symboling

[From Bruce Nevin (2001.10.08 23:50 EDT)]

Bill Powers (2001.10.08.1448 MDT) –

Another way I have described level 9, the
“logic level” for short, is as

the level where symbols are handled according to rules, […]

We agree. We’re approaching the elephant from different angles – I to
say that the formalisms of logicians don’t encompass it, you to say that
it is more encompassing than just logic – but we are agreed that it is
an elephant and neither a tree nor a wall.
Whence a can of worms. Or rather, a bucket of symbols sans bucket. What
are symbols and where do they live? Do perceptions become symbols upon
entry into the input functions at the (hypothesized) 9th level? Any
particular perception – the color yellow – may be symbolic, and it is
not treated as a category before being treated as a symbol but in the act
of doing so. Aha, you say, symbols are category perceptions. But merely
categorizing does not a symbol make. A symbol is a perception, surely
(what else?), and to it we may associate a category, that is,
whatever perceptions we in turn associate to a category, but the symbol
itself is not a member or the category, nor an exemplar, nor a
synecdochic representative (as in “tail plus bark => dog”),
nor any of the diverse perceptions which (in that millipoid hypothesis)
can satisfy the input function for the given category perception. There
is an element of the arbitrary about a symbol, and an element of
convention: Let a represent aardvarks (or the leftmost apex of any
isosceles triangle, or George Washington’s forgotten monacle). Whatever
it is, we agree, else there is no symbol. With categorization one can
speculate about ‘natural kinds’; with symbols there are no natural kinds.
Symbols are not in the hierarchical chain with events, transitions,
configurations, and so on below and sequences, programs (etc.),
principles, and system concepts above. A symbol can be any one of these
perceptions set off as it were to the side. (I say any perception: we
prefer as our symbolic tokens perceptions lower down in the hierarchy –
it is at least conceivable to say e.g. “let the principle stated in
the Golden Rule represent birch wood” but much more wieldy to let
birch wood be a symbol for the principle.) So does this setting to the
side and associating take place at the input functions of each level that
manipulates symbols, sequence and above? Seems implausible. So
where?

Wherever that is, there too is all of language. Not just those
configurations of phonemes that we call words (and their parts), but also
those configurations of word-dependencies that we call phrases and
sentences, and those configurations of word-repetition across arrays of
sentences that make discourse coherent. Those too have meanings, which
are in part referential, and in those embracing configurations what we
imagine to be stable denotative associations of word with perceptual
experience, word meanings, are found to extend and slither among and
through and across one another like … a can of worms. Those meanings –
lexical semantics in natural language – are not and never have been
determined by definitional fiat, “let ‘dog’ represent a quadruped
with four legs, usually a tail, usually fur, that may bark …” and
so on. But they do have in common with symbols this residing on a sort of
parallel track alongside non-symbol perceptions as they flow in ordinary
living experience, to which they refer or are associated. Wherever that
“alongside” is.

    Bruce

Nevin

···

At 15:01 10/08/2001 -0600, Bill Powers wrote:

[From Bill Powers (2001.10.09.0943 MDT)]

Bruce Nevin (2001.10.08 23:50 EDT)--

"" for short, is as
the level where symbols are handled according to rules, [...]

We agree. We're approaching the elephant from different angles

What
are symbols and where do they live?

I think that's not too hard to answer: they're perceptions. They exist in
the same perceptual channels where everything else we perceive exists. When
they're not actively being perceived, the potential to imagine them (to
generate internally a present-time perceptual signal in the appropriate
channel) may exist in memory.

The other question you raise is not so easy to answer: how symbols are
related to the non-symbolic perceptions for which they stand, so that the
symbol can evoke the non-symbolic perception and the nonsymbolic perception
can evoke the symbol. I think the most important thing here is to find
agreement over the nature of the problem we're trying to solve. There are
probably quite a few answers that would work; what we have to do is state
the problem more and more precisely until all the answers (but one, we
hope) have been ruled out.

