By Vladimir REDKO, Dr. Sc. (Phys. & Math.), RAS Institute of Applied Mathematics (named after M. Keldysh)

The process of the Earth's evolution has brought to life some really complicated and surprisingly effectively functioning living organisms. The functioning of their organs is coordinated by what we call biological management systems. What is the principle of their operation in the first place and could it be put to practical uses in our own interests, such as the development of artificial intelligence? What new technologies could appear on the basis of the "intellectual inventions" of Mother Nature? These and a host of other challenging questions are now on the agenda of evolutionary cybernetics-a new branch of science which is developing in this and other countries now.

One of the most complicated problems facing the international scientific community today is the origin of human intelligence. Can we get a more clear picture of the development of the cognitive abilities of living organisms in the process of evolution of the biosphere? In dealing with this challenge we should obviously try and model the evolution of the cognitive properties of living organisms. And the best way to deal with this problem seems to be to begin from the beginning -the origin of life, tracing the whole biological evolution from the primitive organisms to humans, trying to identify the most important of the "inventions" of nature which had led to the development of the first "sprouts" of logic, thinking and intellect. At the next stage we have to try and trace the causes of such qualitative changes by building their mathematical models. But, to begin with, let us try to identify, and briefly describe, certain levels of "intellectual inventions".

Level One-an organism can distinguish conditions of the environment and the memory thereof is recorded in the genome and is inherited. A living organism responds to environmental changes by altering its behavior. One such example can be the regulation of protein synthesis in response to changing nutrients in the environment according to the classical system discovered in 1965 by the French biologists and Nobel Prize winners, Francois Jacob and Jacques Monot. Say, colon bacillus usually feeds on glucose, but if that is missing and there is lactose instead, the bacterium begins to rely on the synthesis of special enzymes which process lactose into glucose. The above mechanism of regula-

Pages. 59

Bacteria magnified 30 - 40 ths times. Photo made with the use of electronic microscope.

Infusoria paramecium.

"Intellectual inventions" in biological evolution indicated rather conditionally.

tion of protein synthesis is an unconditional reflex at the molecular-genetic level.

Level Two-short-term memorizing by an organism of the state of the environment and adequate (also temporary) adaptation to it. That also includes habituation, i.e. gradual decline of the response to a biologically neutral stimulus. This simplest of the acquired habits, or skills, appears in single-cell organisms. Thus, if an infusorium is exposed to drops of water, it develops a fitting reaction, or response. But if the stimulation is repeated many times, the response gradually dies down. And it should be noted that habitation is different from fatigue: if an infusorium, which becomes used to falling drops of water, is exposed to some other neutral stimulus, the response reaction is restored (with infusorium, habitation lasts for about one hour).

Level Three-is the memorizing of stable links between events in the organisms' environment. An example is the classical conditioned reflex (studied by the Russian Nobel Prize winner, Academician Ivan Pavlov, in the first third of the last century). It consists of memorizing for a long time the connections between the conditional and unconditional stimuli (dog gets used to the connections between a sound signal and getting food).

A conditioned reflex is formed in three stages. The first can be described as "pre- generalization", when there is still no response to a conditioned stimulus, but there is an increased electric activity of different sections of the brain. This is followed by generalization, when a response appears both to a conditioned stimulus and also to various similar (differentiated) irritants. The response to the latter gradually weakens with the

Pages. 60

advent of the third stage-specialization; only the response to a conditioned stimulus is retained. The memory of the connection between the conditioned and unconditioned stimuli is long-term: in the lower vertebrates it is preserved for many weeks; in higher animals -up to several years or even for the whole life. The biological significance of a conditioned reflex, which is observed already at the level of mollusks-is anticipation of future events and adaptive use of this ability.

It should be noted that between the classical conditioned reflex and logic, thinking and human intelligence there are several intermediate levels. There is the so-called instrumental conditioned reflex which differs from the classical one in that an animal has to perform a certain act for which it

Pages. 61

had been trained in order to get the expected "reward". Beginning from a certain level animals can "form" models of the surrounding world.

Thus it turns out that we can single out several "key inventions" of nature on the way towards human intelligence and they can be arranged into a succession of evolutionary achievements or improvements. In the course of these "improvements" the process of "cognition" of regularities in the environment was becoming more and more complicated.

