Self-training control block. No brain is able to hold enormous “knowledge bases” on all possible conditions of the entire world around. Therefore, one of the reasons why each species of animals occupies corresponding biosphere niche is the necessity to limit the volume of “knowledge base”. Antelope knows what the seal does not, and vice versa. In each separate ecological niche the quantity of possible situations is much less, than in all ecological niches all together. Therefore, relatively small volume of necessary knowledge is required in separate ecological niches. However, if one tries to somehow input /in the brain/ all the information currently available on all the situations which have already been occurring in the world, it would not help either, because the world alters continually and many situations have never ever arose. The “knowledge base” basically may not have information on what has not yet happened in the world. Naturally, the “base of decisions” cannot contain all the possible options of decisions either. “Genetic knowledge” contains only what the ancestors of animals have experienced. They materially cannot have knowledge of what is going to happen. When new situation arises, the system cannot identify, classify it and make decision on it. Even if this situation will occur repeatedly, if the system is unable of self-training it will every time fail to correctly identify a situation because such situations are not contained in its “knowledge base”. The ant runs along the fence, going up and down, and cannot guess that it is possible to easily bypass the fence. Millions years ago, when its genetically input “knowledge base” was formed the fences were non-existent. If one tries to sink a thread on the web the spider will leave this web and will weave a new one because it is not familiar with such situation and it does not know and cannot learn that it is possible to make a hole in a web so that the thread does not interfere. All this is due to the fact that insects as a class of animals are not capable of learning anything. They may be perfect builders amazing us with their sophisticated and fine webs, nests and other creations of their work. But they can only build based on their innate knowledge. They do have “knowledge base” (instincts), but they do not have cerebral structures (elements of control block) capable of supplementing their own “knowledge base” with new existential situations. They do not have reflexes on new stimuli/exciters/. To be able to identify and classify new situations the control block should be able to enter the descriptions of these situations in its “knowledge base”. But at first it should be able to identify that it is a completely new situation, for example, by comparing it to what already exists in its “knowledge base”. Then it should identify the importance (the value worth) of this particular situation for the achievement of its goal. If there is no any correlation between the new situation and the fulfillment of the goal of the system, there is no sense in remembering this situation, otherwise the brain “will be crammed with trash”. By singling out and classifying external situations (identifying them) and finding interrelation (correlation) between these situations, by decisions made and the achievement of the goal of the system the control block learns to develop appropriate decisions. Thus, the self-training decision-making block continually supplements its “knowledge base” and “base of decisions”. But under the conservation law nothing occurs by itself. In order for the control block to be able to perform the above actions it should have appropriate elements. The major element of the kind is the analyzer-correlator. It is the basis whereon reflex on new stimulus/exciter or a new situation may emerge. Its task is to detect a new situation, identify that it is new, determine the degree of correlation between this situation and its own goal. If there is no correlation between this new situation and implementation of the goal by the system, there is no sense in remembering and loading its limited “database” memory. If the degree of correlation is high it is necessary to enter this situation in the “knowledge base” and develop a decision on the choice of own actions for the achievement of its own goal and thereafter to define whether there is correlation between the decision made and the achievement of the goal. If there is no correlation between the decision made and the fulfillment of the goal by the system it is necessary to arrive at other solution and again determine the correlation between the decision made and the achievement of the goal. And it should be repeated in that way until sufficiently high correlation between the decision made and the achievement of goal is obtained. Only afterwards the correct computed decision should be entered into the “base of decisions”. This is the essence of self-training. Only the analyzer-correlator enables self-training process. As a matter of fact, the system's self-training means the emergence of reflexes to new stimuli/exciters or situations. Consequently, these are only possible when the control block contains analyzer-correlator. Biological analogue of the analyzer-correlator is the cerebral cortex. The presence of cortex determines the possibility of emergence of reflexes to new situations. Cerebral cortex is only present in animals which represent sufficiently high level of development. Non-biological analogues of systems with such self-training control block are unknown to us. Computer self-training systems are built by man and the process of self-training at the end of the day always involves human cerebral cortex. There exist various so-called “intellectual” systems, but full-fledged intelligence is only inherent in human being. Let us specify that there are no self-training systems, but there are their self-training control blocks, because executive elements cannot be trained in anything. There may be systems with simple executive elements, but with control blocks of varying complexity. In order for the control block to be a self-training structure it should contain three types of analyzers: the analyzer-informant with “database”; the analyzer-classifier with the “knowledge base” and “base of decisions” (which is able of classifying external situation on the basis of the information from the “C” informant); the analyzer-correlator (able of identifying the interrelation - correlation between various external situations and the results of actions of the given system and transferring the knowledge obtained and decisions to the analyzer-classifier to enter them in the “knowledge base” and the “base of decisions”). Thus, the system with self-training control block is an object which can learn to distinguish new external influences and situations in which such influence may be exerted. For this purpose it has the analyzer-correlator. In other respects it is similar to the systems with complex control block. It can respond to specific external influence and external situation and its reaction would be stipulated by type and number of its SFU. The result of action of the system is also graduated. The number of gradations is determined by the number of executive SFU in the system. It also has analyzer-qualifier with “knowledge base” and “base of decisions” and the analyzer-informant with “database”, DPC (the “X” informant) and NF (the “Y” informant), which operate the system through the stimulator (efferent paths). In inorganic/inanimate nature there are no analogues of systems with self-training control blocks. Biological analogues of systems with complex control block are all animals with sufficiently developed nervous system in which it is possible to develop reflexes to new situations (should not be confused with conditioned reflexes). The analogue of analyzer-correlator is only the cerebral cortex.
