From: Ben Goertzel (ben@goertzel.org)
Date: Wed Jul 27 2005 - 16:32:23 MDT
Michael,
> Looking at the responses, I think several people haven't actually
> seriously considered a materialist analysis of subjective sensation.
> Yes, they have the general notion that it's all biology and hence
> physics, but they haven't considered what sort of intermediate detail
> there is or what the corresponding causal mechanisms might look like.
> I admitt that if I hadn't thought about this detail the idea that
> subjective experience is something special/nonphysical/noncomputable
> would seem more plausible.
Well, I'm not sure about others, but I know that both Mitchell Porter and I
have thought these things through in a lot of detail, over a period of
decades.
My position is a bit different than Mitchell's, though a bit closer to his
than to yours. Mitch seems a bit more of a conventional quale-iac, whereas
I have crafted a more peculiar view that walks a middle line (though it was
not devised with the intention of walking a middle line -- it was devised to
be correct, and just happened to work out to be a "middle path", just as
Novamente happened to work out to be a sort of middle path between
connectionism and symbolic AI.)
Like you, I *do* think there is an empirical, scientific (I won't say
"mechanistic" due to quantum weirdness) explanation of the behaviors and
utterances of beings like humans (and future AI's) that claim to be
conscious and have qualia. I think one can formulate this explanation in
terms that don't involve qualia, awareness, and the like. Modern science
doesn't have all the details worked out yet, but I'm pretty confident they
will be.
However, this does *not* indicate to me that qualia are meaningless,
nonexistent, yadda yadda.
I think that, as science develops further, we will find that theories
involving qualia form a far *simpler* way of explaining the behaviors and
utterances of beings-that-claim-to-be-conscious-and-have-qualia than
theories that don't involve qualia.
Thus, I think that as future science unfolds, even die-hard materialists
will have to admit that qualia have at least the same sort of "reality" that
electrons do.
Also, I think that future science will allow us to BEGIN with some plausible
assumptions about qualia, and then DERIVE the mathematics of the physical
world therefrom. I.e., to explain apparent objectivity in terms of
subjectivity; just as future neuroscience will allow us to explain
subjectivity in terms of objectivity.
Overall, you can see that my belief in the reality and interestingness of
qualia is not due to a lack of careful thinking about the potential power of
physical explanations to account for the particulars of reported subjective
phenomena. I just don't think these physical explanations, even when
complete, will tell the whole story.
Furthermore, I'd like to note that, while my hypothesized "explanation of
physics in terms of qualia" and "detailed explanation of cognitive systems'
behavior in terms of qualia dynamics" are not yet actualized -- NEITHER IS
THE DETAILED EXPLANATION OF THE BEHAVIORS AND UTTERANCES OF INTELLIGENT
SYSTEMS BASED ON PURE MECHANISM. On all sides we are dealing with
speculations about what future science may provide.
In terms of current science, we have no way to explain subjective experience
in terms of empirical science nor vice versa. To assume that future science
will yield an explanation in one direction (subjectivity in terms of
objectivity) but not the other direction (physics from qualia, a variant of
John Wheeler's "it from bit") is merely to indulge in speculation biased by
one's particular metaphysical assumptions. (I am also indulging in
speculation biased by my own metaphysical assumptions, but I admit it,
whereas positivist/reductionist types seem not to: they seem to accept the
detailed explanation of the subjective in terms of the objective as a done
deal even though it isn't, and dismiss out of hand the possibility of a
detailed reverse explanation).
Regarding Mitch Porter's hypothesis that an AI engineered on purely
reductionist theories might not have qualia, I still doubt it. I think that
if we engineered an AI to reflect on its own structures and dynamics in such
a way that it believed ,based on its own self-study, that the simplest
hypothesis for explaining its own behavior was a hypothesis involving "I
have qualia" -- then this AI would have qualia in exactly the same sense
that human do. Not because qualia are somehow an epiphenomenon of material
phenomena, but because subjectivity and objectivity are different
perspectives on pattern-sets -- meaning they are pattern-sets that often
have significant intersection, and the intelligence-associated patterns in
intelligent self-reflective systems lie in this intersection.
I also think it might be possible to engineer an AI (like Eliezer's
hypothesized "non-sentient highly powerful optimization process") that
lacked the kind of self-reflective cycles correlating with the human use of
qualia for self-modeling, and hence had a very qualitatively different sort
of qualia than we do. Not an AI with no experience at all, but an AI with
such a different sort of experience that we would have a hard time
classifying its experiences as "experience" on an intuitive basis (in a
vaguely similar way to how most people in Western civilizations have a hard
time intuitively classifying a plant's experience as "experience"). AIXItl
would fall into this category if it could be plausibly constructed (which is
not the case).
In conclusion, I do not delude myself that my view of qualia has been
definitively proved; however I have found it a very useful working
hypothesis for AI engineering, study of human psychology, and internal
analysis and modification of my own cognitive and emotional processes.
-- Ben G
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:51 MDT