Every action of information involves the exchange of energy (remember chapter 9). This exchange of energy is used as feedback from the environment. This feedback is important for the evolution of the information.
Information gains or looses energy upon every action. Ongoing loss of energy would cause the destruction, the death of the information structure. Abstract information increases immediately its vertical insulation when it detects a loss of energy. This allows abstract information to "survive" many different experiences. The gain or loss of energy is memorized by the abstract information as experience (related to the used mapping).
The detection of a gain or loss of energy as a base for subsequent reactions is similar to the detection of favorable and unfavorable life conditions in biological life. Therefore I call this joy/pain detection.
A complex abstract information being uses an number of smaller abstract elements. Each of these elements can be instantiated in a number of ways. The number of combinations is often gigantic.
A particular mapping does not describe necessary the instantiation of all elements. Removing elements of a mapping leads to more abstract mapping information.
Each time an instantiation is used, the joy or pain associated with the instantiation (the gain or loss of energy) is memorized associated with the used mapping. This memory is called experience.
Applying knowledge in a given context is not always straight forward. It requires some creativity to use available concrete information as instantiation of an abstract element. In some cases, the result turns out different from expected because the knowledge was a generalization which did not apply to this particular case.
The experience of such failures (pain detected) is memorized together with the abstract information. Successful experience is also memorized.
This experience is necessary because none of the used abstractions is ideal and fully general (remember the hypothesis aspect). The experience describes exceptions and hints for successful use together with all mappings which have been tried out.
We can imagine an ideal information structure as a complex mechanism of abstractions in which the choice is as such that every combination of instantiations makes sense. This can of course only be realized when all used abstractions are completely universal. This is never the case.
Stored experience is in fact a form of patching to correct for lack of universalness of the abstract representation.
The ideal way to correct this should be to review the choice of the abstract elements in a way to explain all observations by another set of abstract elements for which there are no corrections (patches) required. In a real system containing already a lot of information, this is an enormous task which has to be repeated after every observation of something which did not fit.
The reconsideration is only possible for young abstract information which has not been applied a lot yet and thus not yet indurated. In this case, there can be a fundamental reconsideration leading to another choice of abstractions leaving the old set unused and finally dying.
If the abstract information is fully settled in a complex information environment, there is no better way than to store the exceptions as experience. Fundamental reconsiderations in a complex environment would cause an important loss of information.
When applying some information in a given context, the stored experience will be taken into account during the instantiation. The application will not only be an expression of the abstract information but also of the experience. This allows for an external abstraction process (e.g. another observer) to choose for more accurate representations of the abstract information or to complement the already existing representation with a part of the expressed experience.
In our mind, we develop abstract notions by mapping very universal abstract elements on some concrete elements. So, the structure of the abstract elements is not new but the mappings are new. When we observe something we cannot immediately translate by existing mappings, many abstract elements map upon (attempt to grasp) the new phenomenon. The collection of all these abstract elements makes the mapping only applicable to this concrete instance.
By selectively removing some elements of this cluster, the internal representation becomes more abstract. Thus, abstract notions do not develop as something new on top of concrete notions but rather by removing some unnecessary mappings from an initial structure which was too complex.
When we have an abstract insight, we discover that a lot of details in the mappings in our mind are not longer necessary and can be removed. The removal of these unnecessary structures gives rise to the release of energy which is experienced as joy. This makes us enjoy having an insight.
When a child develops the notion of a chair, at first it associates the notion of chair with one particular chair. To do this, a lot of (irrelevant) details are stored. Later on, it is forced to modify the internal representation of chair to fit a new example of a chair. To obtain this, some details in the information structure are removed.
In the early days of mechanical tools, some primitive pair of tongs was invented. Many jobs where attempted to be performed with the primitive tongs. Each time some restrictions where encountered, some small modification was made to the tongs to overcome the restriction. These small modifications are the induration of the experience of the usage in the tongs.
Each time a new pair of tongs was made, an attempt was made to include as much as possible of the experience made by the older tongs in the new one. Some of the modifications which where difficult to make on an old pair of tongs became integrated in the new tongs. By this, the tongs evolved to a quite universal instrument.
When we express some abstract information, we take the experience into account at the same time. By this, the expression seems to originate from a structure closer to the ideal structure (where no exceptions exist).
When a mechanism attempts to extract the structure from such expressions, the new structure will be slightly cleaner than the originating structure. This effect has caused the rectification of some information structure over many generations.
In other words, because the experience is expressed together with the instantiation of an abstract element, the propagated abstract information being is pulled towards a better choice for the abstract information. By this, the new structure will need less experience for the same result. This results in the following important finding:
The evolution of information is guided by the experience gained in the past.
The evolution of information by random mutations and natural selection exists but the effect of the mutations guided by experience is much much higher. The mechanism underlying larger steps of evolution is explained in chapter 17.
Many aspects of structures of information are reflected in the grammatical structures of language. In the structures of language, we can recognize the tendency to reduce the number of exceptions on the grammatical rules.
Due to contacts with other languages, it happens that some expressions of a one language are adopted by another. At first, such adopted expressions are stored as foreign structures not integrated in the language. When these structures are accepted as part of the language, children learn the new expressions as part of their mother tongue. They attempt of course to integrate them in their developing language structure. Because the new expression has a different origin, it does not fully fit and the problem is overcome by memorizing it as some exception. From this point, at least a part of the structure is integrated in the language. The exception causes a kind of "pulling" on the structure of the language to reduce the number of exceptions. After a number of generations, it can evolve to a slightly different expression which does not longer requires an exception. At this point there might be a small influence on the entire structure of the language.
These intermediate stages where a part of one language is half integrated in another causes an important mutual influence between the languages.
More in next chapter on Model of Environment
This is Chapter 11; Interaction with environment of
Behavior of Information
Author: Luc Claeys. All comments welcome, mail to lcl at this site: nanohome.be
Last updated on Nov 12, 1997