ICCS2004 Application Submission/Review

Abstract for
"A new type of models for managed ecosystems"


Using the encoding-decoding metaphor of Robert Rosen, a model in the natural sciences can be characterized as a relationship between a natural and a formal system. This modeling relation with its notion of states from which behaviour can be deduced is the conventional paradigm in dynamic systems theory. The encoding operation from the natural to the formal system requires an intersubjective (among scientists) agreement about observations that can be encoded as states of a formal system. However, the process of encoding faces serious problems when applied to living systems as already pointed out by R. Rosen. Some of the states act as memory and are unobservable for any external agent (at least in all practical situations). In ecological modeling, these models have not been able yet to produce a single counter-intuitive logic entailment (in the formal system) that could be confirmed after decoding into the corresponding natural system; in other words, no successful prediction of ecosystem behavior has been made for this type of models. Most models in ecosystem research are overparameterised and calibrated to observations, thus they are used effectively to interprete and evaluate observations rather than explain and predict them. We propose a new type of models for managed ecosystems which is complementary to the one of Robert Rosen and overcomes the above problems. It is based on a notion of interactive computing that has recently been developed in theoretical computer science. Using sustainable forestry as paradigmatic example, we relate “real” behaviour (encountered in interactive management decisions in silviculture) with “virtual” behaviour in which the same interactive tasks are posed through a model interface. The encoding operation from “real” to virtual behaviour requires an intersubjective (among forestry experts) agreement about aspects within their memories that can be encoded as behavioral patterns of a virtual system. We argue that this process of encoding of the heuristics of experts is the appropriate approach when dealing with interactive processes such as life. Implications for ecosystem modeling are discussed with respect to forest growth models.