sells for a few dollars, gives a simple wiring diagram for solving
Rita’s dilemma. Electrically the problem breaks down into three OR gates
and one AND gate. All Mrs. Farnsworth has to do is set in the conditions
and watch the indicator light. If it glows, she gets the car!
These are of course simple tasks and it might pay to hire a man to
operate the vents, and ride to work on the bus when the car pool got
complicated. But even with relatively few variables, decision-making can
quickly become a task requiring a digital computer operating with
Boolean logic principles.
[Illustration:
_Science Materials Center_
Problem in logic reduced to electrical circuits.
]
The Smith-Jones-Robinson type of problem in which we must find who does
what and lives where is tougher than the car pool—tough enough that it
is sometimes used in aptitude tests. Lewis Carroll carried this form of
logical puzzler to complicated extremes involving not just three
variables but a dozen. To show how difficult such a problem is, an IBM
704 required four minutes to solve a Carroll puzzle as to whether any
magistrates indulge in snuff-taking. The computer did it the easy way,
without printing out a complete “truth table” for the problem—the method
a man would have to use to investigate all the combinations of
variables. This job would have taken 13 hours! While the question of the
use of snuff is perhaps important only to tobacconists and
puzzle-makers, our technical world today does encounter similar problems
which are not practical of solution without a high-speed computer. A
recent hypothetical case discussed in an electronics journal illustrates
this well.
A missile system engineer has the problem of modifying a Nike-Ajax
launching site so that it can be used by the new Nike-Hercules missile.
He must put in switching equipment so that a remote control center can
choose either an Ajax system, or one of six Hercules systems. To
complicate things, the newer Hercules can be equipped with any of three
different warheads and fly either of two different missions. When
someone at the control center pushes a button, the computer must know
immediately which if any of the missiles are in acceptable condition to
be fired.
This doesn’t sound like too big a problem. However, since there are
twelve on-off signals to be considered, and since each has two possible
states, there are 4,096 possible missile combinations. Not all these are
probable, of course, but there is still sufficient variation to make it
humanly impossible to check all of them and close a firing switch in the
split second the control center can allow.
The answer lies in putting Boolean algebra on the job, with a system of
gates and inverters capable of juggling the multiplicity of
combinations. Then when the word comes requesting a missile launch, the
computer handles the job in microseconds without straining itself
unduly.
Just as Shannon pointed out twenty-five years ago, switching philosophy
can be explained best by Boolean logic, and the method can be used not
only to implement a particular circuit, but also to actually design the
circuit in the first place. A simple example of this can be shown with
the easy-to-understand AND and OR gates. A technician experimenting with
an AND gate finds that if he simply reverses the direction of current,
he changes the gate into an OR gate. This might come as a surprise to
him if he is unfamiliar with Boolean logic, but a logician with no
understanding of electrical circuits could predict the result simply by
studying the truth tables for AND and OR.
Reversing the polarity is equivalent to changing a 1 to a 0 and vice
versa. If we do this in the AND gate table, we should not be surprised
to find that the result looks exactly like the OR table! It acts like it
too, as the technician found out.
Boolean logic techniques can be applied to existing circuits to improve
and/or simplify them. Problems as simple as wiring a light so that it
can be turned on and off from two or more locations, and those as
complex as automating a factory, yield readily to the simple rules
George Boole laid down more than a hundred years ago.
Watching a high-speed electronic digital computer solve mathematical
problems, or operate an industrial control system with speed and
accuracy impossible for human monitors, it is difficult to believe that
the whole thing hinges on something as simple as switches that must be
either open or closed. If Leibnitz were alive, he could well take this
as proof of his contention that there was cosmological significance in
the concept of 1 and 0. Maybe there is, after all!
[Illustration:
_Industrial Electronic Engineering & Maintenance_
“Luckily I brought along a ‘loaner’ for you to use while I repair your
computer.”
