Set-up for the Nijhawan 
experiment: from the other side 
of the barrier the observer 
views, through the opening, 
each figure suspended in space.  
The projections of the figures 
onto the picture plane 
correspond to the retinal 
images produced by the figures.

    In a recent ingenious 
experiment conducted at 
Berkeley on the Müller-Lyer 
illusion, Romi Nijhawan used 
three-dimensional figures with 
the arrowheads in a different 
plane than the shaft, as seen at 
left. However, he placed the 
figures in such a way that the 
retinal images reversed the 
actual state of affairs (e.g., the 
shaft in the figure with the 
reversed arrowheads creates a 
retinal image with normal 
arrowheads). When an observer 
viewed the objects with only 
one eye, the true depth could 
not be perceived, and the 
illusion followed what one 
would expect based on the 
retinal images. But with 
binocular vision, which 
allowed veridical perception of 
which way the arrowheads 
pointed, the illusion was 
reversed. Hence the illusion is 
not based directly on the 
retinal patterns normally 
created by the illusion figures, 
but on how the figures are 
perceived. This experiment 
thus rules out a number of 
theories that are based on the 
retinal images these figures 
produce. Instead, to my way of 
thinking, it supports the 
incorrect comparison theory. 
The reversed arrowheads are 
perceived as if to extend the 
shaft’s length and that makes it 
difficult to isolate the shaft 
from them.