119
I
have spoken so far about the physical processes that produce the mind.
Now, to come to the heart of the matter, what is the mind? Brain
scientists understandably dance around this question. Wisely, they
rarely commit themselves to a simple declarative definition. Most
believe that the fundamental properties of the elements responsible for
mind─neurons, neurotransmitters, and hormones─are reasonably well known.
What is lacking is a sufficient grasp of the emergent, holistic
properties of the neuron circuits, and of cognition, the way the
circuits process information to create perception and knowledge.
Although dispatches from the research front grow yearly in number and
sophistication, it is hard to jugde how much we know in comparison with
what we need to know in order to create a powerful and enduring theory
of mind production by the brain. The grand synthesis could come quickly,
or it could come with painful slowness over a period of decades.
Still, the experts cannot resist speculation on the essential natureof
mind. While it is very risky to speak of consesnsus, and while I have no
great trust in my own biases as interpreter, I believe I have been able
to piece together enough of their overlapping opinions to forecast a
probable outline of the eventual theory, as follows.
Mind is a stream of conscious and subconscious experience. It is at
root the coded representation of sensory impresssions and the memory and
imagination of sensory impressions. The .....
120
Direction
and magnitude. For example, a particular taste might be partly
classified by the combined activity of nerve cells responding to
different degrees of sweetness, saltiness, and sourness. If the brain
were designed to distinguish ten increments in each of these taste
dimensions, the coding could discriminate 10 × 10 × 10, or 1,000
substances.
Consciousness consists of the parallel processing of vast numbers of
such coding networks. Many are linked by the synchronized firing of the
nerve cells at forty cycles per second, allowing the simultaneous
internal mapping of multiple sensory impressions. Some of the
impressions are real, fed by ongoing stimulation from outside the
nervous system, while others are recalled from the memory banks of the
cortex. All togrther they create scenarios that flow realistically back
and forth through time. The scenarios are a virtual reality. They can
either closely match pieces of the external world or depart indefinitely
far from it. They re-create the past and cast up alternative futures
that serve as choices for fiture thought and bodily action. The
scenarios comprise dense and finely differentiated patterns in the brain
circuits. When fully open to input from the outside, they correspond
well to all the parts of the environment, including activity of the body
parts, monitored by the sense organs.
Who or what within the brain monitors all this activity? No one.
Nothing. The scenarios are not seen by some other part of the brain.
They just are. Consciousness is the virtual world composed by the
scenarios. There is not even a Cartesian theater, to use Daniel
Dennett’s dismissive phrase, nosingle locus of the brain where the
scenarios are played out in coherent form. Instead, there are
interlacing patterns of neural activity within and among particular
sites throughout the forebrain, from cerebral cortex to other
specialized centers of cognition such as the thalamus, amygdala, and
hippocampus. There is no single stream of consciousness in which all
information is brought together by an executive ego. There are instead
multiple streams of activity, some of which contribute momentarily to
conscious thought and then phase out. Consciousness is the massive
coupled aggregates of such participating circuits. The mind is a
self-organizing republic of scenarios that individually germinate, grow,
evolve disappear and occasionally linger to spawn additional thought.
121
The neural circuits do not turn on and off like parts of an electrical
grid. In many sectors of the forebrain at least, they are arranged in
parallel relays stepping form one neuron level to the next, integrating
more and more coded information with each step. The energy of light
striking the retina, to expand the example I gave earlier, is transduced
into patterns of neuron firing. The patterns are relayed through a
sequence of intermediate neuron systems out of the retinal fields
through the lateral geniculate nuclei of the thalamus back to the
primary visual cortex at the rear of the brain. Cells inthe visual
cortex fed by the integrated stimuli sum up the information from
different parts of the retina. They recognize and by their own patternof
firing specify spots or lines. Further systems of these higher-order
cells integrate the information from multiple feeder cells to map the
shape and movement of objects. In ways still not understood, this
pattern is coupled with simultaneous input from other parts of the brain
to create the full scenaarios of consciousness. The biologist S. J.