In a word, we can practically speak about the evolutionary roots of human intelligence. One of the proofs of that is an interesting analogy between a "deduction" of an animal in the process of formation of a conditioned reflex and elementary logic used by humans in proving mathematical theorems. In actual fact, after the formation of a conditioned reflex, that is the formation in an animal's memory of a link between a conditioned and unconditioned stimuli (CS and US, respectively) it can draw a certain "logical conclusion". If an event CS takes place, an animal anticipating a US event, "concludes" that it is necessary to expect a US event, or formally (CS, CS- >US)=>US. This is similar to using the classical rule of the "modus ponents" logic: "if В follows from A, and A takes place, В also takes place", or "(A, A ->B)=>B".

And that means that a dog in the experiments of Ivan Pavlov and a mathematician proving a theorem are using similar schemes. This, of course, is but an analogy: the "deduction" of a dog rests on a generalization of its experience and is of an individual nature, and the conclusion of a mathematician is purely deductive. Nevertheless in both cases the cause-and-effect links are at work and we are fully entitled to think about the evolutionary roots of our logical deductions, or mental conclusions.

Now, what has already been done in the area of theoretical modelling of the evolution and trying to understand how there appeared and developed in this process the cognitive potentials of biological organisms?

The situation has been as follows. There are plenty of mathematical and computer models describing the "intellectual inventions" of nature. One example is a model of the appearance of an unconditioned reflex at the molecular-genetic level which has been suggested by the author of this article (1990); the model of habituation of J. Staddon (1993, USA); many models of conditioned reflexes by Corresponding Member of the USSR Academy, Alexei Lyapunov (1958); of the American scientists S. Grossberg (1974), A. Bertoand R. Satton (1982), of G. Kloph (1993). All of these, however, are very fragmentary and form no general picture of the evolutionary origin of thinking. One can call them only the "ground work" with the main work still lying ahead. In the late 1980s and early 1990s research was started on the closely connected themes of "Artificial Life" and "Adaptive Behavior".

PROBLEMS OF MODELLING BIOLOGICAL LIFE

The central task before the students of "artificial life" is to try to understand and model the basic principles of organization of biological life. In the opinion of one of the "pioneers" of this research, K. Lengton (USA), the main assumption of specialists is that the "logical form" of an organism can be separated from its material basis.

Supporters of this trend also feel that they are investigating more general life forms than those which exist on the Earth, forms which could exist in principle and not only those which we know already.

Of great importance in studies of artificial life forms is computer modelling.

Although this trend of research was formally proclaimed at the first conference on artificial life forms in Los-Alamos, USA, in 1987, in actual fact models based on similar ideas had already been considered back in the 1950s and 1970s. In the 1960s, for example, Russian scientist Mikhail Tsetlin, suggested and studied models of automatic devices capable of adapting themselves to the environment. His studies initiated an interesting new trend of research which was called "collective behavior of automates". In the 1970s a project called "Animal" was suggested under the leadership of the Russian cybernetics expert, Mikhail Bongard, aimed at studying the adaptive performance of "artificial organisms" produced by mathematical modelling.

The main trend in the "Adaptive Behavior" project which emerged in the early 1990s was designing and building artificial (in the form of computer programs and robots) "organisms" capable of adapting to the environment. They were called "animates" (from the words "animal" and "robot").

The program-minimum for researchers in this field is to identify the principles of functioning of the systems of management, or guidance, which make it possible for animals, or robots, to exist and function in a changing environment. The program- maximum is to try and analyze the evolution of the cognitive abilities of animals and the evolutionary origin of human intelligence.

The studies "Artificial Life" and "Adaptive Behavior" have much in common: a synthetic approach to the "development" of life-like "organisms", attempts to model the formal laws of life and systems of management, orientation on computer and mathematical models and the use of evolutionary concepts and models. Both areas of research use a number of what we call non-trivial computer methods: artificial neuron networks; teaching combined with rewards; genetic algorithms, etc. These are the key area of development of evolutionary cybernetics now.

