SECTION VII Chapter 7. Limitations of the Human Mind "Practically perfect people never permit sentiment to muddle their thinking." -- Mary Poppins ""A great many people think they are thinking when they are merely rearranging their prejudices." -- William James Artificial Intelligence-- or Can Computers be Built with a Human-Like Mind? In this chapter we shall look at the modern debate over whether a computer can be built, which can simulate the learning and creativity of the human mind. Basically three are three possible viewpoints regarding computer, or "ARTIFICIAL" intelligence: (1) There is a divine, "mystical" element within the human brains--which no man-made computer will EVER be able to simulate--with ANY technology! (This is the view of theologians and many religious scientists!) (2) It is only a matter of time, before computers become powerful enough to tackle the simulation of the complex human mind. (3) If computers are used to simulate the human mind--then their architecture must be designed to parallel the natural construction of the human mind--ie multiple central processing units (cpus) etc. These views on artificial intelligence are analyzed in more detail below. I. No Computer Could ever Simulate the Mysterious Actions of the Human Brain! The complexity and mysterious inter-workings of the brain is proof to many theologians and religious scientists that a divine element (ie God) was involved in constructing both man and his powerful brain. Various religions believe in the existence of a SOUL that resides in a physical body, that is the source of the "unknown" mysterious workings of the brain. The famous English physicist, Roger Penrose, was a proponent of this view. In his 1989 book, THE EMPEROR'S NEW MIND, Penrose argued that there are "mystical" forces operating within the mind (by which he is implying a divine-like power). In explaining how there is still much about the workings of the brain that scientists don't fully understand, Penrose noted how scientists have no clue regarding how the human brain intuitively assesses the truth of an argument, or understands the humor in a joke. A computer program currently cannot feel the beauty in music or a piece of art, nor philosophize regarding the meaning of life (such as Descartes famous dictum, "Cognito, ergo sum"). If consciousness could be explained by a complex program, why hadn't neuroscientists gained at least SOME insight by now into the inter-workings of the mind by using these models? Based on this, Penrose concluded that the "quality of understanding and feeling possessed by human beings is not something that can be simulated computationally". and, "It may well be there is something else going on in the brain that we don't have an inkling of at the moment." Consequently, Penrose believed that it was virtually impossible for a human to ever construct a computer/robot that could simulate a HUMAN mind-- as there was more at work here than mere physics and chemistry. II. The Computer as a Good Model for Understanding the Human Mind Some scientists view the computer as a good model for understanding the human mind. This view became especially popular during the 1940's when computers began to be commercially manufactured. Artificial intelligence experts Warren McCullough and Walter Pitts argued during this time, that the digital "on" and "off" states of the computer had their parallels in the human nerve cell which is either in a state of firing, or NOT firing off a signal. That is, just as the computer breaks down all input into symbolic codes, and processes this within its CPU circuitry-- proponents of the "brain-as-a computer" model believe our sense organs convert sensory information into a "code", which is then electrochemically processed by our brain into neuro-physiological signals. Other computer wizards such as Marvin Minsky of MIT, have argued that the brain is simply: "hundreds of different machines...connected to each other by bundles of nerve fibres, but not everything is connected to everything else." Of course the fact that we do not appear to be even close to constructing a computer that has the power and creativity of the human mind, has posed one setback to proponents of this theory. Some have used this as proof that there is some divine element contained within the human brain which is impossible to replicate. Another theory (discussed next), argues that the human mind is PHYSCIALLY constructed much differently than modern computers. Therefore, only until computers are modified to parallel the construction of the human mind, will it be possible to build a computer with artificial intelligence. III. The Computer as a Poor Model for Understanding the Human Mind Proponents of the theory that the computer is NOT a good model of the brain, point out the following: (1) Computers are typically built with ONE central processing unit, whereas the brain appears to be made up of a multiple number of very specialized processing units--which have powerfully integrated actions between them. (2) Unlike computers, humans possess powerful SENSING organs transmitting information from sight, sound, touch, taste, and smell. Humans brains use biochemical/electrical processing, while computers rely entirely on electrical processing. (3) Unlike the computer, the human nerve cell is alive--and dynamically expands (and shrinks) with new experiences. That is, neurons form, dissolve, and reform new chemical and electrical connections in our continued interaction with our own body processes and with the environment. In a computer, this effect can be simulated--however only with extremely complex (software) programming. (4) Even more important, is the fact that human nerve cells are NOT serially linked with one touching the other.