AI: The Technological Trickster
By ai-depot | October 26, 2002
Do Humans Think, or Do We Just Think We Do?
Whether information possessor or processor “…the brain, with its 100 billion or so neurons, has at least 2 possible states…[far] exceeding the number of protons in the known universe…” (Casti, p. 399) But the question of how it turns this complexity into meaning is not known. Attempts to use computer problem solving and information processing as metaphors for brain activities have in some instances served to confuse the issues further.
Ernest Kent suggests we know “…two facts about the brain: there is an evolutionary order of development of its structure and the major functions have representation at all levels… [suggesting] newer structures simply took control of the older ones and used them as subprocessors.” In the process of evolution, a “…general principle of organization evolved in which the higher level structures control the lower …by inhibiting their actions.” (Kent, p. 17) Unfortunately, there is much confusion about which processors control which parts of the brain. Freudian psychology, for instance, is based on the idea that we are motivated by lower orders of complexity acting on the higher orders through drives buried in the subconscious mind. Nevertheless, most brain activity, perhaps 90%, seems to be concerned with inhibition of electrochemical processes.
Kent considers most brain activity precisely analogous to computer activity. He notes there are “…at least two types of learning that the brain permits…[first,] things that have occurred sequentially several times are likely to do so again… [and, second] operant conditioning… that permits us to expand our behavioural repertoire and base such expansions on the quality of the results.” As to the “…physiological mechanisms controlled by the brain… there are only two types, muscles and glands.” (Kent, p. 21-22, 24) Thanks to the complexity of the system and of the chemicals involved, we manage to get quite a range of behaviours out of these few mechanisms.
Research has shown “…the two hemispheres of the human brain are specialized to deal with problems …by the use of two distinct types of paradigms.” One uses a “…sequential approach… that considers only a small portion of the available data…” while the other uses a “…parallel (or gestalt) [process that]…processes data… all at once.” (Fischler and Firschein, p. 13-14) The whole system may be thought of as operating in parallel across the corpus callosum. Sternberg has suggested: “Systems that operate in parallel allow for the simultaneous and independent consideration of competing, and even contradictory, hypotheses…” (Sternberg, p. 145) a human ability as any reading of a newspaper will attest.
The ability to parallel process information likely has survival value, especially with respect to problems that have no ‘right’ answer, or ones that consist of a series of wrong answers with no a priori means of discovering which will be the least damaging. Life situations provide many of these, and computer problem-solving is generally not well equipped for them, being designed in the lab for specific purposes.
The most recent part of the brain to evolve, “…the prefrontal cortex… has two distinct aspects to its operation; feedforward and look-ahead planning.” Kent suggests this has led to “…improvement of system performance in three major areas, motivation, input and output, by the introduction of feedforward mechanisms which can prepare a ‘best guess’ configuration of the system.” (Kent, p. 193, 197-198) This ability to gain understanding of a problem using heuristics, general ideas about a problem rather than lists of priorities or specific attributes, is one of the human brain’s most important characteristics, and one that has eluded programmers and engineers of AI. The algorythyms necessary for programming heuristics depend on a broad base of world experiential knowledge and are probably a result of selective pressures on populations over long periods of time.
Although Kent considers the brain a “…machine for solving real-world problems…” (Kent, p. 233) he doesn’t detract from the debate by insinuating there is nothing more the brain can be than this. According to John Casti, “…Piaget thought of human intelligence as a process of reality construction rather than as a passive receiver of information…” (Casti, p. 238) And Noam Chomsky feels language “…serves essentially for the expression of thought.” (Chomsky, 1977. p. 88) which may be more important than information processing, as such, in any future evolution.
Michael Posner and Marcus Raichle, in the tradition of Roger Penrose of Toronto take the position that particular areas of the brain are responsible for particular activities, that “… different codes… represented in a group of localized brain areas… would be activated in concert…” Their work consists of “…picturing networks of anatomical areas that become active during the performance of mental tasks.” (Posner and Raichle, p. 16, 22)
They have discovered, for instance, that any word impacts the brain in at least three codes relative to “…visual aspects…phonological aspects…or semantic aspects…” (Posner and Raichle, p. 109) They have also discovered “…there can be a considerable distance between the location of the same functional area in different brains…” suggesting “…individuals may differ in the strategies they apply to even relatively simple tasks like reading or remembering words.” (Posner and Raichle, p. 232) This adds another aspect to the debate: exactly where does thinking take place? While specific sites do seem to be specialized for particular types of knowledge or processing, it is the whole brain that seems to be involved in actual thinking. (When the whole brain is thinking, which may be all the time, we tend to want to call it by another name: mind. This has resulted in some confusion that there is somehow a qualitative difference between the two, mind and brain. This paper assumes the two terms are equivalent.)
Research has discovered that in “… children the right hemisphere can relearn lost abilities if the left is damaged. After age 12 to 14, when lateralization of language is almost completed, the effects of aphasia tend to be more permanent. This is also the age when people… find it more difficult to learn a new language.” (Furst, p. 136, 145-146) We may assume language learning ability is intimately involved with thinking, and perhaps may be used to define the limits of human intelligence and/or the ability of the brain to process information. Computer limitations in this regard, if any, may be based in other traits or characteristics.
The differences in the abilities of the two brain hemispheres suggest in the brain’s evolution “…functional asymmetries, as for example in language, visio-spacial and music, in development of the neo-neocortex provide for a variety of behaviours without increasing the size of the brain…” (Eccles, p. 215-216) The trade-off between brain size and birthing ability is a factor of human evolution that may indicate optimum sizes of data bases given a particular size of land mammal. Thus we have arrived after some millions of years at a situation where we may evaluate our own representations with some degree of objectivity. Perhaps.
According to Martin Fischler and Oscar Firschein, “The mind apparently uses two major representations, propositions and images…”(Fischler and Firschein, p. 185) Each is hemisphere dependent in general terms, but either hemisphere is apparently able to learn how to deal with both or either, if necessary, before the age of twelve or so.
Ultimately we must conclude that humans think, because we have defined thinking in a human-centred and specific way that makes it difficult or impossible to conclude otherwise. Thinking is one of those activities that humans and other animals do. The Cartesian paradox is as real today as it was 400 years ago. We are able to doubt everything except our own ability to think. Yet there seems to have been no means to proceed beyond that premise, until the advent of machines that may be capable of some kinds of thought.
Whether this is enough to guarantee what kind of existence this implies is still open for debate. Computer “thinking” may offer a mirror of the thinking process that will provide some clarity to the latter question, although it is more likely to be a distorted fun-house mirror than one that gives an accurate image. Analysis of the distortion may yield some testable images, if we proceed with caution.
How might we derive a means top test as to whether the metaphor of a thinking machine is indeed correct? In past times, the workings of the brain have been compared to the catapult, flour mill, clock, printing press, steam engine, telephone system and, most recently, the computer, depending on the focus of the age. The resulting paradigms may have been adequate for the times. But it was probably the inadequacies of the metaphors that led to problems that defined the succeeding paradigms. As Jerry Mander has asked his detractors, “What comes after the computer?”
Tags: none
Category: essay |