BASIC CONCEPTS OF EVOLUTIONARY CYBERNETICS

It has to be pointed out at this stage that modelling alone is clearly not enough for covering the whole diversity of the evolution of cybernetic sys-

Pages. 62

terns. This being so, the building of specific mathematical and computer models should proceed in conjunction with the development of some general scientific and philosophical concepts and methodology of evolutionary cybernetics. Within this context one should single out two conceptual positions which can provide the basis for the new scientific discipline.

First, this is the theory of functional systems which was formulated in the 1930s- 1970s by the Soviet neurophysiologist, Academician Pyotr Anokhin. In accordance with this theory the management of the organism of an animal consists of: sense of purpose linked with the need to obtain the required result; dominant motivation, setting up the prerequisites (such as those dictated by needs) for the formulation of an objective; afferent synthesis (recognition of a situation and preparation for action); taking the decision; prognostication of action and formation of a notion of the final result of action; implementation of the action itself; assessment of the obtained result; comparison of the prognostication and actual result; learning.

The aforesaid integral parts characterize a general and sufficiently universal scheme of the system of management of an animal's behavior. This scheme can provide the basis for developing models of the "intellectual inventions" in nature.

Second, there is a theory of meta-systemic transitions developed in the late (1960s by a physicist and expert in cybernetics, Valentin Turchin - a former Soviet scientists and now citizen of the United States). Very briefly, it boils down to the following: elevation from the lower levels of the hierarchy to the upper ones goes through meta- systemic transitions. Each can be regarded as bringing together of a number of sub- systems S _i (i=1, 2,...,n) of the lower level and the appearance of an additional mechanism of their management (C). As a result a system S* is formed of a higher level (S*= C+S ₁ +S ₂ +...+S _n ) which, in its turn, can be included as a subsystem for the next meta-systemic transition.

Dr. Turchin pays special attention to the quantitative accumulation of the "development potential" in the subsystems Si before they reach a qualitatively new level of the hierarchy, and the process of multiplication and development of the subsystems of the ultimate layer of the hierarchy after the meta-systemic transition. The latter can be regarded as some cybernetic analogue of a physical phase transition. It has to be stressed that it was Dr. Turchin who brought into the circulation the term of "evolutionary cybernetics".

SEARCHING FOR ANSWERS TO ETERNAL QUESTIONS

The progress of evolutionary cybernetics is linked not only with studies of the cognitive potential of living organisms, but with viewing the future of the human community, especially in the context of the intensive development of modern information or data-handling technologies. Work is now in progress under the international Internet-project "Principia Cybernetica Project". Scientists are trying to find answers to the eternal questions like: Why is the world such as it is today? What is the origin of ourselves (humans, mankind)? Who are we? Where is the evolution taking us to? In the spring of the year 2000 specialists formulated the idea of evolutionary cybernetics as a broad spectrum of studies related to Nature, Thinking, Society and Technologies. The new discipline should provide the basis for studying the cybernetic evolution of mankind, including an analysis of future systems of management of the human community.

One of the subjects most actively discussed is the problem of what is called the universal brain. In the view of the Belgian scientist Prof. Francis Hailighen, "...this is a metaphoric description of the emerging intellectual network which is being formed by mankind on the basis of utilization of computers databases and links which unite everything into a single whole. This network represents a complex, self-organizing system, which not only processes data, but is gradually acquiring functions similar to those of the brain: taking decisions, solving problems, learning, formation of new mental associations and development of new ideas."

The way I see it, the second phrase in this definition is rather debatable, but there can be no doubts that what we need is a profound and comprehensive analysis of the ways of development of the emerging world-wide "intellectual network" and studies of its economic and social consequences.

The year 2001 saw the first international conference on the problem of the universal brain. On the whole an impressive attempt has been made to assess the evolution of the man-machine network technologies in order to be able to anticipate the future of their progress.

* * *

Summing it up, one can say that studies in the field of evolutionary cybernetics are very topical. They make it possible to analyze the process of development of the cybernetic capabilities of biological organisms; clarify the causes of origin of the cognitive potential of man; can provide the basis for a range of applied uses -from the evolutionary methods of optimization of robot management systems to the principles of management of the human community. What is more, studies in the field of evolutionary cybernetics give us hopes for a radical breakthrough in our knowledge and understanding of the surrounding world and its evolution.

Studies have been conducted with the financial support of RGNF.

Code 00 - 03 - 00093.

Постоянный адрес данной публикации: