Postscript to Artificial Intelligence: Reflections on John
Searle’s Rediscovery of the Mind
Ken Wais/12/7/02
As I said in the
original article, my stance on the issue of computer intelligence is
undecided. Searle however is not. He affirms that human intelligence or more
specifically, human consciousness is a subjective phenomenon. He further asserts this phenomenon is not
accessible by the reductionist methodology.
Searle makes a very persuasive argument, but I cannot go as far as he
does. Searle wants us to believe that
consciousness by its subjective nature is unassailable by this method. His objection, while he claims it’s not the
Kantian argument concerning the a priori nature of consciousness, it is
just that. He goes to great lengths to
show that the very subjectivity of consciousness makes it impossible to reduce
ones own consciousness to something else in terms of its constructs. In plainer terms: consciousness is
irreducible. Why? Because by doing this,
we lose the subjective sense of self. It vanishes upon our deconstructing
it into its constituencies. At present,
he admits and I concur, we can’t even do this reduction with the human mind.
But supposing we could, what then?
How does Searle come
to this conclusion? First, he
differentiates consciousness into two philosophic schools: the epistemic and
the ontological arguments. By epistemic,
he means our knowledge of our conscious states.
By ontological he means our being in such conscious states. In the
former, we have our ability to look at consciousness as a phenomenon, which can
be studied through causative inputs and effectual outputs. This study allows us
to understand its process. In the
latter, we have ourselves as beings, which necessarily exist in a conscious
state at any given moment and are not analyzable in an objective third-person
fashion. We can’t, Searle asserts
because we, ourselves are the beings to which we are trying to apply this
analysis. This attempt is doomed because
to be obstructs to know. If this objection sounds similar
to Goedel's Incompleteness paradox I’ve described above, you're right, it
is. It is also very similar to the
argument Kant made in his Critique of Pure Reason. Here is where we get to the heart of
Searle’s debate with AI scientists. His debate is larger than AI proper and the
scientists that are working in the specialized field of cognitive science, but
I'm focusing on them in this article. He
tells us, they believe a form of materialism known as Reductionism. They believe there must be a way to explain
the complex machinery of our minds in terms that are understandable via some
form of analysis. In the case of AI, its
functionalism, sometimes called computationalism. He does not agree. He believes that the subjective nature of consciousness
makes it unassailable by this method or any other for that matter. He explicitly states what he knows is a
cardinal sin to scientific theorists: All things aren’t knowable! He goes
further and asserts this is common sense and should not alarm or disquiet
anyone within the cognitive science discipline.
I have great respect
for John Searle and agree with his conclusion that the Turing Test is flawed
and not sufficient grounds to establish consciousness for an AI. However,
I must point out, that he is on uncertain logical grounds with his objection to
computational AI. But, I can't disagree more, that an AI is theoretically
unattainable because of his stubborn believe that the human mind cannot be
understood via computational methods. My
intuition is he is not as certain or convinced of his own views as he suggests. He just might feel the bold assertion that
consciousness is irreducible to computational functionalism might be
disproved. Note a passage from the book
as evidence:
“Furthermore, when I speak of the irreducibility of consciousness, I am
speaking of its irreducibility according to standard patterns of reduction. No one can rule out a priori the possibility
of a major intellectual revolution that would give us a new—and at present
unimaginable—conception of reduction, according to which consciousness would be
reducible.” A-ha, so there is
still hope for materialists and the functional perspective yet.
The problem with
Searle’s argument (though it may be cogent and compelling) is it doesn’t play
fair. It rules out every possible way to
inspect and verify it. If consciousness is
irreducible because the subjectivity of consciousness is lost when we try to
make this reduction, then consciousness is unknowable by reductionist
methodology. But, notice the
irreducibility principle is not a conclusion of a deductive argument, but an
assertion of a common sense notion, according to Searle. It is put forth as the result of a thought
experiment in which he makes the assumption that subjective consciousness is
indivisible, because doing this would contradict the very nature of being
subjective. Again, this is assumed not
proved. It has the semblance of Goedel’s
Incompleteness Theorem paradox, but lacks the deductive structure. Goedel came to his startling conclusion based
on a set of valid axioms. Searle is
basing his argument on assuming the conclusion.
Let’s not dwell on
this issue. What about the opposite side
of the coin? How can human consciousness
come into being? Here, Searle again falsely
attempts to debunk a materialist argument.
Actually, it was Hegel that first laid the groundwork for what modern
cognitive and computer scientists call the emergence
principle of consciousness. Hegel’s
dialectical process, states explicitly that quantity and quality affect each
other reciprocally. Changes in quantity
lead to commensurate changes in quality, and vice versa. It is this principle that modern scientists
invoke, when they claim that the human brain’s myriad of sub-cellular entities,
cells, tissues, organs and subsystems working in concert, produce an organized
self-aware system we call consciousness, a qualitatively different
phenomenon than any of its constituencies. Searle attacks this notion
too.
To be continued
Return to Portal Philosophies, Science, Mathematics, and Music