Some rules of the game for self-organization

Many systems have specific purposes which only can be fulfilled by a high degree of complexity characterized of many parts which interact in a well-regulated fashion. A good example is how a mind could emerge from a collection of mindless parts in the human brain. What we need to understand is not the behaviour of individuals parts but rather their orchestration. We also must realize that complicated behaviour can be realized on quite different substrates and that the more complex a system is the more it will show traits of what we experience as intelligent behaviour.

A look at high-way traffic, market economies, immune systems, bird flocks, fish shoals, antelope herds and ant colonies show no central authority or coordinator. They show how large scale, orderly patterns emerge by local random interaction among decentralized components and how local rules make global dynamics. Many similarities exist in the behaviours of minds, animals, societies and machines. In the computer area, neural networks are self-organizing in the sense that they learn to make the right computations to accomplish a given assignment. Self- organization and selection can also be studies in Genetic Algorithms and Artificial Life. In each case, the behaviour of the whole is far more sophisticated than the behaviours of the constituting parts. Collective structures behave very differently from the parts which compose them.

Flocking behaviour among birds has been investigated by several biologists. If a flock is going to form, its does so from the bottom up, creating an emergent phenomenon following three main rules.

  • Maintain a minimum distance from objects in the environment, including other birds.
  • Try to match velocities with birds in the neighbourhood.
  • Move toward the perceived centre of a mass of birds in its neighbourhood.

These rules may be compared with the rules influencing intelligent robotic behaviour presented on p. 325.

Ant colonies are particularly illustrative as prototypical example of how complex group behaviour arise from simple individual behaviour. How randomness is replaced with purpose can be seen in an ant colony. Once the number of ants passes a certain number, there is a significant leap in their ability to learn, survive and prosper. Any larger would cause too much confusion. Any smaller and a message would not be passed on at all.

Some general principles, derivered from ant research but relevant for most complex behaviour, has been formulated by Steven Johnsson (2001).

  • More is different. The statistical nature of interaction demands a certain lowest critical mass in numbers for complex behaviour to take place.
  • Ignorance is useful. A densely interconnected system with simple elements where sophisticated behaviour trickle up is more robust than sparse connections with smart elements.
  • Random encounters are useful. Random interactions without predefined orders between simple elements allow them to gauge and alter the macrostate of the system by their mere numbers.
  • Pay attention to neighbors. Local information leads to global wisdom. Neighboring elements stumbling across one another and interchanging data generates overall knowledge.
  • Note patterns in the signs. Simple pattern detection allows meta- information, or signs about signs, to circulate and generate a global behaviour of the system.

Source: Skyttner Lars (2006), General Systems Theory: Problems, Perspectives, Practice, Wspc, 2nd Edition.

Leave a Reply

Your email address will not be published. Required fields are marked *