Singer has drily expressed the matter thus: I link, therefore I am.
Because just to generate consciousness requires an astronomically large
population ofcells, the brain is sharply limited in its capacity to
create and hold complex moving imagery. A key measure of that capacity
lies in the distinction made by psychologists between short-term and
long-term memory. Short-term memory is the ready state of the conscious
mind. It composes all of the current and remembered parts of the virtual
scenarios. It can handle only about seven words or other symbols
simultaneously. The brain takes about one second to scan these symbols
fully, and it forgets most of the information within thirty second.
Long-term memory takes much longer to acquire, but it has an almost
unlimited capacity, and a large fraction of it is retained for life. By
spreading activation, the conscious mind summons information from the
store of long-term memory and holds it for a brief interval in
short-term memory. During this time it processes the information, at a
rate of about one symbol per 25 milliseconds, while scenarios arising
from the information compete for dominance.
Long-term memory recalls specific events by drawing particular persons,
objects, and actions into the conscious mind through a time sequence.
For example, it easily re-creates an Olympic moment: the lighting of the
torch, a running athlete, the cheering of the crowd. It
122
form
of linked concepts simultaneously experienced. Fire is connected to
hot, red, dangerous, cooked, the passion of sex, and the creative act,
and on out through multitudinous hypertext pathways selected by context,
sometimes building new associations in memory for future recall. The
concepts are the nodes or reference points in long-term memory. Many are
labeled by words in ordinary language, but others are not. Recall of
images from long-term banks with little or no linkage is just memory.
Recall with linkages, and aspecially when tinged by the resonance of
emotional circuits, is remembrance.
The capacity for remembrance by the manipulation of symbols is a
transcendent achievement for an organic machine. It has authored all of
culture. But it still fals far short of the demands placed by the body
on the nervous system. Hundreds of organs must be regulated continuously
and precisely; any serious perturbation is followed by illness or
death. A heart forgetful for ten seconds can drop you like a stone. The
proper functioning of the organs is under the control of hard-wired
autopilots in the brain and spinal cord, whose neuron circuits are our
inheritance from hundreds of millions of years of vertebrate evolution
prior to the origin of human consciousness. The autopilot circuits are
shorter and simpler than those of the higher cerebral centers and only
marginally comunicate with them. Only by intense meditative training can
they occasionally be brought under conscious control.
Under automatic control, and specifically through balance of the
antagonistic elements of the autonomic nervous system, pupils of the eye
constrict or dilate, saliva pours out or is contained, the stomach
churns or quietens, the heart pounds or calms, and so on through
alternative states in all the organs. The sympathetic nerves of the
autonomic nervous system pump the body upfor action. They arise from the
middle sections of the spinal cord, and typically regulate target
organs by release of the neurotransmitter norepinephrine. The
parasympathetic nerves relax the body as a whole while intensifying the
processes of digestion. They rise form the brain stem and lower-most
segment of the spinal cord, and the neurotransmitter they release to the
target organs is acetylcholine─also the agent of sleep.
Reflexes are swift automatic responses mediated by short circuits.....
123
or collision. Imagine that you are surprised by a loud noise close by a
car horn blasts, someone shouts, a dog charges in a fury of barking.
You react without thinking. Your eyes close, your head sags, your mouth
opens, your knees buckle slightly. All are reactions that prepare you
for the violent contact that might follow an instant later. The startle
response occurs in a split second, faster than the conscious mind can
follow, faster than can be imitated by conscious effort even with long
practice.
Automatic
responses, true to their primal role, are relatively impervious to the
conscious will. This principle of archaism extends even to the facial
expressions that communicate emotion. A spontaneous and genuine smile,
which originates in the limbic system and is emotion-driven, is
unmistakable to the practiced observer. A contrived smile is constructed
from the conscious processes of the cerebrum and is betrayed by
telltale nuances: a slightly different configuration of facial muscle
contraction and a tendency toward lopsidedness of the upward curving
mouth. A natural smile can be closely imitated by an experienced actor.
