Some considerations on the connection between evolution and knowledge acquisition

The principles of competitive evolution imply some form of locality, in time, of the competition. Local advantages play, on a given time scale depending on many specific factors, a major role in determining what characteristics will dominate in future generations of a given species of surviving systems. This effect is well captured by the bifurcation phenomena in complex systems dynamics.

This character of competitive evolution implies problems similar to those affecting local search of functional maximum by local gradient methods or similar techniques. Mathematical techniques of global research are affected by other well known problems as their incapability of identifying in general sharp maxima. In both cases techniques can be improved in several ways [ ]. Genetic Algorithms (G.A: since now) have been able to improve dramatically the functional maximum search by an implicit mixture of local and global search [ ].
Knowledge acquisition by evolving competing systems is affected as well by the complex interaction between local and global optimisation.
Local, in time, advantages play an important role in determining the survival of system acquiring small, but competitively relevant, advantages independently on the long term consequences on the optimisation of the knowledge itself from a global point of view. The consequences of this local effect are important in the understanding of both the physical and abstract tools that , at a given time, an evolving system uses to monitoring and interpreting the external world. The relevance of the local analysis component in the success of G.A. search with respect to global search techniques shows its important role.
Genetic Programming [ ] captures some aspects of knowledge evolution. It has been shown in facts how the genetic search for programs capable of generalisation of variables relationships is affected by the appearance in the codes of subroutines that appear with a significant frequency [ ]. These subroutines are capable of capturing some essential preliminary relationships between variables, useful for identifying the global formal relationship between them. In other words, those subroutines represent some form of basic knowledge acquires by the system.
From the point of view of competitive evolution, the acquisition of more efficient subroutines can give relevant advantages to some system. Anyhow, also in this case, the complex interaction between local and global dynamics can lead to some form of ?Irrationality? if the evolution trends could be judged by an external observer completely aware of all the characteristics of the external environment. Such a hypothetical observer could see the global inefficiency of subroutines capable of giving nevertheless an important local , in time, competitive advantage [ ] .
Evolution engages not only the knowledge of systems, but their modules responsible for monitoring the environment as well. A better and more complete monitoring device can obviously affect the competitively of the system. We need however to understand what ?Better? means in term of evolution. We should avoid he risk of identifying ourselves with the external observer with a complete a priori knowledge of the environment and its evolution. In so doing we will completely loose any understanding of the complex meaning of the term from the point of view of evolution. The local, in time, efficiency has to be taken again into account. Actually a change of the monitoring module cannot be transformed into a positive evolutionary contribution if the knowledge module cannot make use of it is some competent way. We ignore for the moment the physical module used by the system to transform knowledge into action.
Baldwin?s contribution to the understanding of the evolutionary mechanisms has somehow clarified the points underlined above [ ]. We concentrate here only on the relationship between knowledge and monitoring.
It is clear at this point how some form of co-ordination between the evolution of the knowledge and monitoring modules of a system would represent a powerful tool to improve its competitive success. It's however also evident that, while considering possible mechanism capable of implementing such a co-ordination, the risk of implicitly assuming the position of the omniscient external observer is very high and would lead to major misunderstandings. If such a mechanism exists, it should be based on some simple but effective property of the evolving system, interacting with the relevant properties of the surrounding environment.
From the anatomical point of view there is a wide evidence of the dominance in the brain of re-entry circuitry, which represents the minimal condition to support the hypothesis of co-ordination between monitoring and knowledge modules.

If the above considerations are accepted, the essential unconscious nature of the vast majority of brain processes involved in what we would otherwise define ?Common-sense thinking? Is evident. Old modules are still active in orientation and co-ordination of incoming environmental data as well in decision taking processes. As they represent the historical stratification of epochs were consciousness was not even conceivable, their activity, when taking place, is most probably unconscious. Higher modules can of course register the results of their activity and make it conscious.
There exist, on the other hand, a vast psychological evidence of the importance of a correct co-ordination of emotional reaction in supporting the competent approach to complex activities that rationality could, at least in principle, take care of [ ]. Contemporarily a complex bi-directional interconnection between emotional brain and the neocortex has been made clear


We have a tendency to be affected by a quite cumbersome perspective error when considering the dynamics of evolution. This is due to the obvious fact that the vast majority of the evolution results are actually unavailable to our observations for the simple reason that they were unsuccessful at a given time.
Back to interests
Back to Home Page