Here's my perspective on Walter Fritz's question (apologies to Don M. for
not having yet read Rosen, which I certainly intend to do -- perhaps these
ideas will intersect in some way):
Does it not boil down to the question "what is experience?" Trying to think
un-anthropomorphically even though saddled with anthropomorphic terms, I
ask: does the calculator "experience" its own activity at the level of
calculating numbers? Does your AI machine "experience" it's reply to the
question AT THAT LEVEL? Recall the story of Cleaver Hans (the horse) who
supposedly could add. Hans was able to reply correctly to mathematical
questions. However, it was eventually discovered (via double blind methods,
apparently to the surprise of the owner too) that the horse was being
extremely attentive to subtle cues from the questioner which indicated when
the correct number had been hoofed out. The horse certainly has (I
believe) what I would call "experience of life." But his/her experience was
not integrated to the level of mathematics, only to the level of perceiving
and pleasing the owner's desires. Similarly, Washoe the Chimpanzee (first
to learn American Sign Language) demonstrated a capacity for language which
many traditionalists claimed was a "cleaver Hans" phenomenon. Every
indication was and is that Chimps (Washoe and more accomplished successors)
are actually "experiencing" the language, not just mimicking it. There is
no more informative way of describing this than to refer to the animal's
experience, even though we can't observe it directly. But that is not a
problem epistemologically, because there are other situations in science
where we must assume a hidden reality beyond a veil which prevents
observation and thus description that is tested objectively. The ability to
test descriptions subjectively is less respected by scientists, but in many
fields there is no alternative, and rigorous practices and methods can
reduce the problems. One simply has to be trained in these skills, just as
one needs to be trained for objective analysis. The untrained in either
camp fall victim to sloppy thinking.
Now I go much farther than this, with considerable justification, and claim
that experience is a dual aspect of nature and is present at every level. I
know that "experience of life" can't be defined objectively, only
subjectively (i.e., in the realm of experience), and some people then
reject it from scientific consideration as a dual but unnecessary (i.e.,
irrelevant, even if true) view. The point is that the problem doesn't go
away. The "dual" views have been discovered to not be entirely separable -
they are not completely independent and there are natural phenomena that
seem to involve an intersection of both. I believe, from the evidence, that
"experience" is what is involved in the quantum phenomena, yet we are
justified in delaying that interpretation until the mechanical explanations
are shown to be impossible. By analogy at the level of observed behavior,
many have concluded that the wierd quantum nature is similar to that
exhibited by consciousness, such as in phychological phenomena. What we
might suspect to be willful/purposeful behavior is distinctly non-classical
and involves things like J.A. Wheeler (wierd physicist) described, where "a
thing is not a thing until it is a registered [observed or experienced]
thing." (my interpretation in brackets.) But this does not mean that "will"
or "experience" is necessarily organized at the level of some particular
human experience we might compare it to, nor that it is significant in any
but very specific structures. For example (as in my reply to another
posting on this site) it would be epistemologically unsound to take this
interpretation for classical objects, such as billiar balls or gas
molecules behaving in a regular manner that is well described by Newtonian
concepts of "force." The reality of these "forces" however, are very little
different from the reality of "will/experience" - they are both
metaphysical in origin. As a simple example, gravitational "force" has been
shown to be equivalent to geometrical interpretation (curved relativistic
space). Yet a more appropriate example might be wave-particle duality,
because although it embodies complimentary views (like force vs. geometry)
it is unlike classical or mathematical compliments. With wave-particle
duality, other quantum phenomena, and psychological experience, neither
view (objective vs. experiential) is dispensible and they do affect each
other in important and measurable ways. We need to do the research to
determine where experiential phenomena enters into otherewise mechanical
processes, and how strongly. Can we put some rules on it, as was done in
Quantum? If not, then I am convinced that it will be impossible to develop
a cybernetic paradigm that goes beyond mimikry.
The problem, as I see it, of constructing a "thinking machine" is that this
is a contradiction in terms. "Mechanism" is the opposite of "experience" in
every way, but they intersect in specific phenomena. They do not
circumscribe the same thing at all, but may both be simultaneously involved
in a certain kind of event (e.g., a quantum state produced by an
"experiential" event and resulting in a mechanical state, or a human
decision resulting in certain defined states of consciousness. A computer
is "aware" (being somewhat poetic here) of the activity of electrons, but
its "experience" is not integrated above that level. The patterns that
those electrons are allowed to travel in response to instructions, which
result in meaningful calculations, were created by humans and their meaning
is only experienced by humans. Hence the computer does not think at this
level, but we can say that the human who designed the electrical pathways
did, because he/she experienced the exercise at that level.