It can also be evoked by artificially inducing the appropriate emotion ─
the basic technique of method acting. In ordinary usage it is modified
deliberately in accordance with local culture, to convey irony (the
pursed smile), restrained politeness (the thin smile), threat (the
wolfish smile), and other refined presentations of self.
Much
of the input to the brain does not come from the outside world but from
internal body sensors that monitor the state of respiration, heartbeat,
digestion, and other physiological activities. The flood of “gut
feeling” that results is blended with rational thought, feeding it, and
being fed by it through reflexes of internal organs and neurohormonal
loops.
As
the scenarios of consciousness fly by, driven by stimuli and drawing
upon memories of prior scenarios, they are weighted and modified by
emotion. What is emotion? It is the modification of neural activity that
animates and focuses mental activity. It is created by physiological
activity that selects certain streams of information over others,
shifting the body and mind to higher or lower degrees of activity,
agitating the circuits that create scenarios, and selecting ones that
end in certain.........
124
experience.
Current experience and memory continually perturb the states of mind
and body. By thought and action the states are then moved backward to
the original condition or forward to conditions conceived in new
scenarios. The dynamism of the process provokes labeling by words that
denote the basic categories of emotion ─ anger, disgust, fear, pleasure,
surprise. It breaks the categories into many degrees and joins them to
create myriad subtle compounds. Thus we experience feelings that are
variously weak, strong, mixed, and new.
Without the stimulus and guidance of emotion, rational thought slows
and disintegrates. The rational mind does not float above the
irrational; it cannot free itself to engage in pure reason. There are
pure theorems in mathematics but no pure thoughts that discover them. In
the brain-in-the-vat fantasy of neurobiological theory and science
fiction, the organ in its nutrient bath has been detached from the
impediments of the body and liberated to explore the inner universe of
the mind. But that is not what whould ensue in reality. All the evidence
from the brain sciences points in the opposite direction, to a waiting
coffin-bound hell of the wakened dead, where the remembered and imagined
world decays until chaos mercifully grants oblivion.
Consciousness satisfies emotion by the physical actions it selects in
the midst of turbulent sensation. It is the specialized part of the mind
that creates and sorts scenarios, the means by which the future is
guessed and courses of action chosen. Consciousness is not a remote
command center but part of the system, intimately wired to all the
neural and hormonal circuits regulating physiology. Consciousness acts
and reacts to achieve a dynamic steady state. It perturbs the body in
precise ways with each changing circumstance, as required for well-being
and response to opportunity, and helps return it to the original
condition when challenge and opportunity have been met.
The reciprocity of mind and body can be visualized in the following
scenario, which I have adapted from an account by the neurologist
Antonio R. Damaiso. Imagine that you are strolling along a deserted city
street at night. Your reverie is interrupted by quick footsteps drawing
close behind. Your brain focuses instantly and churns out alternative
scenarios ─ ignore, freeze, turn and confront, or escape. The last
scenario prevails and you act. You run toward a lighted store-
125
conscious
response triggers automatic changes in your physiology. The
catecholamine hormones epinephrine (“adrenaline”) and norepinephrine
pour into the bloodstream from the adrenal medulla and travel to all
parts of the body, increasing the basal metabolic rate, breaking down
glycogen in the liver and skeletal muscles to glucose for a quick energy
feed. The heart races. The bronchioles of the lungs dilate to admit
more air. Digestion slows. The bladder and colon prepare to void their
contents, disencumbering the body to prepare for violent action and
possible injury.
A few seconds more pass. Time slows in the crisis: The event span seems
like minutes. Signals arising from all the changes are relayed back to
the brain by more nerve fibers and the rise of hormone titers in the
bloodstream. As further seconds tick away, the body and brain shift
together in precisely programmed ways. Emotional circuits of the limbic
system kick in ─ the new scenarios flooding the mind are charged with
fright, then anger that sharply focuses the attention of the cerebral
cortex, closing out all other thought not relevant to immediate
survival.
The storefront is reached, the race won. People are inside, the pursuer
is gone. Was the follower really in pursuit? No matter. The republic of
bodily systems, informed by reassuring signals from the conscious
brain, begins its slow stand-down to the original calm state.
