Advertisement

Collective Intelligence in Natural and Artificial Systems

  • Jean-Louis Deneubourg
Conference paper

Abstract

Insect societies are described as “factories within fortresses.” The society is a well defended factory which gathers food (raw material) from outside world, and uses them to maintain and extent its nest (infrastructure) and produces new individuals. The problems and solutions (transportation, development of network, task allocation, synchronisation,...) of insect societies exhibit great analogies with “human factories”. However despite these similarities, the precepts and the rules of the organisation involved in insect societies and artificial systems are very different.

In artificial systems, interactions between units are, for the most part, weak and/or strongly centralized (hierarchy). In insect societies, the interactions are numerous, very often based on positive feed-back and highly decentralized. A second difference appears when we compare the “knowledge” of the units. In artificial systems, when we need to manage a large group of units, the first idea is often to use a centralized and hierarchical chain of command. For exemple in the case of transportation, a “chief’ knows at any moment the position, the goals,... of each unit. Based on this information, the centre issues its orders. This idea is pushed to the extreme with the advent of computers, leading to the dream of organization based on perfect knowledge.

What the insects have selected and what we explore in many respect is diametrically opposed. Rather than one solitary central control unit that is specific, complex and omniscient, society is self-organized through a multitude of local interactions between units but also with their environment. For instance, insects are locally informed (no global knowledge), interact with all others units (no hierarchy), and these interactions are not explicitly coded. When transposing this model to artificial systems, this eliminates the central organizer, exploiting real time interactions between agents.

These observations raises two key questions:
  • What level of complexity at the individual level is required to generate the observed complexity at the group level?

  • How much of the observed complexity at the group level is a reflection of complexity of the environment rather than complexity at the level of the individual?

In this context, research in biological and artificial systems (related to building, foraging, transportation,...) focuses on the type of information used by agents to provide a good statistical sampling, the link between the local treatment of the information by the agents and the modulation of the communications, and the different procedures of transmission of the information inside the group. Theoretical and experimental result demonstrate that collective choice can appear very easily even without any coding of information by individual ants and that these problems must be considered not only at the level of the algorithms used by the agents, but also at the level of the physical or chemical characteristics of the communications and of the environment.

Keywords

Artificial System Task Allocation Global Knowledge Perfect Knowledge Collective Choice 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

References

  1. [1]
    E. Bonabeau, G. Theraulaz. Intelligence Collective, Hermès, 1995.Google Scholar
  2. [2]
    R. Beckers, J.L. Deneubourg, S. Goss, Trails and U-turns in the selection of a path by the ant Lasius niger, Journal of Theoretical Biology, Vol. 159, 1992, pp. 397–415.CrossRefGoogle Scholar
  3. [3]
    R. Beckers, O.E. Holland, J.L. Deneubourg, From local actions to global tasks: stigmergy and collective robotics. Proceedings of ALIFE IV, Eds R.A. Brooks, P. Maes, MIT Press, Cambridge, Massachusetts, 1994, pp. 181–189.Google Scholar
  4. [4]
    S. Camazine, J. Sneyd, A model of collective nectar source selection by honeybees: self-organization through simple individual rules, Journal of theoretical Biology, Vol. 147, 1991, pp. 553–571.CrossRefGoogle Scholar
  5. [5]
    J.L. Deneubourg, N. Franks, Collective control without explicit coding: the case of communal nest excavation, Journal of Insect Behavior, Vol. 8, 1995, pp. 417–432.CrossRefGoogle Scholar
  6. [6]
    J.L. Deneubourg, S. Goss, N. Franks, A. Sendova-Franks, C. Detrain, L. Chrétien, The dynamics of collective sorting robot-like ants and ant-like robots, From Animals to Animats, Eds. J.-A. Meyer, S. Wilson, MIT Press, Cambridge, Massachusetts, 1991, pp. 356–365.Google Scholar
  7. [7]
    Cl. Detrain, J.L. Deneubourg, Scavenging by Pheidole pallidula: a key for understanding decision-making systems in ants, 1996. (in press)Google Scholar
  8. [8]
    S. Goss, S. Aron, J.L. Deneubourg, J.M. Pasteels, Self-organized shortcuts in the argentine ant, Naturwissenschaften, Vol. 76, 1989, pp. 579–581.CrossRefGoogle Scholar

Copyright information

© Springer Japan 1996

Authors and Affiliations

  • Jean-Louis Deneubourg
    • 1
  1. 1.Centre for Non-Linear Phenomena and Complex SystemUniv. Libre de BruxellesBelgium

Personalised recommendations