]
------------------------------------------------------------------------
“_Whatever that be which thinks, understands, wills, and
acts, it is something celestial and divine._”
—Cicero
6: The Electronic Brain
The idea of a man-made “brain” is far from being new. Back in 1851, Dr.
Alfred Smee of England proposed a machine made up of logic circuits and
memory devices which would be able to answer any questions it was asked.
Doctor Smee was a surgeon, keenly interested in the processes of the
mind. Another Britisher, H. G. Wells, wrote a book called _Giant Brain_
in 1938 which proposed much the same thing: a machine with all knowledge
pumped into it, and capable of feeding back answers to all problems.
If it was logical to credit “human” characteristics to the machines man
contrived, the next step then was to endow the machine with the worst of
these attributes. In works including Butler’s _Erewhon_, the diabolical
aspects of an intelligent machine are discussed. The Lionel Britton
play, _Brain_, produced in 1930, shows the machine gradually becoming
the master of the race. A more physical danger from the artificial brain
is the natural result of giving it a body as well. We have already
mentioned Čapek’s _R.U.R._ and the Ambrose Bierce story about a
chess-playing robot without a built-in sense of humor, who strangles the
human being who beats him at a game. With these stories as models, other
writers have turned out huge quantities of work involving mechanical
brains capable of all sorts of mischief. Most of these authors were not
as well-grounded scientifically as the pioneering Dr. Smee who admitted
sadly that his “brain” would indeed be a giant, covering an area about
the size of London!
The idea of the giant brain was given new lease by the early electronic
computers that began appearing in the 1940’s. These vacuum-tube and
mechanical-relay machines with their rows of cabinets and countless
winking lights were seized on gleefully by contemporary writers, and the
“brain” stories multiplied gaudily.
Many of the acts of these fictional machines were monstrous, and most of
the stories were calculated to make scientists ill. Many of these
gentlemen said the only correct part of the name “giant brain” was the
adjective; that actually the machine was an _idiot savant_, a sort of
high-speed moron. This opinion notwithstanding, the name stuck. One
scholar says that while it is regrettable that such a vulgar term has
become so popular, it is hardly worth while campaigning against its use.
An amusing contemporary fiction story describes an angry crowd storming
a laboratory housing a “giant brain,” only to be placated by a calm,
sensibly arguing scientist. The mob dispersed, he goes back inside and
reports his success to the machine. The “brain” is pleased, and issues
him his next order.
“Nonsense!” scoff most computer people. A recent text on operation of
the digital computer says, “Where performance comparable with that of
the human brain is concerned, man need have little fear that he will
ever be replaced by this machine. It cannot think in any way comparable
to a human being.” Note the cautious use of “little,” however.
Another authority admits that the logic machines of the monk Ramón Lull
were very clever in their proof of God’s existence, but points out that
the monk who invented them was far cleverer since no computer has ever
invented a monk who could prove anything at all!
The first wave of ridiculous predictions has run its course and been
followed by loud refutations. Now there is a third period of calmer and
more sensible approach. A growing proportion of scientists take a
middle-of-the-stream attitude, weighing both sides of the case for the
computer, yet some read like science fiction.
Cyberneticist Norbert Wiener, more scientist than fictioneer, professes
to foresee computerized robots taking over from their masters, much as a
Greek slave once did. Mathematician John Williams of the Rand
Corporation thinks that computers can, and possibly will, become more
intelligent than men.
Equally reputable scientists take the opposite view. Neuro-physiologist
Gerhard Werner of Cornell Medical College doubts that computers can ever
match the creativity of man. He seems to share the majority view today,
though many who agree will add, tongue in cheek, that perhaps we’d
_better_ keep one hand on the wall plug just in case.
_Thinking Defined_
The first step in deciding whether or not the computer thinks is to
define thinking. Far from being a simple task, this definition turns out
to be a slippery thing. In fact, if the computer has done no more than
demand this sort of reappraisal of the human brain’s working, it has
justified its existence. Webster lists meanings for “think” under two
headings, for the transitive and intransitive forms of the verb. These
meanings, respectively, start out with “To form in the mind,” and “To
exercise the powers of judgment ... to reflect for the purpose of
reaching a conclusion.”