Damasio, in depicting the mind holistically in such episodes, has
suggested the existence of two broad categories of emotion. The first,
primary emotion, comprises the responses ordinarily called inborn or
instinctive. Primary emotion requires little conscious activity beyond
the recognition of certain elementary stimuli, the kind that students of
instinctive behavior in animals call releasers ─ they are said to
“release” the preprogrammed behavior. For human beings such stimuli
include sexual enticement, loud noises, the sudden appearance of large
shapes, the writhing movements of snakes or serpentine objects, and the
particular configurations of pain associated with heart attacks or
broken bones. The primary emotions have been passed down with little
change from the vertebrate forebears of the human line. They are
activated by circuits of the limbic system, among which the amygdala
appears to be the master integrating and relay center.
Secondary emotions arise from personalized events of life. To meet
126
the
limbic circuits of primary emotion, but only after the highest
integrative processes of the cerebral cortex have been engaged. We must
know who is friend or enemy, and why they are behaving a certain way. By
this interpretation, the emperor’s rage and poet’s rapture are cultural
elaborations retrofitted to the same machinery that drives the prehuman
primates. Nature, Damasio observes, “with its tinkerish knack for
economy, did not select independent mechanisms for expressing primary
and secondary emotions. It simply allowed secondary emotions to be
expressed by the same channel already prepared to convey primary
emotions.”
Ordinary words used to denote emotion and other processes of mental
activity make only a crude fit to the models used by the brain
scientists in their attempts at rigorous explanation. But the ordinary
and conventional conceptions ─ what some philosophers call folk
psychology ─ are necesarry if we are to make better sense of thousands
of years of literate history, and thereby join the cultures of the past
with those of the future. To that end I offer the following
neuroscience-accented definitions of several of the most important
concepts of mental activity.
What we call meaning is
the linkage among the neural networks created by the spreading
excitation that enlarges imagery and engages emotion. The competitive
selection among scenarios is what we call decision making. The
outcome, in terms of the match of the winning scenario to instinctive or
learned favorable states, sets the kind and intensity of subsequent
emotion. The persistent form and intensity of emotions is called mood. The ability of the brain to generate novel scenarios and settle on the most effective among them is called creativity. The persistent production of scenarios lacking reality and survival value is called insanity.
The explicit material constructions I have put upon mental life will be
disputed by some brain scientists, and reckoned inadequate by others.
That is the unavoidable fate of synthesis. In choosing certain
hypotheses over others, I have tried to serve as an honest broker
searching for the gravitational center of opinion, where by and large
the supporting data are most persuasive and mutually consistent. To
include all models and hypotheses deserving respect in this tumultuous
discipline and then to clarify the distinctions among them.
127
that
in places I chose badly. For that eventually I apologize now to the
slighted scientists, a concession I comfortably make, knowing that the
recognition they deserve and will inevitably receive cannot be blunted
by premature omission on the part of any one observer.
THE
SUBJECT thus qualified, I will next describe the deeper problems that
must be resolved before the physical basis of mind can be said to be
truly solved. The one universally judged to be the most difficult of all
is the nature of subjective experience. The Australian philosopher
David Chalmers recently put the matter in perspective by contrasting the
“easy problems” of general consciousness with the “hard problem” of
subjective experience. In the first group (easy, I suppose, in the sense
that Mont Blanc is more readily climbed in beachwear than Everest) are
the classical problems of mind research: how the brain responds to
sensory stimuli, how it incorporates information into patterns, and how
it converts the patterns into words. Each of these steps of cognition
are the subjects of vigorous contemporary research.
The hard problems is more elusive: how physical processes in the brain
addressed in the easy problems give rise to subjective feeling. What
exactly does it mean when we say we experience a color such as
red or blue? Or experience, in Chalmers’ words, “the ineffable sound of a
distant oboe, the agony of an intense pain, the sparkle of happiness or
the meditative quality of a moment lost in thought. All are part of
what I am calling consciousness. It is these phenomena that compose the
real mystery of the mind.”