Do perceptions become symbols upon
entry into the input functions at the (hypothesized) 9th level?

Is this in fact the only way symbols are used -- to indicate categories? I
have proposed that they do indicate categories, but is that _all_? I'm
looking at a person over there and thinking "Mary." Is that person over
there -- that particular familiar person, not just the class of things that
can be called a person -- a category perception? Perhaps so, with the
category perception then
elaborating into higher-level perceptions. But perhaps symbolic labels can
be attached to lower-level perceptions, too.

There is an element of the arbitrary about a symbol,
and an element of convention: Let a"" but much more wieldy to let birch
wood be a symbol for the principle.)

That's what happens to HTML files on my computers.

I have proposed that a category is simply the "or" of a bunch of
perceptions. We often justify categories by giving a rule for generating
them: all things with three legs and a place to sit, for example. But such
rules (ninth order)are already made up of categories which themselves, I
think, are arbitrary. I've referred to the category of all things that are
"mine." Is there any necessity for the things in this category to have any
direct relation to each other? No, yet I can treat them as equivalent under
certain operations such as buying and selling.

Here are some justifications for this model -- some phenomena that I think
do occur, and their significance for this interpretation of symbol-category
relationships:

If a category is the 'or' of an arbitrary set of perceptions, then one of
the perceptions could be the symbol (perception) that "stands for" the
category. All that is required is for the symbol to be included as one of
the "or"-ed elements. Then either the symbol or a non-verbal example of the
category will give rise to the same category signal. This happens very fast
because the input function is very simple. When you encounter a symbol that
is not part of any category, you immediately know it has no meaning,
because no category signal results.

To go the other way is a little more complex: given the nonverbal member of
the category, to find the symbol for it, or given the symbol, to find at
least one nonverbal member of the category. An active process is required
in either case. The basic problem is to detect the category signal when
either the verbal or the nonverbal element is presented, and then to come
up with the corresponding nonverbal or verbal element, respectively, that
will reproduce exactly the same signal. That is, read or hear "a chair" and
imaging seeing any chair, or see any chair and imagine reading or hearing
"a chair."

That doesn't handle any other relationships between symbol-perceptions and
non-symbol perceptions, but maybe it's a start.

While we're at it, this is an opportunity to re-propose a project that only
a linguist can do properly. That is to explore languages to see what
hierarchical functions are implied by features they all have in common. I'm
not talking about the Chomskyian approach, but of a PCT approach.

For example, in PCT we assume that all human beings have the capacity to
perceive sequences. This implies both being able to recognize a sequence as
matching a reference sequence, and being able to say when one sequence
differs from another. Many activities require this ability, starting with
such simple things as executing the two processes "going though a door" and
"opening the door" in the proper sequence. In mathematics, we have
non-commutative processes in which the order of doing operations (like
rotations about different axes) makes a great difference in the outcome.
The ability to perceive in terms of sequence is essential in many kinds of
behavior -- including language. In language, I therefore ask you, is the
ordering of words _always_ important? If so, then the ability to perceive
and control what we call sequence is probably an essential ingredient in
any organism that has language. By the same token, if any organism can be
shown to control sequence, then that organism possessed at least one of the
prerequisites for acquiring language.

We can also ask "What is a word?" I have proposed that a spoken word is a
perception that exists at the event level, just above transitions, and that
a printed word exists at the configuration level. This would explain why it
is that we can usually read language faster than we can hear it, since the
fastest possible speed at a higher level is always less than the fastest
possible speed at a lower level. Is this true of all languages with a
written version?

The capacity to perceive configurations or transitions would seem essential
for any organism that has language, although it would not be sufficient.
While this may seem too obvious to merit much discussion, we must still
establish that all languages entail the ability to perceive and control
configurations or transitions, or both.

I won't belabor the rest. Is it possible to give an account of all
languages in terms of PCT levels, either those I have proposed or newly
defined levels? Note that in doing so, it is not necessarily to make
reference to or propose linguistic facts per se, like "deep structure."

Best,

Bill P.