The Effect of Contrast on the Mueller-Lyer Illusion
Anne Benson 04ALBENS@alma.edu
Lisa Miller 03lkmill@alma.edu
Mary Rosalez 02merosa@alma.edu
Methods | Results | Discussion | References
Previous research indicates conflict concerning the effect of contrast on the strength of the Mueller-Lyer illusion. While some believe that a background closer in shade to the stimulus-which reduces contrast-weakens the illusion, others contend that the contrast of the background affects the illusion no more than it affects a regular straight line. This research tests the Mueller-Lyer illusion on three backgrounds with increasingly less contrast.
Previous research has shown that the strenghth of the Mueller-Lyer illusion is affected by the contrast between the background and the eye-lines: "the lightness contrast level of the Mueller-Lyer figures inducing lines affects the magnitude of the illusion as a function of the amount of contrast (Ebert & Pollack 1972). Pollack's research, as well as the research of Coren and Girgus (1978), found a significant difference between the strength of the Mueller-Lyer illusion as a function of brightness contrast. Their belief was that, "Since lateral inhibition is a function of brightness contrast, it follows that the Mueller-Lyer illusion should be a function of brightness contrast" Coren and Girgus (1978). Their findings were that the strength of the Mueller-Lyer illusion decreases as contrast decreases. However, this research contains several variables, and Butler (1981) claims that Pollack's research does not contain "necessary control groups" to determine the effect of brightness contrast alone. Thus Butler researched the contrast effect on the Mueller-Lyer illusion, using what he felt were the needed controls, and found that the contrast between figure and background had no more affect for the illusion lines than for the control lines. Because decreasing contrast is found to cause assimilation (Leeuwenberg 1982), and thus causes increasing difficulty judging line length, participants will judge any line less accurately. While Pollack hypothesized that the effect he found was the result of lateral inhibition in the retina, which caused displacement in the Mueller-Lyer figures at the intersections of lines, Butler's research shows that this is not necessarily the case. The present research is designed to re-test the effect of contrast on the Mueller-Lyer illusion, using necessary controls.
Using a Power Macintosh G3 and the EyeLines software (Beagley 2000), we set
up the Mueller-Lyer illusion experiment in the following way: We have three
tails-out Mueller-Lyer versions, one with white background and black eye-lines,
one with .67 gray background and black eye-lines, and one with .33 gray background
and black eye-lines. There are also three controls: one with white background
and black eye-lines, one with .67 gray background and black eye-lines, and
one with .33 gray background and black eye-lines. Each stimulus is represented
three times in random order in the experiment. Figure 1 shows what these each
looked like. The starting length of the adjustable line varied by plus or
minus 20%.
The participants were 3 male and 16 female college students between the ages
of 19 and 22. Participants were instructed to choose a three-letter subject
code, were instructed on how to use the EyeLines program by clicking and dragging
the mouse from side to side until the bottom line looked like it was the same
length as the top line or the shaft of the line with tails (the Mueller-Lyer
figure) presented on the computer screen.
 |
 |
 |
Figure 1. The Mueller-Lyer stimulus (A - E) and control line (F)
The mean of the stimulus on white background was 270 and the control mean on white background was 29, resulting in an overall effect of 241 for the white background. Quite similarly, the mean of the stimulus on the gray.67 background was 289 and the control mean on the gray.67 background was 48, also resulting in an overall mean effect of 241 for the gray.67 background (see figure 2). Since these two resulted in the same overall mean effect, we collapsed the two groups together and compared them to the third variable, the stimulus on the gray.33 background, which had a mean of 274, and the control on gray.33 which had a mean of 63, resulting in an overall mean effect of 211 for the gray.33 background. However, the collapsed white/gray.67 mean compared to the gray.33 mean, while it does show a difference, is not significant at p < .05 level for a two-tailed t-test, t(18) = 1.397.
Figure 2. Effect of background contrast on Mueller-Lyer illusion
As expected, the participants misjudged the length of the Mueller-Lyer stimulus much more than they did the control. When the contrast error is subtracted from the stimulus error, there seems not to have been any difference regarding the background contrast, which would disprove Pollack et al. (1972) and would support the findings of Butler (1981). The strength of the illusion seems to not be significantly affected by brightness contrast. However, Butler's contention was not that contrast had no effect on the illusion, but that it had no more effect on the illusion than it did on the straight lines. Our findings indicate that contrast brightness actually had a stronger effect on the control than it did on the illusion. The overall mean of the stimulus on gray.33 background is lower, but only after subtracting the control. If the effect of contrast on the stimulus and the control are looked at separately, as contrast decreases for the stimulus, the strength of the illusion is unaffected. As contrast decreases for the control, however, participants increasingly misjudged line length. The results seem inconclusive; perhaps a larger pool of participants would produce some significant findings.
Butler, D. (1981). Does brightness contrast really affect the Mueller-Lyer illusion. Perceptual and Motor Skills 52, 623-629.
Ebert, C., & Pollack, R. (1972). Magnitude of the Mueller-Lyer illusion as a function of lightness contrast, viewing time, and fundus pigmentation. Psychnomic Science 26, 347-348.
Leeuwenberg, E. (1982). The perception of assimilation and brightness
contrast as derived from code theory. Perception and Psychophysics 32, 345-352.