An imaginary experiment proposed by the philosopher Frank Jackson in
1983 illustrates the supposed unattainability of subjective thought by
the natural sciences. Consider a neurobiologist two centuries hence who
understands all the physics of color and all the brain’s circuitry
giving rise to color vision. But the scientist (call her Mary) has never
experienced color; she has been cloistered all her life in a
black-and-white room. She does not know what it is like for another
person to see red or blue; she cannot imagine how they feel about color.
According to Jackson and Chalmers, it follows that there are qualities
of conscious experience that cannot be deduced from
128
Although it is the nature of philosophers to imagine impasses and
expatiate upon them at book length with schoolmasterish dedication, the
hard problem is conceptually easy to solve. What material description
might explain subjective experience? The answer must begin by conceding
that Mary cannot know what it feels like to see color. The chromatic
nuances of a westering sun are not hers to enjoy. And for the same
reason she and all her fellow human beings a fortiori cannot know
how a honeybee feels when it senses magnetism or what an electric
field. We can translate the energies of magnetism and electricity into
sight and sound, the sensory modalities we biologically possess. We can
read the active neural circuits of bees and fish by scanning their sense
organs and brains. But we cannot feels as they do ─ ever. Even the most
imaginative and expert observers cannot think as animals, however they
may wish or deceive themselves otherwise.
But incapacity is not the point. The distinction that illuminates
subjective experience lies elsewhere, in the respective roles of science
and art. Science perceives who can feel blue and other sensations and
who cannot feel them, and explains why that difference exists. Art in
contrast transmits feelings among persons of the same capacity. In other
words, science explains feeling, while art transmits it. The majority
of human beings, unlike Mary, see a full color spectrum, and they feel
its productions in reverberating pathways through the forebrain. The
basic patterns are demonstrably similar across all color-sighted human
beings. Variations exist, owing to remembrances that arise from the
personal memories and cultural biases of different people. But in theory
these variations can also be read in the patterns of their brain
activity. The physical explanations derived from the patterns would be
understandable to Mary the confined scientist. She might say, “Yes, that
is the wavelength span classified by others as blue, and there is the
pattern of neural activity by which it is recognized and narned.” The
explanations would be equally clear to bee and fish scientists if their
species could somehow be raised to human levels of intelligence.
Art is the means by which people of similar cognition reach out to
others in order to transmit feeling. But how can we know for sure that
art communicates this way with accuracy, that people really, truly feel
129
weight
of our cumulative responses through the many media of art. We know it
by detailed verbal descriptions of emotion, by critical analyses, and in
fact through data from all the vast, nuanced, and interlocking
armamentaria of the humanities. That vital role in the sharing of
culture is what the humanities are all about. Nevertheless, fundamental
new information will come from science by studying the dynamic patterns
of the sensory and brain systems during episodes when commonly shared
feelings are evoked and experienced through art.
But surely, skeptics will say, that is impossible. Scientific fact and
art can never be translated one into the other. Such a response is
indeed the conventional wisdom. But I believe it is wrong. The crucial
link exists: The common property of science and art is the transmission
of information, and in one sense the respective modes of transmission in
science and art can be made logically equivalent. Imagine the following
experiment: A team of scholars ─ led perhaps by color-challenged Mary ─
has constructed and iconic language from the visual patterns of brain
activity. The result resembles a stream of Chinese ideograms, each one
representing an entity, process, or concept. The new writing ─ call it
“mind script” ─ is translated into other languages. As the fluency of
its readers increases, the mind script can be read directly by brain
imaging.
In the silent recesses of the mind, volunteer subjects recount
episodes, summon adventure in dreams, recite poems, solve equations,
recall melodies, and while they are doing this the fiery play of their
neuronal circuity is made visible by the techniques of neurobiology. The
observer reads the script unfolding not as ink on paper but as electric
patterns in live tissue. At least some of the thinker’s subjective
experience ─ his feeling ─ is transferred. The observer reflects, he
laughs or weeps. And from his own mind patterns he is able to trans-
130
Accurate transmission of the mind script depends as much as
conventional language does on the commonality of the users’ culture.