I am gennerally aware of the arguments for/aginst dualism, and I do not
argue as a dualist, though it may seem so here. There is undoubtedly a
greater reality to discover; however, my argument is that we have not
progressed to that discovery yet, and are still involved in attempting to
reject one or the other half of a necessary whole. This was acceptable in
both science and religion for many years, and may still be acceptible in
specific disciplines. But we have exhausted the objective paradigm in some
fields, and now science has to grow. Similarly, I would say religion has
exhausted the strictly theological paradigm, and also must grow. I am
primarily concerned with global phenomona and man's influence on the
"environment" - and we now must consider that the environment includes
mankind (both body and mind) to an extent where it is no longer possible to
restrict all scientific inquiry to studying the two as separate phenomena.
But I shall not launch into that.
Now one could tie this philosophy to any number of references because so
many conflicting ideas have been generated through the ages that many are
bound to reveal a fit (not to rule out intent here, just to say there isn't
much point in providing the citations because by now we all should be
familiar with the mind-body philosophy problem -- and also to suggest that
we might have something to contribute to the discussion if we don't
incessantly tie everything to a past statement and declare it over while
still unresolved). I also suspect that this view does not match current
cybernetics thinking very well, although I see both sides of the debate on
this web site. There are some who will say "yes, just so" and others who
will claim that "experience" is unproven and unobservable and does/should
not enter science. There are yet others who will insist, with the firmest
of grounds, that science cannot be inclusive. Still others who remind us
that if anything in the universe is to be found as a holism, then
consciousness (and thinking) must be, and thus it is unlikely that it could
be contained entirely within an objective model.
Yet I do believe, as my own personal conclusion after much questioning,
that "experience" must be counted as scientifically real and its effects
(i.e., objective results) must be accounted for within mechanistic views.
In other words, the mechanistic view is not an independent alternative.
Just as others developed necessarily limited and inadequate (but
nonetheless useful) models for the part of quantum causation we can observe
- which can only be described from the objective view behaviorally, as
probabilistic and ontologically uncertain - we should be able to develop
useful models of similar phenomena where objective description must account
for uncertainties resulting from the subject's own direct causation
(experience and resulting decisions). Understanding that causation requires
direct experience (which we can gain only from analogy with our own).
Duplicating it in a computer (rather than mimiking it) would require
building a mechaniism that is capable of such indeterministic behavior and
incorporating that structure into the mechanism. This may be possible by
learning how to incorporate bose-einstein condensates (which do exhibit
this behavior) into the information processing architecture of an otherwise
mechaniical computer. This means that it will not be doable in software
alone, with our current computer structures. Quantum computing is emerging
and may hold some promise.
To make it more clear what I am saying, we can describe a gorilla's
behavior in mechanical terms. We can do so better in statistical terms and
even better (more predictive) if we develop probabilities. This will
satisfy some people who don't want to look any farther and disgust others
who do. The probabilities will be useful, but they will not tell us what is
really going on in the gorilla, and it is quite apparent that what is
really going on in there actually does affect the objective results
(probabilities) and is something as wierd as human experience. It behaves
according to the same "doesn't exist until I say so" nature as the quantum
phenomena. The statistics then cannot be more accurate than the
probabilities suggest, and noticable and important uncertainties remain.
Yet humans, who have endeavored to share as much as possible
psychologically with the gorilla (i.e., attempt to participate in and share
thoughts) have learened far more about gorillas than can be captured in
mechanical descriptions. The problem of anthropomorphasizing was well
considered in this work, and prevented it for an inappropriately long time.
The problem did/does not go away, but again, with careful methods it can be
reduced.
>
>Walter Fritz wrote:
>
>> I have a question on simulation, modelling etc., for those who
>> understand philosophy better that I do:
>>
>> When we take a pocket calculator and type in 5732 dived by 31 and it
>> shows us 184 and some decimals, what did the calculator do. Did it
>> simulate calculating, did it emulate calculating, did it model
>> calculating or did it calculate?
>>
>> Similarly when we take the artificial intelligent system called "General
>> Learner" and type in: "show me a small triangle" and the computer shows
>> a small triangle on the screen (because it had previously been taught
>> the concept of triangle and the concept of small).
>> What did it do, did it simulate thinking, did it emulate thinking, did
>> it model thinking or did it think?
>>
>> I would highly appreciate any help you can give me on this, and better
>> even if you can state your reasoning behind the answer (that is, why you
>> think so).
>>
>> Walter Fritz
>
>
-----------------------------------------------
John J. Kineman, Physical Scientist/Ecologist
National Geophysical Data Center
325 Broadway E/GC1 (3100 Marine St. Rm: A-152)
Boulder, Colorado 80303 USA
(303) 497-6900 (phone)
(303) 497-6513 (fax)
jjk@ngdc.noaa.gov (email)
(303) 497-6513 (fax)