Two-dimensional filtering, oriented line detectors, and figural aspects as determinants of visual illusions

Abstract

Quantitative data of Muller-Lyer illusions from the literature were analyzed according to three different models. All three models predict the illusion effect, although with different magnitude and different parameter dependency. First, a filter model describing a certain amount of blurring of the retinal picture seems partly responsible for the observed illusion. With reasonable estimation of the filter constants, however, a sufficient magnitude of illusion cannot be obtained. A second model of oriented line or bar receptors is even less effective in explaining the observed length illusions. A third model, consisting of a size-constancy operator triggered by depth cues, may predict effects larger than actually observed. It is concluded that figural aspects such as depth-inducing cues are mainly responsible for the illusion effects observed in Muller-Lyer figures.

Geometrical illusions: study and modelling

“…Comparison of the effects of various analyzed models and experimental data fail to establish a quantitative match: Eijkman et al (1981) found that a model consisting of a size-constancy operator triggered by depth cues predicted effects larger than those actually observed; a filter model seemed partly responsible for the observed illusion, but a sufficient magnitude of illusion could not be obtained; oriented bar detectors were even less effective in explaining the observed length illusions. Therefore some investigators (Day 1972;Restle and Decker 1977;Pressey and Di Lollo 1978;Eijkman et al 1981) suggest that simple straightforward explanations are questionable and that there may be more than one mechanism contributing to the observed effect. Others, though (Predebon 1992), still emphasize the lack of necessity of invoking different causal mechanisms.…”

“…Therefore we have constructed a filter model that employs the neurophysiological data on the properties of retinocortical pathways, the principles of spatial organization of simple and complex cortical receptive fields, and the scheme of their distribution in the retina -which makes our model essentially different from that of Eijkman et al (1981) or Ginsburg (1975). The variability of the parameters of the filters of our model depends on the eccentricity of the visual field.…”

Influence of color and luminance on the Müuller-Lyer illusion

“…When arrows and line are of a different color, there is, due to chromatic aberration, more retinal blur. This blur could lead to a larger illusion (Coren, Ward, Porac, & Fraser, 1978;Eijkman, Jongsma, & Vincent, 1981;Zucker, 1980)and a greater variability.…”

The effect of perceived length on visuomotor localization

“…The overestimation of the array length could have lead to overshoots with spectacles and with contact lenses. Illusions of extent such as the Oppel±Kundt illusion have been attributed to errors in judging the depth of the illusory ®gure and to spatial ®ltering in the visual system (Eijkman et al, 1981;Bulatov et al, 1997;Bulatov and Bertulis, 1999). If the pointing overshoot is related to an illusion of length, then this implies that extraretinal eye position information is not the primary determinant of visual direction in the presence of the horizontal array.…”

“…The Mu Èller±Lyer illusion has been attributed to overestimation of ®gure distance (Eijkman et al, 1981). This overestimation triggers the size-constancy operator mentioned above, and as a result the size of the ®gure is also overestimated.…”