching gaps between misaligned pointers demonstrates that lines, both transverse and parallel, enhance Poggendorff illusions

“…They do: adding parallels increases the effect, and when parallels are present, larger effects occur with line transversals than with dots. Note that this last finding seems to negate a recent theory of the Poggendorff effect couched in terms of Miiller-Lyer illusions (Greist-Bousquet & Schiffman, 1981, because the Miiller-Lyer components essential to that theory occur in the figure both when the transverse segments are lines and when they are dots.…”

Visual illusions with acute and obtuse angles: Configurational effects and observer strategies

“…Two experimental groups recently have attributed the Poggendorff illusion obtained with the full figure and its amputations to an underestimation of the distance between the inner tips of the obliques (Greist-Bousquet & Schiffman, 1981;Quina-Holland, 1977), following the earlier suggestion by Zanuttini (1976) that the parallels attract, or assimilate, to produce the apparent misalignment. Quina-Holland proposes that parallel attraction explains the illusion, whereas GreistBousquet and Schiffman prefer to explain it in terms of another illusion, the Mueller-Lyer.…”

The logic of misperceived distance (or location) theories of the Poggendorff illusion

“…Thus, v retains its meaning as developed in the discussion concerning Figure 2. However, in Figure 3, the termination point of the perceptual displacement vector is no longer constrained to lie along the imaginary intertransversalline (nor is objective intertransversal distance the sum of perceived distance plus pdv length) as in Greist-Bousquet and Schiffman (1981) and in Figures 2A and 2B of this presentation. Therefore, the tip of the pdv, which represents the perceptual location of the transversal intersection associated with the right half of the display, is not specified in advance.…”

“…An array of explanatory principles have been proposed for this effect. Greist-Bousquet and associates (e.g., Greist-Bousquet & Schiffman, 1981, 1985Greist-Bousquet, Schiffman, Dorsett, & Davis, 1989a) have been actively promoting a distance-misperception hypothesis grounded on the Milller-Lyer (arrowhead) display, arguing that misperception of the distance between the arrowheads is applicable to the Poggendorff display. They have been challenged by Trueman and Wilson (1989) and Day, Jolly, and Duffy (1987).…”

“…Since Greist-Bousquet and Schiffman (1981) attributed the Poggendorff illusion to the perceptual shrinkage of the objective intertransversal distance as the result of a wings-in Miiller-Lyer effect, Figure 2A represents the assumed perceptual displacement. The two transversals (including the attached vertical lines) are displaced toward each other by an amount v, half of which (Le., v/2) is attributable to the displacement of each side of the display, leaving the shortened perceptual distance, d. Here, lowercase letters will always represent perceptual variables.…”