--Instead they are grouped into clusters of nerve cells. Thus the simplistic "on-off" mechanism of the computer is not relevant here--because instead there is a "conversation" of signals going on, simultaneously within hundreds, even thousands of nerve cells in other clusterlike networks. (Restak, THE MIND, p 257) This organization makes it possible to cross-reference ideas and concepts in a large variety of ways. Indeed, this explains why a common mnemonic (memory) trick--such as remembering a person's name--calls for one to try and associate the memorized name with as many ideas as possible. (For example, I may try to remember the name Bryce by noting it rhymes with "ice" and thus visualize ice cream with this person's name). This theory also explains why we can struggle to remember some item-- again say a person's name--when suddenly in a flash, the name and other associate memories with this person is now "remembered". And last, this would also explain the human creative process--often called "brainstorming" where new connections are made between existing pieces of knowledge. It has been noted before that when brainstorming, a person does not INVENT new facts--but instead new ways of combining existing facts known to them. This is very different with how a computer SERIALLY recalls information-- Computers are amazingly fast and efficient at retrieving information and performing calculations. However, computers are very poor at establishing multiple relationships between different sets of information--unless this is PRE-programmed into the computer. For example, while a six month child can recognize the face of his mother, a computer trying to simulate this simple example of pattern recognition--must grind through massive algorithms to find an "exact" match. There is, however, a NEGATIVE aspect to our human pattern recognition ability! And that is, humans are SO good at finding patterns, that they will find "patterns" EVEN WHEN CLEARLY NO PATTERNS EXIST AT ALL. Psychologists have performed studies, for example, that have asked test subjects to stare at RANDOM static on a tv screen or monitor.--In many cases these test subjects INSISTED that they saw REAL patterns. In one study performed at the University of Arlington at Texas, a computer game was developed whereby students could either REWARD or PUNISH the actions of a fictitious character. Although in reality the computer model was programmed to generate RANDOM results, most students INSISTED that their rewards and punishments had made a "difference" in the results of the character's actions. In some cases, even after confronted that there was no REAL cause and effect--that all was randomly generated--students refused to accept this as an answer. What these students were doing was to IGNORE those situations when the computer results did NOT confirm that their reward/ punishments made a difference on the outcome of the game. That is, the students only counted the "successes" where they perceived a cause and effect, and ignored the "misses". (See Section VII, Chapter 1.) The implications of these psychological tests are that the human mind appears to be constructed so as to search for and find patterns, EVEN WHEN CLEARLY NONE EXISTS! Other DIFFERENCES Between Human and Computer Processing --Recording Meaning, Instead of Sequences of Letters and Symbols Another example that demonstrates the difference in architecture between computers and humans, can be seen by interpreting the following statements: Descartes: "I think therefore I am" Einstein: "E=MC2" To the computer, the above sentences are only an encoded series of bits of information. Only humans attribute any REAL meaning to these statements. Nor were humans BORN where this information had any meaning--A child for example, will not understand either of these statements, without any previous exposure to these ideas. Where a person has NO experiencing relating the above sentences to other ideas and concepts, these are nothing more to him than meaningless gibberish--ie they register little to no meaning. --Unlike Computers, Most Humans Are Naturally Terrible at Logic Various psychological tests have demonstrated how most humans, unless properly trained, are naturally TERRIBLE at logic. For example, consider the following problem put forth by John E. Taplin and Herman Staudenmayer of the University of Colorado (footnote as described by Morton Hunt, THE UNIVERSE WITHIN, (A TOUCHSTONE BOOK, published by Simon and Shuster, New York, 1982), p 126) Which statement(s) are true: I."If there is a Z, then there is an H. There is no Z. Therefore there is no H. II.If there is a Z, there is an H. There is an H. Therefore there is a Z." Close to two-thirds of the testing subjects (who had no formal training in logic) thought both cases were true. The answer is that both of these are false. (Footnote: For example, in I, if we said "If there is a cat there is no animal" and "There is no cat, therefore there is no animal." One could now relate to this pictorially to see, that there can be other animals that are not cats. Likewise in II, "If there is an animal, there is a cat too", can be demonstrated to be false, say if the animal is instead a dog.) (Footnote, Morton Hunt also refers to a study by Wason where there are four cards that have been placed down that show an "A", a "D", a "4" and a "7". The subject has been told that "If a card has a vowel on one side, then it has an even number on the other side." The problem is to state which cards must be turned over to determine if the rule is true or false. The answer: * the "A" (which is a vowel--card must be turned over to test rule) * the "D" (is not a vowel--the respondent should not turn over the card because this does not test the rule) * the "4" (must turn over because a non-vowel would disprove the rule) * the "7" (must be turned over because a vowel would disprove the rule) Only 5 out of 128 subjects got the correct answer. Ibid, p 126-7 end of footnote) --Unlike Computers, Humans Often Use Emotion Rather than Reason to Arrive at Answers. Two psychologists Daniel Kahneman and Amos Tversky have demonstrated that most human minds appear to be constructed in such a way that it INTUITIVELY GETS THE WRONG ANSWER (compared to using mathematics and logic) when confronted with real life decisions that involve statistical analyses. For example, Kahneman and Tversky posed the following problem to a sample group: A general surrounded by a superior enemy force must escape via one of two escape routes, to keep from having all his men killed: 1) If he chooses the first escape route, 200 of his 600 men will survive, with 400 dying. 