Whether or not we have an 
incorrect memory schema of an 
object in most cases, there can 
be no doubt that we have 
difficulty in achieving imagery 
from memory that is good 
enough to allow us to draw 
many objects adequately. The 
kind of object to which this 
difficulty applies is, I believe, 
precisely the kind that was 
discussed above in the case of 
copying—–namely, one in which 
the shape is qualitatively 
subtle, such as the human 
body, a dog or other animal, a 
geographical shape, a bicycle, 
or an automobile. The kind to 
which it would not apply is one 
in which the shape is 
relatively simple or unsubtle—–
a book, a simple wooden chair, 
the outline of a house, a 
staircase, a lamp, or the like. It 
is true that our own drawing 
becomes an object for 
recognition, and this is all the 
more important in the case of 
drawing from memory than it is 
in copying. Since the memory 
schema may well be a rather 
faithful representation of the 
object, even if it is not easily 
evoked without the help of a 
stimulus figure, we can 
compare the drawing to the 
schema. The mismatch tells us 
the drawing is not right, but 
not necessarily in what 
respect. In theory, if we 
continued revising the 
drawing, we should be able to 
achieve an excellent likeness. 
However, as was noted with 
respect to copying pictures, few 
if any people ever proceed in 
this way.