Generalizations and Models in Ecology: Lawlikeness, Invariance, Stability, and Robustness

Research output: ThesisDoctoral ThesisMonograph

Abstract

The question at issue in this dissertation is the epistemic role played by ecological generalizations and models. I investigate and analyze such properties of generalizations as lawlikeness, invariance, and stability, and I ask which of these properties are relevant in the context of scientific explanations. I will claim that there are generalizable and reliable causal explanations in ecology by generalizations, which are invariant and stable. An invariant generalization continues to hold or be valid under a special change – called an intervention – that changes the value of its variables. Whether a generalization remains invariant during its interventions is the criterion that determines whether it is explanatory. A generalization can be invariant and explanatory regardless of its lawlike status. Stability deals with a generality that has to do with holding of a generalization in possible background conditions. The more stable a generalization, the less dependent it is on background conditions to remain true. Although it is invariance – rather than stability – of generalizations that furnishes us with explanatory generalizations, there is an important function that stability has in this context of explanations, namely, stability furnishes us with extrapolability and reliability of scientific explanations. I also discuss non-empirical investigations of models that I call robustness and sensitivity analyses. I call sensitivity analyses investigations in which one model is studied with regard to its stability conditions by making changes and variations to the values of the model’s parameters. As a general definition of robustness analyses I propose investigations of variations in modeling assumptions of different models of the same phenomenon in which the focus is on whether they produce similar or convergent results or not. Robustness and sensitivity analyses are powerful tools for studying the conditions and assumptions where models break down – and they are especially powerful in pointing out reasons as to why they do this. They show which conditions or assumptions the results of models depend on.
Original languageEnglish
Place of PublicationHelsinki
Edition1
Publisher
Print ISBNs978-952-10-6767-9
Electronic ISBNs978-952-10-6768-6
Publication statusPublished - 2 Mar 2011
MoE publication typeG4 Doctoral dissertation (monograph)

Fields of Science

  • 611 Philosophy

Cite this