sort of PHYSIOLOGICAL abstraction. Mnemic homophony gives us, without
the addition of other processes of thought, a picture of our friend
X which is in a certain sense abstract, not the concrete in any one
situation, but X cut loose from any particular point of time. If the
circle of ekphored engrams is drawn even more widely, abstract pictures
of a higher order appear: for instance, a white man or a negro. In my
opinion, the first form of abstract concepts in general is based upon
such abstract pictures. The physiological abstraction which takes
place in the above described manner is a predecessor of purely logical
abstraction. It is by no means a monopoly of the human race, but shows
itself in various ways also among the more highly organized animals."
The same subject is treated in more detail in Chapter xvi of "Die
mnemischen Empfindungen," but what is said there adds nothing vital to
what is contained in the above quotations.

It is necessary, however, to distinguish between the vague and the
general. So long as we are content with Semon's composite image, we MAY
get no farther than the vague. The question whether this image takes us
to the general or not depends, I think, upon the question whether, in
addition to the generalized image, we have also particular images
of some of the instances out of which it is compounded. Suppose, for
example, that on a number of occasions you had seen one negro, and
that you did not know whether this one was the same or different on
the different occasions. Suppose that in the end you had an abstract
memory-image of the different appearances presented by the negro on
different occasions, but no memory-image of any one of the single
appearances. In that case your image would be vague. If, on the other
hand, you have, in addition to the generalized image, particular images
of the several appearances, sufficiently clear to be recognized as
different, and as instances of the generalized picture, you will then
not feel the generalized picture to be adequate to any one particular
appearance, and you will be able to make it function as a general
idea rather than a vague idea. If this view is correct, no new general
content needs to be added to the generalized image. What needs to be
added is particular images compared and contrasted with the generalized
image. So far as I can judge by introspection, this does occur in
practice. Take for example Semon's instance of a friend's face. Unles