Signaling systems. The appearance in the control block of the analyzer-correlator enabled the possibility to enhance its personal experience by self-training and continually update its “knowledge base” and “base of decisions”. But it cannot transfer its experience to other systems. Personal experience is limited howsoever an individual would try to expand it. In any case collective experience is much broader than that of an individual. In order for one individual to be able to transfer his/her experience to other individual separate device is needed enabling “downloading” the information from one “knowledge base” to another. For example, the antelope knows that the cheetah is very dangerous because it feeds on antelopes and wishes to transfer this knowledge to its calf. How can it be done? For example, the antelope can simulate a situation playing a performance in which all characters are real objects, i.e. it should expose itself to cheetah so that the calf could see it to gain its own experience by the example of its mum. The calf will see the situation and new reflex to new situation will be developed and the calf will be on its guard against the cheetahs. Of course, it is an absurd way as it does not solve the problem of survival. Anyway, only one out of the two antelopes will survive. So, what can be done in principle? How one self-training system can transfer its individual experience to other self-training system? It is necessary to simulate a situation by making a show in which all characters are abstract objects and replace real objects with others, which are conferred conventional connection/link between them and the real objects (abstracting of objects). Such abstract objects are prearranged signals. The systems “agree” (stipulate a condition) that if such-and-such signal occurs, it will speak of something agreed upon. It is the development of conditioned reflex that represents replacement of real influence for abstract influence. It is a so-called first signaling system which is based on conditioned reflexes. The appearance of cheetah causes producing a panic sound by an antelope. Consequently such sound is associated with the appearance of cheetah and it becomes an abstract substitute of cheetah itself, i.e. prearranged signal. Any motional signal may be an abstract substitute of danger, i.e. raising or dropping of tail, special jumps, producing special sounds, mimicry, etc. These motional signals affect the systems in the herd and based on this signal they may know about a danger nearby. In other words, there was a replacement of real external influence by some abstract thing associated with this object. Abstracting of real action by its symbol (vocal, motional, etc) took place. For such abstracting the control block needs to have an additional device - the analyzer-abstractor which should contain the “base of abstraction” (“base of prearranged signals”). The “base of abstraction” contains a set of descriptions of certain signals which are perceived as conditional situations and correspond to other certain situations. A prearranged signal is the appearance of some object or movement (situational signal) which usually does not appear in common routine situation. The occurrence of prearranged signal does not in itself affect in any way the achievement of the goals by the systems. For example, raising and fluffing out a tail does not influence in any way neither food intake, nor running, etc. But the occurrence of a signal is connected with the occurrence of such situation which can affect the achievement of goals by the systems. Given the ability to abstract from concrete situations, then not even seeing a cheetah, but having seen the lifted tails, may be conducive to guessing that a cheetah is nearby. Abstracting of real external influence by vocal or motional symbol is performed by the first signaling system. It supplements the analyzer-correlator and operates similarly to it, i.e. is self-training. Unlike the “knowledge base” the “base of abstraction” of a newly born system is empty. It is being filled out during the system's lifetime on account of possibility of self-training, and the newly obtained knowledge is then downloaded in the “knowledge base”. Sometimes behavior of animals seems to be indicative of their possibility to transfer the information from one to another even before the occurrence of the respective situation. For example, some lions go to an ambush, others start driving the antelopes, so they kind of foresee the situation. But they only know about ambush possibilities based on their own experience. They do not have other means of transfer of such information to their younger generation except for demonstrating this situation to them. A new way for the development of systems (or rather their control blocks) is being opened at this point, the way of socialization - associations of animals in groups for the enhancement of their own experience because prearranged signals are only intended for an information transfer from one system (subject) to another. There are probably several levels of such analyzer-abstractor and the degree of abstraction which may be attained by this or other subject depends on the number of these levels. One may abstract external influences, external situations, real objects and even process of self-training proper. But in any case one should be able to abstract and understand abstract symbols. This is what analyzer-abstractor does. Abstracting of real external