Hans Moravec wrote:
>
> Russell Standish <R.Standish.domain.name.hidden>:
> > We have discussed a number of scenarios whereby being too thorough in
> > an investigation will reveal a concious object as nonconcious.
>
> Every one a canard.
>
> **** Consciousness is an attribution, not an objective fact ****
>
> But any attributed consciousness is conscious in its own eyes. This
> includes your consciousness, my consciousness and theconsciousnesses
> of characters in a movie, or the character represented by a
> Turing-test-passing HLUT.
>
> Sherlock Holmes, as I read the books, is conscious in his own
> eyes and has feelings and beliefs (I can tell you what some of
> them are), and so was my teddy bear when I thought about it a
> certain way.
I think that this is not only basically wrong, it is fundamentally
useless. I agree that it is often not useful to talk about
consciousness as an objective fact, and we have seen where the
distinction can be blurred by replay-type devices or HLUTs. But
even the reflective aspects of self-consciousness are, IMO,
produced by some process - and I find computation to be a compelling
hypothesis.
A teddy bear or printed words on a page have no mechanism by which
they could be aware of themselves. Now, that's not the same as
saying that there are no universes in which a detective named
Sherlock Holmes actually exists, fitting all the right descriptions.
I'm sure that there are, but they are not accessible by us. In this
world-line in which we find ourselves, Sherlock Holmes is fictional.
Non-fictional entities here-now, to which I would ascribe conscious-
ness, implement computational processes which I could, in theory,
objectively study. Each person has a brain that could be dissected
and studied. If our tools were sophisticated enough, we could
figure out what that creature was experiencing at that moment,
independent of his or her report. We would have no such possibility
with a teddy bear.
Now, that is true of any "computation", or computer program. Given
enough information, we could, in theory, decipher what that program
was subjectively experiencing. That wouldn't necessarily give us
enough so that we would "know what it's like" to be that program.
It may never be possible for us to, for example, know what it's like
to be a bat. But we can certainly understand an awful lot about
that bat, and provide detailed, objective descriptions of that bat's
subjective experiences.
Also, for any computation, we could do the same sorts of dissections.
In the case of an abstract computation, that dissection is
independent of any actual instantiation of the program. That's
where a lot of the confusion has been lately, and is the reason for
the subject of this thread: "Implementation". As far as replay
devices, Olympia's, and HLUTs, here is my take:
A replay device is not an instantiation of a computation in any
sense. It is more analogous to ink on a page, and is therefore
in no sense conscious. On could think of it as a particular run
of a program (Olympia), but as we have seen, in that case, the
program which is actually implemented is a trivial one. Note that
in this, I thoroughly reject Hal's and Jacques' suggestion that
Olympia may be conscious.
HLUTs are clearly conscious, because they implement a complex
computation.
> Should this idea be hard to grasp by people who imagine that whole
> universes exist simply by virtue of seeming to exist to the characters
> implemented in them?
>
> YOU exist simply by virtue of seeming to exist in your own self-model
> (which exists only insofar as you exist).
--
Chris Maloney
http://www.chrismaloney.com
"Donuts are so sweet and tasty."
-- Homer Simpson
Received on Wed Jul 28 1999 - 05:17:22 PDT