Cognitive process

The beginning of evolution, the passive assimilation of information, is at the same time the beginning of building up a model which the system requires in order to deal with its environment. A cognitive process ensues, spanning from the first passive phylogenetic learning steps all the way to conscious realization on the part of the system, and to organized research undertaken by an entire society. The first consequence of information assimilation for a system is that a targeted selection of different system conditions can take place. The reason for this first information assimilation, evolutionary pressures, continues and remains in existence for the learning process of higher-level systems. Playful and curious behavior also came into existence as a consequence of these pressures.

One of the first consequences of the system-changing phylogenetic learning process might have been the acquisition of adaption functions which made a reaction to environmental changes possible. The first steps from passive to active information reception were done. Certainly, at the outset there were chemical sensors which enabled, for the first time, the individual and not just the species to adapt to environmental changes. Here, in the transit from passive to active information reception, we can clearly recognize the enormous acceleration in the speed of evolution.

Control functions - no matter how they are realized - which enable the individual to react to environmental stimuli represent the first primitive theory the system has passively acquired from its environment. The system has "grasped" an interrelation, for example between chemical stimuli and the presence of food. These interrelations describing and imaging the environment provide an additional advantage, with regard to natural selection, so that evolutionary pressures point in the direction of active acquisition of such interrelations. Therefore, the next big step are active learning systems. Of course this represents continuous evolution because the functions are taken over into the system in an individually hierarchized manner, according to their relative significance. Consequently, there will be very highly developed adaptive functions in a system which is otherwise not yet capable to react to the environment, which is not yet capable of moving in many dimensions of the state space environment, and which therefore still has to be adapted via selection. Evolutionary pressures affecting single functions are the determinant. Optimization according to the orientation processes gains in significance.

Consequently, the system has already assimilated part of the structure of the state space surrounding it in an informative manner, that is, it is in possession of a theory of the environment which enables it to actively adapt to the state space. On closer inspection, this environmental model for the fight for survival presents itself in a hierarchically structured way. Control functions of a general nature are settling on "higher-level" states than more specified, mostly older functions. They can also utilize a "lower-level" function on a mutual basis. For example, motor activity functions are used for different purposes. First, the simplest and most important functions, subject to the strongest evolutionary pressures, are adopted. Later on, functions of a more general nature will join and resort to lower hierarchy levels. They can easily be understood as command variables for existing functions/control loops of "lower" levels. See Fig. on pg. [*]

Another reason for the development of a hierarchical model structure is the next crucial step in nature's cognitive process. The reason for this step seems to be an economical one. The multiple use of lower-level hierarchical functions by others seems to be advantageous. The possibility to combine similar interrelations to a more general form becomes a necessity in the face of consistent accrual of memory requirements. Evolutionary pressures work in the direction of abstraction of the stored information. Next to the advantage of less memory requirements, abstraction offers the very notable possibility of extrapolation beyond the immediate sphere of experience. As long as an abstract environmental model is realistic it can provide information hitherto inaccessible to the experience. The memory-efficient, more general presentation enforces a backward projection making the information useful for the special case at hand. This backward projection now permits the insertion of many specific input data. Therefore, abstract model and backward projection provide considerably more information than that which was supposed to be stored originally. In this way, information becomes accessible via conditions for which no experiential data exist at that time. However, this enormous gain introduces a certain risk. The theory does not have to be adequate, thus realistic - at least not for those cases which have been extrapolated. The theory therefore can be correct for all cases which it was abstracted from, yet it still carries a risk of error for all extrapolated cases. For example, these differences are being exploited by predators (enemies) for the purpose of setting traps.

Eventually, this is the basic objective of all scientific work: the creation of a theory which is free from contradictions, adequate, and preferably all-embracive.

Thus, tracing the information reception within the evolutionary process leads in one line from the primordial inception of life all the way to scientific activities of man and/or society.