The effect of estimation errors, the size constancy in two different depth
cues.
The Introduction:
The Ponzo illusion also widely known as ‘railway lines’ is a visual illusion where geometrical
characteristics accumulated by the retina deceive the brain creating a fake perception. It is
suggested by Gregory (1994) that the dependent variable formulates a change in the supposed
distance implying that the size constancy is an imperative factor in this study. The aim of the
study is to imitate Ponzo’s illusion to have a better understanding in the effect of estimation
errors due to optical illusions. One of the past researches closely related to the Ponzo illusion is
the Muller-Lyer illusion, where two of the exact length lines look different due to the addition
of dissimilar directions arrows. His research was first presented in 1889, since there have been
many versions of this illusion, for examples using circles instead of arrows. Another
explanation given by R.H. Day is that the size of constancy is due to opposing cues, Day, R. H.
(1989). These illusions help psychologists establish the reflexivity of the brain when judging
length and sizes. Dr. Michael Proulx also proposed that these illusions are merely what the
brain wants us to see, Proulx, M. J. & Green, M. (2001). Likewise there is the Zollner illusion
implying that having a background influences the appearances of straight lines as it forms an
image of depth, Zollner, F. (1860). His research consisted of parallel lines with short segments
(angled). The tilted lines do not look parallel however they really are. Then angle of the short
lines compared to long lines creates a false image that various lines have different distances
due to the depth. Furthermore, a replica of the Ponzo illusion was conducted in 2001 by
Prinzmetal, Shimamura & Mikolinski, implementing a fresh theory that associating traits of
contour and position. This is merely due to Gibson’s (1937) study on the visual tilt effect. In
this experiment, the participant spent four minutes observing an oriented line and then had to
modify another line to a vertical orientation. This gave an opening to a new argument that
unlike Pressey and Epp’s, (1992) proposition that Ponzo’s illusion was due to distortion of size;
the real illusion was due to orientation of implicit lines, therefore contradicting assimilation
theory and encouraging tilt constancy theory. Another well-known research to be an excellent
example of the Ponzo illusion is the ‘Moon Illusion’, station that objects horizontal to us
appears bigger than objects vertical to us, Spekkens, K. (2002). Therefore, in this research, the
hypothesis is that the stronger the cues the higher the estimation error meaning that the illusion
will be greater, making it a directional hypothesis. The null hypothesis would be that there will
be no variation between the two cues and the results will be the same for both strong cue and
weak cue.
The Method:
Design:
The dependent variable in this experiment is size constancy which also is the estimation error.
The independent variable is the depth cue consisting of two cues: strong and weak. This is
related as the two variables have natural connection. The illusion has four lines, two which are
equal lengths consecutive but merging and another two horizontally parallel lines; one at the
, end of the merging, looking longer and the other line on the opposite side appearing shorter.
The participant has to adjust the lower line to match with the upper line; both measured in
pixels on a computer screen in two diverse conditions. The first condition also knows as the
strong cue is trying to adjust the length of the lower line with a background containing trees
and bushes. The second condition, weak cue is trying to match the length white a without a
background, so it’s just a plain assortment of lines. There are many advantages to this design as
it is very repetitive making the experiment more accurate as well as having more authority over
the perplexing variables. Obviously, there are disadvantages such as the participant could get
tired by the time they do both conditions.
Participants:
For this research 23 first year psychology students from City University London, situated in the
heart of London at Northampton Square, took part in the experiment, amongst them were three
male and 20 female. The age range was 18-25 with a mean age of 18.
Apparatus:
For this experiment, computers screens were used to present the illusion with a software
package along with a mouse to move the cursor around the screen to change from one cue to
another.
Procedure:
So firstly, participants were asked to sit in a comfortable position adjusting the computer screen
to an appropriate distance from the eye. Participants were explained on the various operations
on this experiment that were to take place. They were also allowed to ask any questions they
had about the research and experiment. Experiments allowed the participants to have four trials
to settle in with the software and the keys and practice the research. There were eight trials on
the screen with two converging lines and two parallel lines. The task was to accustom the
length of the bottom line to match the length of the top line. The cursor was used to click on
‘Shorter’ and ‘Longer’ to adjust the lines to suit the top line. Every time the participant pressed
‘shorter’, the line decreased, and every time they clicked on ‘longer’ the line increased in
length.
This had to be done quick as the participants were not allowed to see the results, therefore they
were asked to turn away from the screen, during which the experimenters recorded the results
on paper. This procedure was done in two different variation; strong cue (supplementary
scenery) and weak cue (no scenery).
Refer to appendix for exact instruction.
The results:
After having used to Wilcoxon test, ‘non parametric equivalent of the t-test for related
samples’, Miller, S. (1943). The T value I worked out from adding up the rankings of the less
frequent sign’s difference, I got 57. Looking at the critical value for N=21 is 59. Since this
value is smaller than my T value, the null hypothesis has been rejected since there is
significance between the extent of the illusion and the depth cues. Refer to appendix for data.
