From newshub.ccs.yorku.ca!torn!cs.utexas.edu!usc!rpi!scott.skidmore.edu!psinntp!psinntp!dg-rtp!sheol!throopw Sat Oct 24 20:44:54 EDT 1992
Article 7378 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!torn!cs.utexas.edu!usc!rpi!scott.skidmore.edu!psinntp!psinntp!dg-rtp!sheol!throopw
>From: throopw@sheol.UUCP (Wayne Throop)
Newsgroups: comp.ai.philosophy
Subject: grounding and the entity/environment boundary
Summary: fuzzy boundary renders "direct experience" questionable concept
Message-ID: <719720414@sheol.UUCP>
Date: 22 Oct 92 00:05:06 GMT
References: <1asq47INNr9o@smaug.West.Sun.COM> <1992Oct5.195433.9320@spss.com> <718611244@sheol.UUCP> <1992Oct12.203359.8713@spss.com>
Lines: 72

: From: markrose@spss.com (Mark Rosenfelder)
: Message-ID:  <1992Oct12.203359.8713@spss.com>
: Groundedness doesn't diminish one whit when you close your eyes, any more
: than your capacity for vision does.  What could change things is if you grew
: up blind: then you'd lack grounding in (say) colors.

Agreed, but I'll have more to say about the reasons for
the lack of grounding in (as an example) colors mentioned here.

: If you want a computer whose statements about color are grounded, you must
: provide it with something like a color TV camera, an algorithm which can
: acquire experience with the world using it, and time to do so.

Two things.  I don't understand why the camera, or for that matter the
algorithm that aquires the experience, needs to run on the same
computer that is to be grounded.  And I don't understand why the
experience needs to be "with the world".

Especially:

: I think I'd allow that it remains grounded even if you
: remove the camera (tho' if the robot is conscious that might not be a
: very nice thing to do...)

If it'd remain grounded with the camera removed (and presumably you'd
agree that a human would remain grounded after having the eyes
surgically removed (eg, due to retinal cancer or other adequate
reason)), and if grounding isn't something that needs to be "kept up",
then why did the camera ever need to be attached to that computer in
the first place?  Certainly, a human can only "get grounded" in color
as above by once having eyes, but the supposition that that's the
only way seems wrong to me.

Of course, I'm even more radical, in that I wonder why the algorithm
that gains sensual experience ever needs to run, if the results of that
running can be computed by some other means (though I don't see a more
reasonable way of getting the required behavioral capabilities).
But we can leave that aside for now.

Back to why it is important to "have eyes": what about two
robots/computers that share a single camera on a network? (We can avoid
the problems of lack of interaction noted in the experiments with cats
by making the two robots/computers motor peers instead of
master/slave...) Or how about any pool of cameras that any AI can patch
into at need, but with fewer cameras than AI engines? Or how about
*more* cameras than AI engines, but still no unique asociation of
camera with algorithm or motor system? Can the AIs in such schemes "get
grounded" by use of a shared camera? Even though any given AI's share
of the camera can arbitrarily approach zero? If they *can* "get
grounded", why does the camera's NTSC (or whatever) signal need to be
"parsed" by each AI, when one could "parse" it, and (presumably) spoon
feed the predigested experience to the others?

As in the subject line, the problem I see with this notion of direct
experience being needed for "getting grounded" is that it ignores the
arbitrary nature of the boundary between an entity and its environment,
and hence it overlooks the very fuzzy, blurry nature of what it might
*mean* to have direct experience.  It only seems a sharp distinction
to us, because humans have limited ways of "getting grounded".

And if you don't think such entity/environment problems are relevant to
humans, think of siamese twins, or people who have had their brain
hemispheres segregated to control epilepsy.  We really have something
VERY much like a network of N AI engines controlling M(less-than-N)
"peripherals" (such as arms, legs, etc).  Is the left hemisphere
grounded by skills aquired by the right hemisphere's interaction with
the brainstem and peripheral nervous system?  Or not?

So, what does "direct experience of the world" really mean?  I propose
that ultimately it is an empty phrase, signifying nothing.
--
Wayne Throop  ...!mcnc!dg-rtp!sheol!throopw


