The effects of illusory contour figures on the Ponzo illusion

Melissa Desjarlais and Sarah Keen

Alma College

Illusory contours are the subjective perception of edges or forms that are not delineated by such forms in an actual stimulus. One of the most well-known of these is the Kanizsa triangle, produced by three "pacmen" figures that appear to form the points of a triangle. Numerous explanations for how this illusion occurs have been hypothesized, including apparent depth theories (Coren, 1972), brightness contrast (Shapley & Gordon, 1987), and visual synthesis (Bradley & Petry, 1977). The strength of the illusion can vary with the figure being used and the components used to induce it.
Illusory contours usually are capable of influencing the strength of other visual illusions as well. Kanizsa (1976) presents cases of the Poggendorf and Ponzo illusions as being induced by illusory contour figures with similar strengths as the actual lines. Most of the studies done on this topic show that illusory contour figures are capable of producing equal strength illusions (Meyer, 1986), although the Poggendorf illusion has returned some mixed results. The Ponzo illusion, angled lines that distort the perception of line length of two interior parallel lines, has returned stronger illusion effects, although they are less than with physical lines (Meyer, 1986). This is the phenomenon that is being tested in this experiment. In addition to the illusory nature of the inducing lines, the illusory nature of the comparison lines will also be tested. It is hypothesized that when the comparison lines are induced against a background of radiating lines that the illusion will still be present.

Methods
Twenty-seven male and female students from a small liberal arts college participated in this experiment. Each participant was presented with the same set of stimuli in random order as determined by the computer. The Eye Lines computer program was used to create and run the experiment. The stimuli consisted of a set of parallel lines (control figure), a Ponzo illusion, an illusory contour triangle Ponzo illusion and its inverse (Figure 1), and a Ponzo illusion with radiating lines and its inverse (Figure 2). Each stimulus was shown three times for a total of eighteen presentations.
Participants were asked to compare the length of the two Ponzo lines and adjust the lower line so that it appeared to be the same length as the comparison above it. They were allowed to proceed at their own pace but were told not to spend too much time on any one image.

Results
The error in line segment judgement was computed for each of the eighteen stimuli and the values were collapsed within subject for each of the six different stimuli. The mean error for each of the stimuli were computed and graphed (Figure 3). As expected, the mean error for the parallel lines, one of the control groups, was the smallest. The means for the Ponzo illusion and the two Kanizsa illusions were very similar, and the standard deviations were smaller than the standard deviation for the parallel lines. The radial Ponzo and its inverse both have higher mean error scores and higher variability of scores than the other conditions.

Discussion
  The similarity in mean error scores for the standard Ponzo and the Kanizsa triangle Ponzo illusions suggests that the illusory contours had a positive effect on the perception of line length. This would support our first hypothesis that illusory contours have a similar effect as the actual Ponzo lines. The illusory contour still appeared to function even though it was crossed by the interior parallel lines. With the particular stimuli used in this case, the circles that are necessary to induce the illusory lines may have influenced the perception of line length for the interior lines, i.e.- subjects were paying attention to the circles and not the induced lines. However, because all of our participants reported being able to see the illusory lines, this possibility would require further study to be validated.
  The mean error scores for the radial Ponzo illusions are higher than any of the other conditions. This suggests that the illusory contour lines produced a Ponzo illusion, but the higher scores may be due to the fact that participants reported not being able to clearly distinguish the ends of the interior parallel lines. However, by adding more radials, it is hypothesized that there would be less variability of scores. As expected there were no significant effects for the inverse conditions in this experiment. They were introduced mainly to control for any possible impact that shading could have had on perception of the illusions.
  Further research on illusory contours and illusions could involve producing other illusions such as the Poggendorf, the vertical-horizontal or the Muller-Lyer illusions with illusory contours. There have been mixed results from previous experiments about whether or not these illusions can be produced with illusory contours. Since the hypothesis about the effect of polarity of the illusion was not proven or disproven, that hypothesis should be investigated further. Also, since in one instance the illusory contours induced the parallel lines and in the other instance the illusory contours actually produced the Ponzo illusion which contours were actually induced could have some effect on the strength of the illusion, and this could be researched in the future.

References