Date of Award
Doctor of Philosophy (PhD)
Professor P.L. Newbigging
Nine experiments involving 147 subjects are described. Experiments 1 to 5 were designed to produce evidence relevant to the question "do the classic Poggendorff figure and its variants activate the same perceptual process(es) or are they distinct and give rise to different phenomena?" Experiments 6 to 8 constitute an attempt to determine the ways in which the perceptual geometry of the Poggendorff differs from its Euclidean geometry. The ninth and final experiment is addressed to the cause of the illusion.
Experiments 1 and 2 show that the classical Poggendorff figure and three of its variants all produce the same illusory effect and that, for all four figures, that effect is increased as the angle of interception of the oblique decreases and as the separation between the parallels increases. Experiment 3 shows that the differences in the magnitude of illusion produced by the variants is due to the particular component retained and not the total amount of the illusion figure present. Experiments 4 and 5 extended this analysis to study the effects of practice and the extent to which practice on the variants transfered to the classical figure. Decrements with practice were obtained for all figures and the figures were ordered, in terms of magnitude of illusion produced, precisely as they were for the initial two experiments. This result is noteworthy because it shows that the effects obtained are not dependent on either the details of the stimulus, psychophysical method, or mode of presentation. Also, equal or greater positive transfer to the classical figure is obtained from the variants as from the classical figure itself. The conclusion is reached that the classical figure and the variants arouse the same single perceptual process. Unfortunately, the data do not permit the identification of that process.
The results of Experiments 6 to 8 provide a model of what the visual system does when confronted with the classical Poggendorff figure. It was determined that the presence of a single transversal is all that is required to produce the full illusory effect (Experiment 7). The angle of interception of this transversal is perceptually enlarged (Experiment 8). This perceptual representation is then projected linearly across the space between the parallels (Experiment 6). An across-experiment test shows that the model fits the data reasonably well. Although the experiments were not designed to test any theory of the cause of the illusion, the resulting model has some features that are similar to angular distortion explanations. The ninth and final experiment was direct test of Burns and Pritchard's (1971) proposal that lateral inhibition in the visual system is the basis for the distortion of cortical images and thus the Poggendorff illusion. The result was a significant reduction, but not the elimination, of the illusory effect.
In the final chapter it is noted that while lateral inhibition seems to figure causatively in the classical figure, certain configurations, particularly those that do not include acute angles, present some difficulty. Suggestions for further research which will explicate the phenomenon are also noted.
MacKay, Darius Cyrus, "The Poggendorff Illusion: Variants and Perceptual Geometry" (1977). Open Access Dissertations and Theses. Paper 735.