When the overlap is slight, the script may be limited in use to a
hundred characters; when extensive, the lexicon can expand to thousands.
At its most efficient, the script transmits the tones and flourishes
indigenous to particular cultures and individual minds.
Mind script would resemble Chinese calligraphy, not only a medium
employed for the communication of factual and conceptual information,
but also one of the great art forms of Eastern civilization. The
ideograms contain subtle variations with aesthetic and other subjective
meanings of their own shared by writer and reader. Of this property the
Sinologist Simon Leys has written, “The silk or paper used for
calligraphy has an absorbent quality: the lightest touch of the brush,
the slightest drop of ink, registers at once ─ irretrievably and
indelibly. The brush acts like a seismograph of the mind, answering
every pressure, every turn of the wrist. Like painting, Chinese
calligraphy addresses the eye and is an art of space; like music, it
unfolds in time; like dance, it develops a dynamic sequence of
movements, pulsating in rhythm.”
AN
OLD IMPASSE nonetheless remains: If the mind is bound by the laws of
physics, and if it can conceivably be read like calligraphy, how can
there be free will? I do not mean free will in the trivial sense, the
ability to choose one’s thoughts and behavior free of the will of others
and the rest of the world all around. I mean, instead, freedom from the
constraints imposed by the physiochemical states of one’s own body and
mind. In the naturalistic view, free will in this deeper sense is the
outcome of competition among the scenarios that compose the conscious
mind. The dominant scenarios are those that rouse the emotion circuits
and engage them to greatest effect during reverie. They energize and
focus the mind as a whole and direct the body in particular courses of
action. The self is the entity that seems to make such choices. But what
is the self?
The self is not an ineffable being living apart within the brain.
Rather, it is the key dramatic character of the scenarios. It must
exist, .........
131
conscious
actions. The self and body are therefore inseparably fused: The self,
despite the illusion of its independence created in the scenarios,
cannot exist apart from the body, and the body cannot survive for long
without the self. So close its this union that it is almost impossible
to envision souls in heaven and hell without at least the fantastical
equivalent of corporeal existence. Even Christ, we have been instructed,
and Mary soon afterward, ascended to heaven in bodies ─ supernal in
quality, but bodies nonetheless. If the naturalistic view of mind is
correct, as all the empirical evidence suggests, and if there is also
such a thing as the soul, theology has a new Mystery to solve. The soul
is immaterial, this Mystery goes, it exists apart from the mind, yet it
cannot be separated from the body.
The self, an actor in a perpetually changing drama, lacks full command
of its own actions. It does not make decisions solely by conscious,
purely rational choice. Much of the computation in decision making is
unconscious ─ strings dancing the puppet ego. Circuits and determining
molecular processes exist outside conscious thought. They consolidate
certain memories and delete others, bias connections and analogies, and
reinforce the neurohormonal loops that regulate subsequent emotional
response. Before the curtain is drawn and the play unfolds, the stage
has already been partly set and much of the script written.
The hidden preparation of mental activity gives the illusion of free
will. We make decisions for reasons we often sense only vaguely, and
seldom if ever understand fully. Ignorance of this kind is conceived by
the conscious mind as uncertainty to be resolved; hence freedom of
choice is ensured. An omnicient mind with total commitment to pure
reason and fixed goals would lack free will. Even the gods, who grant
that freedom to men and show displeasure when they choose foolishly,
avoid assuming such nightmarish power.
