Jim Balter <jqb at sandpiper.net> wrote in message
news:378D9AE0.95566048 at sandpiper.net...
> flefever wrote:
> >
> > Well...not to put too fine a point on it, but... Do you really mean to
> > say that ability to "identify the object in the world" is the critical
> > requirement?
>> That does seem to be what these "symbol grounding" folks seem to mean.
> But I think it clear from "brain in the vat" considerations that
> this is the wrong track.
I'm not sure how. While a brain in a vat might have qualia, its qualia
cannot stand in relationship to objects in the world in a way consistent
with
reference except coincidentally. Qualia don't have meaning.
> "snow is blue" might "really mean"
> (I'm using their language) "7 is greater than 91", but this would have
> no consequence in re whether we understand the sentence "snow is blue".
I'm sure people on all sides are confused about this. If "really mean" is
to have any meaning it is due to intentionality. If the system lacks
intentionality
its performances can't stand in relationship to the system in a way where
the system really means anything.
Determining if a system has intentionality may not be possible. In certain
cases it is possible to determine that a system does not intend a particular
interpretation of its performance. A system cannot intend an interpretation
that relies upon entities to which it has no reference (except
coinsidentally.)
> Looking at a computer and seeing that it's "snow is blue" corresponds
> to our "7 is greater than 91" and therefore denying that it
> "understands" is a confusion of viewpoints and levels of description.
> People only get away with it because we don't yet have a formal
> model of human cognition where we can point to the obvious counterpart.
Are you relying upon your intuition to tell you this?
That alternative interpretations exist only indicates we lack the tools to
determine intent. One would hope that with sufficient performance
alternative
interepretations could be ruled out. With ungrounded computational systems
this won't happen since the most obvious alternative interpretation relates
to the internal structure of the system. (For the most part this is true
for
humans as well as computers. The difference is that humans are connected
to the world in a way in which they can make reference to it. Some machines
are connected to the world in a way in whch they can make reference to the
world if they can make reference at all.)
> > Would it suffice to add some input resulting from an
> > object's knocking at the room's front door? Would this allow a
> > "semantic association"? (e.g., "hey, Chinese room: you know that thing
> > we call a 'solid object', that is related to this, that, and the other
> > thing we talked about? well, that's whatt just bumped up against your
> > door."))
>> Indeed, it's all a matter of *relationships*. "symbol grounding"
> is just another form of absolutism.
>> > >Just look at Jim Balter, who even plonked me now
> > >because I told him (via PM) that I don't consider insults as
> > >convincing arguments.
>> I plonked you because ...
What in the heck does "plonk" mean? I must lead a sheltered life.
I've never seen it before. Webster's 10th says its a variation of "plunk"
and "plunk" says it means "to come out in favor of something." I don't
see how coming out in favor of something someone said could be
considered an insult so it doesn't seem likely this is the intended
interpretation. I might be plonking you without even knowing it but
if I wrote "I plonked you because..." you could be sure I didn't
intend what you intended when you wrote this sentence because I
don't know what it means.
