Other things being equal, we 
also tend to group together 
units that are similar to one 
another. Similarity can be of 
color, of lightness, or of size. 
The similarity of the shapes of 
the units to one another is 
more problematic. Apparently, 
small line segments of one 
orientation tend to be 
perceived as a group when 
embedded in other line 
segements of a different 
orientation. However, arrays of 
T-shaped units, composed of 
horizontal and vertical line 
segments, are not readily 
perceived as a segregated array 
when embedded in arrays of L-
shaped units, which are 
composed of similarly oriented 
segments. Differences in 
orientation seem to be of great 
importance in the grouping of 
arrays of units, as well as in 
forming boundaries segregating 
otherwise similar unit arrays. A 
well-known test for color 
vision is based on segregation 
as a function of color 
similarity. If one is colorblind, 
a grouping by color similarity 
will not occur. In experiments 
using this technique, it has 
been found that similarity of 
line orientation of the units or 
of parts of the units will lead to 
segregation, but that, 
generally, similarity of form 
where line orientation remains 
the same will not lead to 
segregation.

    Similarity is a principle that 
is clearly important in 
camouflage, such as in 
countershading in fish. A fish 
is typically dark on its dorsal 
surface so that, when viewed 
from above, it blends with the 
dark appearance of water 
below; when viewed from 
below, its light ventral surface 
blends with the light color of 
the water’s surface and the sky 
above.