Even a fairly simple computer would seem to qualify as a thinker by
these yardsticks. The storing of data in a computer memory may be
analogous to forming in the mind, and manipulating numbers to find a
square root certainly calls for some sort of judgment. Learning is a
part of thinking, and computers are proving that they _can_ learn—or at
least be taught. Recall of this learning from the memory to solve
problems is also a part of the thinking process, and again the computer
demonstrates this capability.
One early psychological approach to the man-versus-machine debate was
that of classifying living and nonliving things. In _Outline of
Psychology_, the Englishman William McDougall lists seven attributes of
life. Six of these describe “goal-seeking” qualities; the seventh refers
to the ability to learn. In general, psychologist McDougall felt that
purposive behavior was the key to the living organism. Thus any computer
that is purposive—and any commercial model had better be!—is alive, in
McDougall’s view. A restating of the division between man and machine is
obviously in order.
Dr. W. Ross Ashby, a British scientist now working at the University of
Illinois, defines intelligence as “appropriate selection” and
goal-seeking as the intelligent process _par excellence_, whether the
selecting is done by a human being or by a machine. Ashby does split off
the “non goal-seeking” processes occurring in the human brain as a
distinct class: “natural” processes neither good nor bad in themselves
and resulting from man’s environment and his evolution.
Intelligence, to Ashby, who long ago demonstrated a mechanical
“homeostat” which showed purposive behavior, is the utilization of
information by highly efficient processing to achieve a high intensity
of appropriate selection. Intelligent is as intelligent does, no
distinction being made as to man or machine. _Humanoid_ and _artificial_
would thus be meaningless words for describing a computer. Ashby makes
another important point in that the intelligence of a brain or a machine
cannot exceed what has been put into it, unless we admit the workings of
magic. Ashby’s beliefs are echoed in a way by scientist Oliver Selfridge
of Lincoln Laboratory. Asked if a machine can think, Selfridge says,
“Certainly; although the machine’s intelligence has an elusive,
_unnatural_ quality.”
“Think, Hell, COMPUTE!” reads the sign on the wall of a computer
laboratory. But much of our thinking, perhaps some of the “natural”
processes of our brains, doesn’t seem to fit into computational
patterns. That part of our thinking, the part that includes looking at
pretty girls, for example, will probably remain peculiar to the human
brain.
_The Human Brain_
Mundy Peale, president of Republic Aviation Corporation, addressing a
committee studying the future of manned aircraft, had this to say:
Until someone builds, for $100 or less with unskilled labor, a computer
no larger than a grapefruit, requiring only a tenth of a volt of
electricity, yet capable of digesting and transmitting incoming data in
a fraction of a second and storing 10,000 times as much data as today’s
largest computers, the pilots of today have nothing to worry about.
The human brain is obviously a thing of amazing complexity and fantastic
ability. Packed into the volume Mr. Peale described are some 10
_billion_ neurons, the nerve cells that seem to be the key to the
operation of our minds. Hooked up like some ultra-complicated
switchboard, the network of interconnections stores an estimated
200,000,000,000,000,000,000 bits of information during a lifetime! By
comparison, today’s most advanced computers do seem pathetically
unimpressive.
We have discussed both analog and digital computers in preceding
chapters. It is interesting to find that the human brain is basically a
digital type, though it does have analog overtones as well. Each of the
neurons is actually a switch operated by an electric current on a
go/no-go, all-or-nothing basis. Thus a neuron is not partly on or partly
off. If the electrical impulse exceeds a certain “threshold” value, the
switch operates.
Tied to the neurons are axons, the long “wires” that carry the input and
output. The axons bring messages from the body’s sensors to the neurons,
and the output to other neurons or to the muscles and other control
functions. This grapefruit-size collection of electrochemical components
thus stores our memories and effects the operation we call thinking.
