[From Bill Powers (931105.1015) --
Rick Marken (931105.0800) --
I may just be categorically challenged, but I still don't
have a clear idea of the difference between a "category"
perception (as a perceptuual type currently located at level
seven of the HPCT model) and any other categorical perception
(for example, all the perceptual "constancies" where some
aspect of a perception -- such as shape [configuration] --
remains constant despite variations in lower level aspects of
the perception -- such as the component sensations of the
shape).
I'll try to explain the distinction I'm trying to make. I wish it
were clearer. Maybe it shouldn't be made in the first place.
First, you're right in saying that at all levels, all perceptions
that are produced by a many-to-one transformation involve what we
call categories. Something remains the same while its components
can be in various states. But this specific relationship between
levels is not itself perceived. You have to have the concept of a
category with interchangeable members to PERCEIVE THAT a category
exists. When you apply this mode of perception to lower levels in
the diagram of the hierarchy, you can see this categoryness, even
though the lower levels themselves don't experience it.
Somehow, category perception involves both the thing categorized
and the identity of the category. That is, you can look at one
specific screwdriver and see not only that it is what it is, but
that it is "a screwdriver," an example of a class of things. This
class includes members of very different appearance which would
never be mistaken for each other if examined directly (i.e., at a
lower level of perception). Martin Luther King and B. B. King are
both members of the category "black," but in almost every other
respect they differ from each other.
I thought of this level originally because a way was needed to
create symbols standing for lower-level perceptions, symbols that
could be used as words or other symbolic indicators. As Martin
Taylor has observed, we tend to use symbols in an either-or way:
that is, you are either "inside" or "outside" the house; there is
no word for "both inside and outside." The word "inside" does not
designate any particular perception but a whole set of
perceptions of differing amounts of insideness. At the category
level, they are all the same perception.
At lower levels we perceive specific things in variable states.
Each perception is simply itself, as we experience it, now. If we
see a 1-inch cube and then a 1.1 inch cube, these two perceptions
have nothing to do with each other. They are not different-sized
examples of the same perception. The concept of "the same
perception" doesn't exist at the configuration level. Every
experiential field is unique and isolated from every other
experiential field.
What the category level does is reduce a world of continuous
variation and complete uniqueness to a far smaller number of
discrete symbols, which can then be used in place of a range of
lower-level perceptions. You can look at a dog or a mouse and
perceive the label "animals." Then, at higher levels, you can
reason about "animals" without the reasoning being confined to
the properties of either the dog or the mouse. Or you can look at
a few people and categorize them as "psychologists." Given that
label, you can then reason about "psychologists" without
considering the differences between actual examples of
psychologists.
Perhaps in choosing the term "category" I have pulled a red
herring across the trail. A more explicit term might be "naming"
or "symbolizing." I had thought of those terms, but at the time I
felt that they might be _too_ explicit, only examples of some
more general operation. But now I am wondering whether they
shouldn't be used to designate this specific level, because it is
the function of naming or symbolizing that seems to be what we
keep coming back to: the substitution of a general-purpose
variable for the states of many more explicit analog variables.
I am still troubled by the ease with which we can name things:
let "X" be everything with drawers that have socks or shirts in
them. Maybe such fanciful categories aren't really useable until
they become habitual ways of perceiving the world, so the
apparent ease of inventing categories isn't operationally
significant. It may be much harder to acquire a new label that
has enough permanent meaning to be used in a control process. Yet
when we program, we can assign "dblprmpt" as the name of a
subroutine that prompts the user twice, and have no great
difficulty in using that name immediately in subsequent
programming, knowing exactly what operation it means.
Bob Clark has been doggedly insisting on bringing memory into
practically everything, and I've more or less ignored that
because memory is in principle a part of every control system and
doesn't need explicit mention in many cases. But this may be a
case where we must explicitly refer to memory. Perhaps this level
that we're trying to define relies heavily on memory
associations. Martin Taylor has suggested association as the
result of positive-feedback cross-connections among systems at
this level; I think the concept is promising, but that memory
association may be a more general answer. Memory associations are
basically one-way; we have to create both directions
independently, as far as I know. Also, I have a problem with
positive cross-connections in that I think they would tend just
to drive all the signals either to maximum or to minimum.
However, real-time association has to play a part if we can
create labels _ad libitum_.
Our discussions of this level of perception may become clearer if
we think of the problem as that of creating and using labels to
stand for lower-level perceptions.
ยทยทยท
------------------------------
In my experience, the Necker cube alternates between two
mutually exclusive (and discrete) states; one corner of the
cube is either pointing "toward" me or "away from" me. So the
cube is always in one of two (discrete) states-- "toward" or
"away from"; and never in both (mutually exclusive).
I've just thought of another explanation for the Necker cube
phenomenon. The actual figure presented to the eyes is a two-
dimensional drawing with no abiguities in it. To make it into a
three-dimensional figure, depth information must be _imagined._
In looking at a particular corner, one must imagine that it is
closer to the eyes or farther from the eyes, out of the plane of
the paper. If this imagined depth perception is carried by a
single imagination signal, then that signal can have only one
magnitude at a time. This would explain the mutual exclusiveness
without invoking a flip-flop action. This could also explain the
face-vase figure-ground switch: the figure is whichever one is
imagined to be closer than the other, the background.
However, this doesn't explain the hag-beauty switch (Boring's
marvellous drawing). In this illusion, one can feel the
perceptual system making its way from one stable perception to
another with a potential barrier between them. It's clearly a
mutually-exclusive situation and requires perceptible effort to
make the switch. I think this is evidence that some perceptual
systems are a lot more complex than a flip-flop, and that we have
some way of altering the way the same perceptual unit is being
used. Yet some perceptual units are clearly independent and can
be used freely in all possible combinations. There is a lot more
to the process of perception that our feeble efforts have
revealed.
The harder this gets, the more I'm inclined to stop all this
guessing and get back to simple experiments. Although it's almost
irresistible to try to look ahead to what we might find.
--------------------------------------------------------------
Best,
Bill P.