influence, object or situation by means of situational prearranged signal (a pose, a sound, a movement, some kind of action) may be performed by the first signaling system. Abstracting of real external influence, an object or a situation by means of sign /emblematic/ prearranged signal (symbol) can only be performed by second signaling system. Control block having the second signaling system is an intellectual control block. Intelligence depends on the presence and the degree of development (number of levels) of analyzer-abstractor. In animals the second signaling system is very poorly developed or undeveloped at all. If the horse dashes aside from a whip, it is not even the first signaling system that works in this case, but rather a reflex on the new situation which the horse has learnt when it first encountered a whip. If the horse is coarsely shouted at even without showing a whip to it, it will draw necessary conclusions. That's the point at which the first signaling system takes effect. But if the horse is shown an inscription which reads that it now will be beaten, the animal will not react to in any way because it cannot and will never be able to read since it does not have second signaling system. There are animals which apparently are capable of speaking and understanding words, written symbols and even making elementary arithmetic operations. But the second signaling system is very poorly developed in them and is literally “in embryo” condition. When the trainer demonstrates the dog's counting up to five, he bluffs in a way as in fact the dog picks up some motional signals from him, i.e. the second rather than the first signaling system takes effect. The second signaling system is developed to the utmost extent only in human beings. In human beings it is developed to the extent that it makes it possible to transfer all necessary information on our further actions to us in the nearest or even quite a distant future only by means of sign symbols. We can read a book containing just mere squiggles only, however such a full-blown and colorful pictures are open before us that we forget about everything on earth. Your dog for sure is surprised that its master looks for hours at a strange subject (the book) and does not move, run or make any sounds. And even if you try to explain to it that it is a book the dog will not understand it anyway, because it has not yet “matured”, it does not have second signaling system. Thus, the system with self-training control block containing the first signaling system is an object which can abstract external influences and situations by means of abstract situational prearranged signal. For this purpose it has an analyzer-abstractor of the first order. But it can inform of the presence of such action or situation only at the moment of their occurrence. It may transfer its experience to other systems only with the help of the situational prearranged signal which possibilities are limited. Such block has the “knowledge base” and “base of abstraction” which it accumulates in its brain within the lifespan. In the communities of systems with first signaling system accumulation of personal knowledge is possible, whereas accumulation of social knowledge is impossible because this knowledge is accumulated only in the control block (cerebrum) which possibilities are limited. The system which has self-training control block containing the second signaling system is an object which can abstract external influences and situations by means of abstract sign /symbolic/ prearranged signal. For this purpose it has an analyzer-abstractor of Z-order. It can transfer its experience to other systems by transfer of information to them in the form of conventional signs. Such blocks accumulate “knowledge base” outside its cerebrum in the form of script thanks to the developed “base of abstraction”. It gives an opportunity to absolve from dependence of accumulation of knowledge on the lifespan of an individual subject. In communities of systems with the second signaling system accumulation of social knowledge is possible and it strengthens the accumulation of individual knowledge. In other respects the control block with signaling systems is similar to the self-training control block examined above. It can react to definite external influence and learn to react to new external influence and an external situation, and its reaction is determined by type and number of its SFU. The result of action of the system is also graduated. The number of gradations is determined by the number of executive SFU in the system. It also has the analyzer-correlator, the analyzer-classifier with “knowledge base” and “base of decisions”, the analyzer-informant with the “database”, DPC (with the “Х” informant) and NF (the “Y” informant) which through a stimulator (efferent paths) operate the system. In an inanimate/inorganic nature there are no analogues of systems with control block having signaling systems. Biological analogues of systems with control block containing the first signaling system are all animals with sufficiently developed nervous system in which conditioned reflexes may be developed. As a rule such animals do already have social relations (flocks, herds and other social groups), as signals are transferred from one animal to another. Biological analogue of systems with control block containing the second signaling system is only the human being.
Страницы: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
|