napolitano.pap2 - Minds & Machines Home

advertisement
James Napolitano
Philosophy of AI
Selmer Bringsjord
The popular viewpoint of computationalism assumes that all forms of cognition
and consciousness are logically equivalent to the execution of the correct algorithms.
The very possibility of philosophers’ zombies, however, shows that this is a very
unreasonable assumption.
In the zombie attack against computationalism, a patient’s brain is systematically
replaced, piece-by-piece, by silicon computer chips that function in the same manner.
One logically possible outcome of this thought experiment is that, while the patient’s
external behavior remains the same, his conscious experience and mental life gradually
diminish to nothing. He has become a zombie. Is such a turn of events physically
possible? If it is, then one has a clear counterexample to the functionalist philosophy.
One can construct an argument to suggest that zombies are indeed physically
possible. It is generally accepted that certain objects, like people and animals, are
conscious, while other objects, like hammers and silicon chips, are not. The latter have no
“inner lives” at all. Additionally, if one takes two hammers and moves them about in any
manner or places them in any arrangement, the hammers do not together become
conscious. With these two premises, one can now formally present a zombie argument
against functionalism, where instead of hammers or silicon chips, one uses a non-specific
“object x”:
(1) Object x is not conscious at all.
(2) Two non-conscious objects arranged in any manner are still entirely
without consciousness.
by induction:
(3) Any arrangement of any number of object x’s is not conscious.
(4) If functionalism is true, then if the flow of information through an
arrangement of object x’s is the same as the flow of information through
a human brain, then the arrangement of object x’s is a conscious person.
instantiate (3): (5) An arrangement of object x’s through which the flow of information
is the same as through a human brain is not conscious; it is a zombie.

(6) Functionalism is false.
It follows that consciousness cannot be reduced to a proper arrangement of nonconscious entities. It would be like trying to represent a charged particle as a
combination of neutral particles, or to find a finite solution to the equation x 1  x ; no
solution to the problem exists. It is sometimes said that the qualia people experience are
merely their brain’s interpretations of the underlying biochemistry. The problem here is
that the above argument applies to any computer program; a mindless program that
believes it is conscious and experiencing, say, happiness, does not suddenly become
conscious or happy.
Note that this argument parallels the arbitrary realization arguments typically
given against functionalism. For instance, imagine a billion Chinese workers that hand
buckets of water to each other in a manner that corresponds exactly with the firing of
synapses in a human brain. According to functionalism, these people together would
form one giant consciousness. One odd thing to notice is that the atoms in the brains of
the Chinese workers would then be part of two distinct consciousnesses!
The above arguments suggest that computation is not a sufficient condition for
consciousness. Is computation, then, even a necessary condition for conscious
experience? The answer to that question may be a “no” as well, as can be shown with
another thought experiment. Consider a hypothetical person named Murry. Murry
possesses consciousness, emotions, and qualia, although perhaps not to the same degree
as other people. The problem with Murry is that whenever his brain makes some sort of
higher-order computation, it obtains the wrong answer (one needs to allow the primitive
portions of Murry’s brain to function normally or else it would not be able to regulate his
body functions). For instance, whenever Murry adds 1 and 1, he obtains 3. When Murry
wonders if he exists or not, he decides that he does not. And so on. Is such a being
physically possible? Based on the fact that people typically make many mistakes each
day (especially in comparison to a computer), it would seem so. It follows that
consciousness and computation can actually occur independently of each other.
There is a problem with all of this, however. One can simply let object x be a
subatomic particle. Consciousness, unlike mass and charge, is certainly not one of the
properties that physicists ascribe to subatomic particles; they are believed to be lifeless.
By the zombie argument presented earlier, however, it would follow that anything
composed of matter, such as a human brain, is not conscious, and hence people should be
zombies too! Such a dilemma clearly points out that an error has been made somewhere,
and the computationalist would gladly suggest premise (2) and hence (3) as well. Does
the very consciousness of humans doom the zombie attack?
Fortunately, there is another way out. The zombie argument is deductively valid,
so one of the premises must be at fault. The only premises are propositions (1), (2), and
(4). Proposition (4) is a definition. There were other arguments given earlier to support
(2), and these are necessary to deny computationalism. The only other choice is to deny
premise (1); one must concede that consciousness is a property or a result of matter itself.
It may very well be the case that certain particles or certain interactions among
particles are what causes consciousness to occur. It would then be not just the flow of
information, but the specific arrangement of matter that would determine if something
were conscious or not. Everything would depend on whatever as-yet-unknown laws of
physics there are that deal with consciousness. If any new physics were needed to
explain consciousness, then surely no amount of computer science would be able to
reduce cognition to the actions of a Turing machine. There is also the possibility that, if
the Universe as a whole were ordered in the right manner, then it too would be conscious.
The Universe would possess “the mind of God,” although this would be a very different
“god” that the one of Judeo-Christian tradition; it could even be of sub-human
intelligence. One would be led to pantheism.
There is another possibility, however. The previous paragraphs assumed that
people are arrangements of subatomic particles—and nothing more. If one is willing to
concede that there could be more to reality than that which can be observed in the
laboratory, then this opens up the possibility that there is something in people that
enables them to be more than zombies—something that people have and presumably
computers never will. One need not even go as far as to reject materialism—just certain
strict versions of it.
Download