Since brain impulses are electrical in nature, we speak of them in
electrical terms. The impulses have an associated potential of 50
millivolts, that is, fifty thousandths of a volt. The entire brain
dissipates about 10 watts, so that each individual neuron requires only
a billionth of a watt of power. This amount is far less than that of
analogous computer parts.
A neuron may take a ten-thousandth of a second to respond to a stimulus.
This seemingly rapid operation time turns out to be far slower than
present-day computer switches, but the brain makes up for this by being
a “parallel operation” system. This means that many different
connections are being made simultaneously in different branches, rather
than being sequential, or a series of separate actions.
Packaging 10 billion parts in a volume the size of a grapefruit is a
capability the computer designer admires wistfully. Since the brain has
a volume of about 1,000 cubic centimeters, 10 million neurons fit into a
space of one cubic centimeter! A trillion would fit in one cubic foot,
and man-made machines with even a million components per cubic foot are
news today.
Even when we are resting, with our eyes closed, a kind of stand-by
current known as the alpha rhythm is measurable in our brains. This
current, which has a frequency of about 10 cycles per second, changes
when we see or feel something, or when we exercise the power of recall.
It disappears when we sleep soundly, and is analogous to the operating
current in a computer. Also, there is “power” available locally at the
neurons to “amplify” weak signals sufficiently to trigger off following
branches of neurons.
Philosophers have proposed two general concepts of the human brain and
how it functions. The _a priori_ theory presupposes a certain amount of
“wired-in” knowledge: instincts, ideals, and so on. The other theory,
that of the _tabula rasa_ or clean slate new brain, argues that each of
us organizes an essentially random net of nerves into ordered
intelligence. Both theories are being investigated with computers, and
as a result light is beginning to be shed on the workings of our brains.
[Illustration:
_The Upjohn Company, Ezra Stoller Associates Photo_
“A moment at a concert” is diagrammed by brain model, showing eyes,
ears, nerves, and structures analogous to brain. Picture at top
represents perception.
]
There is another division of philosophical thought in the mechanistic
versus _elan vital_ argument. In other words, is the entire mind to be
found in its constituent parts, or is there an intangible extra
something that really breathes life into us? Whatever the correct
concept, the brain does record impressions it can later recall. No one
yet knows just how this is done, but several theories have been
advanced. One of these describes a “chain circuit” set up in a neuron
network by messages from the body’s sensors. This circuit, once started,
continues to circle through the brain and is on tap whenever that
particular experience needs to be recalled. The term “reverberate” is
used in connection with this kind of memory, seeming to be a good
scientific basis for the poetic “echoes of the past.” Reverberation
circuits also provide the memory for some computers.
Among other explanations of memory is that of conditioning the neurons
to operate more “easily,” so that certain paths are readily traversed by
brain impulses. This could be effected by chemical changes locally, and
such a technique too is used in computers.
However the brain accomplishes its job, it is certain that it evolved in
its present form as a result of the environment its cells have had to
function in for billions of years. Its prime purpose has been one of
survival, and for this reason some argue that it is not particularly
well adapted to abstract reasoning. Although the brain can do a wide
variety of things from dreaming to picking out one single voice amid the
hubbub of noise at a social gathering—a phenomenon scientists have given
the descriptive name of “cocktail party effect”—men like Ashby consider
it a very inflexible piece of equipment not well suited to pure logic.
As a test of your brain as a logical device, consider the following
problem from the Litton Industries “Problematical Recreations.”
If Sara shouldn’t, then Wanda would. It is impossible that the
statements: “Sara should” and “Camille couldn’t” can both be true at the
same time. If Wanda could, then Sara should and Camille could. Therefore
Camille could. Is this conclusion valid?
If your head starts to swim, you are not alone. Very few humans solve
such problems easily. Interestingly, those who do, make good computer
programmers.
