How can these seemingly 
contradictory examples be 
explained? After all, in these 
cases there is no real change in 
perceived orientation when the 
figure is tilted along with the 
head because we remain 
perfectly aware of which region 
is the top of the face or of the 
word. But, if these figures 
remain upright and are not 
tilted along with the head, a 
difficulty does arise. For simple 
material it is easy enough to 
achieve the proper description 
despite the retinal abnormality 
of orientation of the image. In 
viewing a square with the head 
tilted 45 degrees, for example, 
it is easy enough to recover the 
fact that its sides are 
horizontal and vertical, which 
is the essence of perceptual 
"squareness." But this process 
of correction is far more 
difficult to achieve when the 
stimulus consists of multiple 
parts, as is the case with the 
letters making up words or the 
features making up a face. If we 
cannot correct all these parts 
and the relation among them in 
one operation, many will 
remain uncorrected and thus 
will be seen as if their tops and 
bottoms were governed by 
retinal orientation. That leads 
to the wrong description. Our 
initial lack of awareness of the 
extreme distortion of the face 
in the left-hand photograph 
shown here is testimony to the 
inadequacy of our perception 
when the image is inverted on 
the retina. Similarly, when 
printed words are viewed upside 
down, various letters are 
incorrectly "described" on the 
basis of their retinal 
orientation--a u becomes an n, 
or a d a p, and so forth.