[From: Bruce Nevin (Thu 931118 16:28:04 EST)]
Bill Powers (931028.1115 MDT)
Mary pointed out off-line that "edible" is an example of a
nonverbal category. While an animal or person can control for
eating things by learning one at a time which specific things
make you sick, which leave you starving, etc., it would be much
more efficient to be able to perceive a class of things which
prove satisfactory as things to eat or are most prudently
avoided. Then, at lunch-time, one need only activate a search for
"something to eat," a reference-category which will be matched as
soon as anything edible is perceived. Naming isn't necessary for
making use of such a category perception.
I am hungry. Any particular perception that would reduce that error is a
member of the category "things that assuage hunger." Even if I have
never encountered a particular given perception before, after a suitable
process of trial and (often) error I recognize it as "food", not by some
process of creating a new perceptual function for it on the category
level, certainly not by any rationcination, but simply by the very
process of satisfying hunger (without ill side effects), something I
could do only in memory and imagination while I was hungry and looking
for food. Categorization is present whenever error may be reduced by
diverse means. (Note that diverse means are discrete, variable means are
continuous.)
The perception of a category is present whenever I imagine controlling
some perception without imagining precisely which of diverse possible
means I use for the purpose. "Something to pry this cover up with."
I do not see a need for a separate category level.
What am I missing?
Bill Powers (931105.1015) --
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.
[...]
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.
Looking at symbolization as a social process, the simplest case is an
exemplar, a member of the category used as a token to represent any
member of the category or even the category itself. I put a banana down
to represent food. I make noises and movements of eating.
At one remove is an iconic sign: a clear representation of an exemplar.
The representation could be sculptural, graphical, gestural, or
conceivably in some other medium. It must be recognized by sender and
receiver as a representation of an exemplar standing for members of a
category or for the category itself. I put a banana-shaped piece of
found wood down to represent food. I scratch a banana shape in the dirt.
I utter a syllable "nuhm" that conventionally stands for noises and
movements of eating.
At a farther remove is a symbol: a conventional abbreviation. Over time,
this may become increasingly stylized and remote from iconic
representativeness. The elements of Chinese ideograms show this. The
origins of the Phoenician alphabet, whence most of the other alphabetic
systems we know of today, probably were originally iconic. I utter a
word meaning (to those who know my language) something that we might
translate into English as "food" or "that which one eats."
Symbols are needed for social communication and cooperation. Symbols are
not needed for logic-level perception. All that is needed is the
equivalent of an imagined perception passed up from some ECS. We might
label this perception "food" or "assuaging hunger" when we analyse the
system from the outside, but the logic-level ECS does not need to do
this. All it needs is the signal. And to provide that signal, control
systems that are involved in assuaging hunger need only imagine that they
are doing so and pass a signal along to the logic level ECS. All that is
needed is an additional function for the imagination loop. No category
level required.
What am I missing?
Bruce
bn@bbn.com