2) If he takes all 600 men along a second escape route, there is a 1/3 chance that all the men will be saved and a 2/3 chance that they will all die. The wording of this problem was stated in terms of LIVES that could be SAVED, and (despite the fact in statistical terms the options are identical) 75% of the group sampled by Kahneman and Tversky chose the first option. Now, when the SAME problem was reworded in terms of lives LOST not lives saved--That is: The general surrounded by a superior enemy force must choose between two escape routes to keep from having all of his men killed: 1) If he chooses the first route 400 men of his 600 men will die. 2) If he chooses the second route there is a 1/3 chance that no soldiers will die and a 2/3 chance that all soldiers will die. In this situation, only 20% of the group sampled now opted for the first choice, (ie 80% chose the second option) as they wanted to avoid death. Even when the same groups were given both problems and recognized the contradiction, they gave conflicting answers. ("Decisions, Decisions" DISCOVER, June 1985 by Kevin McKean p 22) The point is, that the humans were solving the problem along emotional (ie minimizing death/maximizing life) lines, and thus did not appear to recognize that statistically both options were identical. How the Human Mind Easily Confuses Cause and Effect. Kahneman discovered another example of how the mind misinterprets naturally occurring statistics. While teaching a course on the psychology of training to Israeli air force flight instructors at the Hebrew University, he cited studies done on pigeons that demonstrated that reward was a better teaching tool than punishment to his class. A student then interrupted him: "With respect, Sir, what you're saying is literally for the birds. I've often praised people warmly for beautifully executed maneuvers, and the next time they almost always do worse. And I've screamed at people for badly executed maneuvers, and by and large the next time they improve. Don't tell me that reward works and punishment doesn't. My experience contradicts it." The other air force instructors in the class nodded in agreement with the student. Kahneman realized that his student was describing a real life example of the statistical principle of regression to the mean--ie where random studies show that in a series of similar events, an extraordinary one tends to be followed by a more ordinary event, and vise versa. Thus extra tall fathers, tend to have slightly shorter sons; brilliant wives tend to have slightly duller husbands; great movies tend to have less great sequels; etc. Regarding the pilots' observations that their students did better only after yelled out, Kahneman explained that as student pilots were learning aviation skills, their performance from maneuver to maneuver tended to be largely a matter of luck. Thus, regression would predict that a student who made a high score on the first maneuver, would do more poorly the next time-- regardless of whether he was praised or not. The same principle applied whether the pilot had been criticized or not. (This appears to be an especially difficult concept for most people to understand.) Thus, the human mind is constructed far differently than a computer which has been programmed to mathematically compute the correct statistical probabilities. Implications of Illogical Processing by Humans According to Morton Hunt, humans tend to naturally look for a PROOF, rather than a DISPROOF of what they are thinking: "we tend to think only in terms of tests that will confirm a hypothesis we hold but not tests that might disconfirm it." Indeed, many of us reach the conclusion we want, and short-circuit any information that might contradict this: "... most human beings earn a failing grade in elementary logic. But we're not just frequently incompetent, we're also willfully and skillfully illogical. When a piece of deductive reasoning leads to a conclusion we don't like, we often rebut it with irrelevancies and sophistries of which, instead of being ashamed we act proud." "Recently, a new mortality study released by the State Mutual Life Assurance Company of America reported that at all ages the death rate among smokers is more than twice as high as that among nonsmokers. That night on a television news program, a reporter was shown asking various smokers what they thought of the findings: one man sarcastically replied, "So nonsmokers don't die, right?"--and looked immensely pleased with himself; a young woman with equal self-satisfaction, said, "Nobody lives forever, anyway." Such rebuttals are not at all unusual; many psychological studies have shown that smokers tend to reject logical inferences about smoking by means of a variety of distortions and rationalizations. They may assert that the evidence is incomplete or biased, or cite the case of someone they know who smoked heavily and lived to be ninety, or, like the man and woman on television, rebut conclusions other than the one that was actually drawn." (Hunt, op cit., p 128) Likewise, other studies have concluded that people in general "insulate" themselves from reasoning around them,--unless this conforms to their preexisting attitudes and beliefs. (ibid, referencing Jeanne B. Herman of the Graduate School of Management at Northwestern University). This has implications in the political arena (where people vote as often for the personalities as they do the real issues), foreign policy, law, and of course religion. Role of Logic in Great Philosophical Systems Throughout history, individuals such as Aristotle, Spinoza, and Leibniz believed that, through logic, one could reason the answers to great metaphysical and moral questions. By the eighteenth century, such philosophers as Kant had shown that pure reason in the realm of ordinary existence could be used to produce conflicting conclusions. By the nineteenth and twentieth centuries, various mathematicians had invented other non-Euclid (but equally valid) geometric systems. From the work of Alfred North Whitehead, Bertrand Russell, Rudolf Carnap and others, it was proven that even the laws of logic had NO direct bearing on reality! Whereas logic and mathematics are important TOOLS for describing knowledge that we have--they can in themselves, produce NO NEW knowledge outside of their own system. This is not to infer that logic is valueless. On the contrary, logic is an indispensable tool for making conclusions from our hypotheses. The point is that even deductive logic, cannot solely in itself, prove if our hypotheses are correct. (This should not really be so surprising, because it is in essence, a restatement of the Garbage-In/Garbage-Out Principle--ie Where the perfectly logical computer will give the wrong answers if its input is wrong. See Section VIII, Chapter 4).