_The Computer’s Brain_
Just as we have an anthropomorphic God, many people have done their best
to endow the computer with human characteristics. Not only in fiction
but also in real life, the electronic brains have been described as
neurotic and frustrated on occasion, and also as being afraid and even
having morning sickness! A salesman for a line of computers was asked to
explain in understandable terms the difference between two computers
whose specifications confused a customer. “Let’s put it this way,” the
salesman said, “The 740 thinks the 690 is a moron!”
We can begin to investigate the question of computer intelligence by
again looking up a definition. The word “compute” means literally to
think, or reckon, with. Early computers such as counting sticks, the
abacus, and the adding machine are obviously something man thinks with.
Even though we may know the multiplication tables, we find it easier and
_safer_ to use a mechanical device to remember and even to perform
operations for us.
These homely devices do not possess sufficient “intelligence” to raise
any fears in our minds. The abacus, for example, displays only what we
might charitably call the property of memory. It has a certain number of
rows, each row with a fixed number of beads. While it is not fallible,
as is the human who uses it, it is far more limited in scope. All it can
ever do is help us to add or subtract, and if we are clever, to
multiply, divide, do square roots, and so on. If we are looking for
purposive behavior in computing machines, it is only when we get to the
adding machine that a glimmer appears. When a problem is set in and the
proper button pushed, this device is compelled to go through the
gear-whirring or whatever required to return it to a state of
equilibrium with its problem solved.
So far we might facetiously describe the difference in the goal-seeking
characteristics of man and machine by recalling that man seeks lofty
goals like climbing mountains simply because they are there, while the
computer seeks its goal much like the steel ball in the pinball machine,
impelled by gravity and the built-in springs and chutes of the device.
When we come to a more advanced computer, however, we begin to have
difficulty in assessing characteristics. For the JOHNNIAC, built by Rand
and named for John von Neumann, can prove the propositions in the
_Principia Mathematica_ of Whitehead and Russell. It can also “learn” to
play a mediocre game of chess.
If we investigate the workings of a digital computer, we find much to
remind us of the human brain. First is the obvious similarity of on-off,
yes-no operation. This implies a power source, usually electrical, and a
number of two-position switches. The over-all configuration of the
classic computer resembles, in principal if not physical appearance,
that of the human brain and its accessories.
As we have learned, the electronic computer has an input section, a
control, an arithmetic (or logic) section, a memory, and an output.
Looking into the arithmetic and memory sections, we find a number of
comparisons with the brain. The computer uses power, far more than the
brain. A single transistor, which forms only part of a neuron, may use a
tenth of a watt; the brain is ahead on this score by a factor of
millions to one.
Electronic switches have an advantage over the neuron in that they are
much faster acting. So fast have they become that engineers have had to
coin new terms like nanosecond and picosecond, for a billionth and a
trillionth of a second. Thus, the computer’s individual elements are
perhaps 100,000 times faster than those of the brain.
There is no computer in existence with the equivalent of 10 billion
neurons. One ambitious _system_ of computers does use half a million
transistors, plus many other parts, but even these relatively few would
not fit under a size 7-1/2 hat. One advanced technique, using a “2-D”
metal film circuitry immersed in liquid helium for supercooling,
hopefully will yield a packaging density of about 3-1/2 million parts
per cubic foot in comparison with the brain’s trillion-part density.
We have mentioned the computer memory that included the “delay line,”
remindful of the “chain circuit” in the brain. Electrical impulses were
converted to acoustic signals in mercury, traversed the mercury, and
were reconverted to electrical impulses. Early memory storage systems
were “serial” in nature, like those stored on a tape reel. To find one
bit of information required searching the whole reel. Now random-access
methods are being used with memory core storage systems so wired that
any one bit of information can be reached in about the same amount of
time. This magnetic core memory stores information as a magnetic field,
again analogous to a memory theory for the human brain except that the
neuron is thought to undergo a chemical rather than magnetic change.
[Illustration:
_General Electric Co., Computer Dept._
Tiny ferrite cores like these make up the memory of some computers.
Each core stores one “bit” of information.
]