Free will as a side product of illussion would seem to be free will
enough to drive human progress and offer happiness. Shall we leave it at
that? No, we cannot. The philosophers won’t let us. They will say:
Suppose that with the aid of science we knew all the hidden processes in
detail. Would it then be correct to claim that the mind of a particular
individual is predictable, and therefore truly, fundamentally
determined and lacking in free will? We must concede that much in
132
interval
of a microsecond the active networks composing the thought were known
down to every neuron, molecule, and ion, their exact state in the next
microsecond might be predicted. But to pursue this line of reasoning
into the ordinary realm of conscious thought is futile in pragmatic
terms, for this reason: If the operations of a brain are to be seized
and mastered, they must also be altered. In addition, the principles of
mathematical chaos hold. The body and brain comprise noisy legions of
cells, shifting microscopically in discordant patterns that unaided
consciousness cannot even begin to imagine. The cells are bombarded
every instant by outside stimuli unknowable by human intelligence in
advance. Any one of the events can entrain a cascade of microscopic
episodes leading to new neural patterns. The computer needed to track
the consequences would have to be of stupendous proportions, with
operations conceivably far more complex than those of the thinking brain
itself. Furthermore, scenarios of the mind are all but infinite in
detail, their content evolving in accordance with the unique history and
physiology of the individual. How are we to feed that into a computer?
So there can be no simple determinism of human thought, at least not in
obedience to causation in the way physical laws describe the motion of
bodies and the atomic assembly of molecules. Because the individual mind
cannot be fully known and predicted, the self can go on passionately
believing in its own free will. And that is a fortunate circumstance.
Confidence in free will is biologically adaptive. Without it the mind,
imprisoned by fatalism, would slow and deteriorate. Thus in organismic
time and space, in every operational sense that applies to the knowable
self, the mind does have free will.
FINALLY,
given that conscious experience is a physical and not a supernatural
phenomenon, might it be possible to create an artificial human mind? I
believe the answer to this philosophically troubling question to be yes
in principle, but no in practice, at least not as a prospect for many
decades or even centuries to come.
Descartes, in first conceiving the question over three centuries ago
declared artificial human intelligence to be impossible. Two
133
machine
from a real mind. It could never “modify its phrases to reply to the
sense of whatever was said in its presence, as even the most stupid men
can do,” and it could never “behave in all the occurrences of life as
our reason makes us behave.” The test was recast in operational terms by
the English mathematician Alan Turing in 1950. In the Turing test, as
it is now generally called, a human interpreter is invited to ask any
question of a hidden computer. All he is told is that either another
person or a computer will answer. If, after a respectable period of
time, the questioner is unable to tell whether the interlocutor is human
or machine, he loses the game; and the mind of the machine is accorded
human status. Mortimer Adler, the American philosopher and educator,
proposed essentially the same criterion in order to challenge not just
the feasibility of humanoids but also the entire philosophy of
materialism. We cannot accept a thoroughly material basis for human
existence, he said, until such an artificial being is created. Turing
thought the humanoid could be built within a few years. Adler, a devout
Christian, arrived at the same conclusion as Descartes: No such machine
will ever be possible.
Scientists, when told something is impossible, as a habit set out to do
it. It is not, however, their purpose to search for the ultimate
meaning of existence in their experiments. Their response to cosmic
inquiry is most likely to be: “What you suggest is not a productive
question.” Their occupation is instead exploration of the universe in
concrete steps, one at a time. Their greatest reward is occasionally to
reach the summit of some improbable peak and from there, like Keats’
Cortez at Darien, look in “wild surmise” upon the vastness beyond. In
their ethos it is better to have begun a great journey than to have
finished it, better to make a seminal discovery than to put the final
touches on a theory.
The scientific field of artificial intelligence, Al for short, was
inaugurated in the 1950s hard upon the invention of the first electronic
computers. It is defined by its practitioners as the study of
computation needed for intelligent behavior and the attempt to duplicate
that behavior using computers. A half century of work has yielded some
impressive results. Programs are available that recognize objects and
faces from a few select features and at different angles, drawing on
rules of geometric symmetry in the manner of human cognition.
134
classify novel objects on the basis of cumulative experience ─ much in the manner of the human mind.
Some programs can scan and choose options for particular courses of
action according to preselected goals. In 1996 Deep Blue, an advanced
chess-playing computer, earned grand master status by narrowly losing a
six-game match to Gary Kasparow, the reigning human world champion. Deep
Blue works by brute force, using thirty-two microprocessors to examine
two hundred million chess positions each second. It finally lost because
it lacked Kasparov’s ability to assess an opponent’s weakness and plan
long-term strategy based in part on deception. In 1997 a reprogrammed
and improved Deep Blue narrowly defeated Kasparov: the first game to
Kasparov, the second to Deep Blue, then three ties and the final game to
Deep Blue.
