--- layout: default --- Publication details Sensor Evolution in Artificial Systems: Towards a more appropriate model of the relationship between organism and environment Tim Taylor 2003 Abstract An approach to modelling open-ended evolutionary systems is discussed, in which an attempt is made to soften the distinction between organism and environment. Most existing artificial evolutionary systems impose a fairly inflexible structure upon organisms; at the same time, they generally pay little attention to the dynamics of the environment. As a result, the "problem" of sensor evolution arises: organisms are unable to evolve new ways of interacting with the environment, beyond those mechanisms provided by the model’s designer. An alternative model is discussed, which attempts to overcome this limitation and thereby increase evolvability. This comprises a cellular automata array (i.e. a simple dynamical system) as a model of the environment. A population of genomes interacts with this environment by specifying constraints upon its dynamics. An organism’s body (phenotype) is the set of environmental dynamics that are initiated by its genome. In this way, there is no representational distinction between phenotypes and the rest of the environment. As a model of the evolution of life, this approach has a number of advantages over previous work, for example: with no "hard-wired" specification of what form an organism’s body should take, this structure is free to evolve; because the environment is a dynamical system with certain self-organisational properties, chain reactions, etc., an organism’s genome can evolve to exploit almost any process supported by the environment (thereby avoiding many of the problems faced by other approaches in evolving news ways of interacting with the environment, e.g. sensor evolution); because organisms and the environment are represented as a single system, organisms form part of the environment experienced by other organisms, thus creating rich co-evolutionary dynamics; and, while retaining the same basic model, it is very easy to incrementally add more physical realism to the dynamics of the environment, such as entropy increase, conservation of matter, energy flow, membrane structures, etc., as well as informational properties such as computational universality. The relevance of this model to designing other artificial evolutionary systems, in software and in hardware, will be discussed. Full text Presentation slides: pdf Reference Taylor, T. (2003). Sensor Evolution in Artificial Systems: Towards a more appropriate model of the relationship between organism and environment. In J. F. Miller, D. Polani, & C. L. Nehaniv (Eds.), Abstracts from the Evolvability and Sensor Evolution Symposium, held at University of Birmingham, UK, in April 2003. University of Hertfordshire Computer Science Technical Report No. 384. BibTeX @incollection{taylor2003sensor, author = {Taylor, Tim}, title = {Sensor Evolution in Artificial Systems: Towards a more appropriate model of the relationship between organism and environment}, booktitle = {Abstracts from the Evolvability and Sensor Evolution Symposium, held at University of Birmingham, UK, in April 2003}, publisher = {University of Hertfordshire Computer Science Technical Report No.~384}, year = {2003}, month = apr, editor = {Miller, Julian F. and Polani, Daniel and Nehaniv, Chrystopher L.}, category = {workshop}, keywords = {evoca, meaning} } Related publications
  1. Taylor, T. (2009). A Creative Dance: Symbols, Action and the Bringing Forth of Meaning. In M. Boden, M. D’Inverno, & J. McCormack (Eds.), Computational Creativity: An Interdisciplinary Approach. Retrieved from http://drops.dagstuhl.de/opus/volltexte/2009/2207
    PDF Full details
  2. Taylor, T. (2004). Redrawing the Boundary between Organism and Environment. In J. Pollack, M. A. Bedau, P. Husbands, R. A. Watson, & T. Ikegami (Eds.), Artificial Life IX: Proceedings of the Ninth International Conference on the Simulation and Synthesis of Living Systems (pp. 268–273). https://doi.org/10.7551/mitpress/1429.003.0045
    PDF Full details
  3. Taylor, T. (2003). Evolving Interaction in Artificial Systems: An historical overview and future directions. In P. McOwan, K. Dautenhahn, & C. L. Nehaniv (Eds.), Abstracts from the Evolvability and Interaction Symposium, held at Queen Mary, University of London, UK, in October 2003. University of Hertfordshire Computer Science Technical Report No. 393.
    Full details
  4. Taylor, T. (2002). An Alternative Approach to the Synthesis of Life. Poster presented at the 8th International Conference on the Simulation and Synthesis of Living Systems (ALIFE 8), Sydney, Australia.
    Full details
  5. Taylor, T. (2002). The Control of Dynamical Systems by Evolved Constraints: A New Perspective on Modelling Life (Informatics Research Report No. EDI-INF-RR-0148). University of Edinburgh.
    PDF Full details
« Return to publications list