PeI:> To test the power of "visual reasoning", here is a rough visual
explanation on two very different ways for "symbols" to get their
meaning:
http://nars.wang.googlepages.com/wang.semantics-figure.pdf
Wow, I have to stop talking but this is really stimulating. Your
paper/illustrations are v. useful as far as they go, but they are almost
literally the tip of the iceberg. Your Experience-Grounded Semantics
represents a flower/pot as a tree or net of attached symbols
"plant - containing - blossom - round" etc
Now can we please have the VAST attached clusters/ trees of images of
flowers and pots that your brain has, and uses, to understand and process
flowers/plants - the huge image and movie databank that it also brings to
bear. When you see a plant wilting, or with a strange bit of Xmas decoration
on it, you don't, I suggest, understand that by consulting any symbolic
network - a direct image comparison of a normal erect healthy plant and a
drooping plant is what produces that "wilting" conclusion. And there's
nothing that could conceivably be (at least not normally) in your verbal
network to know that that decoration doesn't belong in February. Again the
conclusion comes from a visual comparison of most normal unladen plants with
this decoration-laden one. ["Laden", "unladen," "festooned" or comparable
concepts just won't exist in your immediate network of verbal associations].
And such embodied image processing of the world is actually (though it
certainly doesn't appear so at first) much more efficient. You'll find, I
suggest, that a lot of words you're suggesting like "containing" and "round"
don't even need to be, and aren't, there. You can see them directly from
the images - e.g.. if you see a plant in a pot without earth but with some
weird plastics or toys, you (i.e. mainly here your unconscious brain) will
know directly from comparisons of the images in your mind, that that pot
does not "contain" what it should, and not from flipping down some network
of words.
Here lies the value of Lakoff and co - a lot of this has been experimentally
tested/ supported, even though, rather like Blakeslee and body maps, we are
not culturally on top of it at all. For example, there are now a very large
body of experiments to show that you cannot think about moving in whatever
way without the relevant motor or pre-motor areas coming into play, or think
about plates and presumably flowers without the relevant image and not just
symbol areas coming into play. ( I'm not just talking about direct visual
perception of, but also thinking in words about, objects as dependent on
image-processing).
One sort of tangential but utterly crucial way you can test this is the
following:
please define "plant". Verbally.
Try it.
I think you'll find that you don't have a consistent definition of any
concept or any word whatsoever. (You might produce one today, but you
wouldn't be able to produce the same one another time, unless it has been
formally learned by heart, like your AGI definition).
Nor can you offer a *method* of defining and/or redefining "plant" or
"table" or "AI" or "intelligence" or "red" - and you CANNOT OFFER A METHOD
OF DEFINING ANY CONCEPT FROM A VERBAL NETWORK. (Sorry for shouting, but, if
true, it's so important - and the evidence of cognitive science's failure to
pin down how we conceptualise, suggests it is true)
Why? Because your definitions of plant, when you attempt them will be
largely dependent on consulting not words, but your "experience" - your
images - of plants and the rest of the world.
And there is no way to bind those diverse images into a single image, let
alone word. A word/concept is merely a label on a box containing a v.
complex and dynamic network of images of objects that cannot be reduced to a
single configuration. You have a truly vast set of images for "plant", and
somewhat less vast set for a more specific concept like "tulip." Neither can
be reduced to a single definitive image/form/ structure, or indeed set of
images etc. When you try, you always run into trouble - for example, your
"round" is false as a fixed definition of pots! And your image bank knows
that.
What you're doing above is confusing the label on the box with the image
contents, (or the tip with the whole iceberg). Nothing personal - nearly
everyone does.
-------------------------------------------
agi
Archives: http://www.listbox.com/member/archive/303/=now
RSS Feed: http://www.listbox.com/member/archive/rss/303/
Modify Your Subscription:
http://www.listbox.com/member/?member_id=8660244&id_secret=95818715-a78a9b
Powered by Listbox: http://www.listbox.com