The search is on for quantum advances in the simulation of all domains
of human thought. In evolutionary computation, AI programmers have
incorporated an organismlike procedure in the evolution of design. They
provide the computers with a range of options in solving problems, then
let them select and modify the available procedures to be followed. By
this means the machines have come to resemble bacteria and other simple
one-celled organisms. A truly Darwinian twist can be added by placing
elements in the machines that mutate at random to change the available
procedures. The programs then compete to solve problems, such as gaining
access to food and space. Which mutated programs will be born and which
among the the neonates will succeed are not always predictable, so the
“species” of machines as a whole can evolve in ways not anticipated by
the human designer. It is within the reach of computer scientists to
create mutable robots that travel about the laboratory, learn and
classify real resources, and thwart other robots in attaining their
goals. At this level their programs would be close to the instinctive
repertories not of bacteria but of simple multicellular animals such as
flatworms and snails. In fifty years the computer scientists ─ if
successful ─ will have traversed the equivalent of hundreds of millions
of years of organic evolution.
But for all that advance, no AI enthusiast claims to have a road map
from flatworm instinct to the human mind. How might such an immense gan
be closed? There are two schools of thought. One, rep-
135
Technology,
takes a bottom-up approach. In this version, the designers would follow
the Darwinian robot model to higher and higher levels, gaining new
insights and elaborating technology along the way. It is possible that
in time, humanoid capability might emerge. The other approach is
top-down. Favored by Marvin Minsky, a founding father of AI and
colleague of Brooks at MIT, it concentrates directly on the
highest-order phenomena of lerarning and intelligence as they might be
conceived and built into a machine without intervening evolitionary
steps.
In the teeth of all pessimistic assessments of human limitation likely
to be raised, human genius is unpredictable and capable of stunning
advances. In the near future a capacity for at least a crude simulation
of the human mind might be attained, comprising a level of brain
sciences sophisticated enough to understand the basic operations of the
mind fully, with computer technology advanced enough to imitate it. We
might make up one morning to find such a triumph announced in the New York Times,
perhaps along with a generic cure for cancer or the discovery of living
organisms on Mars. But I seriously doubt that any such event will ever
occur, and I believe a great majority of AI experts are inclined to
agree. There are two reasons, which can be called respectively the
functional obstacle and the evolutionary osbtacle.
The functional obstacle is the overwhelming complexity of inputs of
information to and through the human mind. Rational thought emerges from
continuous exchanges between body and brain through nerve discharges
and blood-borne flow of hormones, influenced in turn by emotional
controls that regulate mental set, attention, and the selection of
goals. In order to duplicate the mind in a machine, it will not be
nearly enough to perfect the brain sciences and AI technology, because
the simulation pioneers must also invent and install an entirely new
form of computation ─ artificial emotion, or AE.
The second, or evolutionary, obstacle to the creation of a humanoid
mind is the unique genetic history of the human species. Generic human
nature ─ the psychic unity of mankind ─ is the product of millions of
years of evolution in environments now mostly forgotten. Without
detailed attention to the hereditary blueprint of human nature, the
simulated mind might be awesome in power but
136
And even if the blueprint were known, and even if it could be followed,
it would serve only as a beginning. To be human, the artificial mind
must imitate that of an individual person, with its memory banks filled
by a lifetime’s experience ─ visual, auditory, chemoreceptive, tactile,
and kinesthetic, all freighted with nuances of emotion. And social:
There must be intellectual and emotional exposure to countless human
contacts. And with these memories, there must be meaning, the expansive
connections made to each and every word and bit of sensory information
given the programs. Without all these tasks completed, the artificial
mind is fated to fail Turing’s test. Any human jury could tear away the
pretense of the machine in minutes. Either that, or certifiably commit
it to a psychiatric institution.
