Colloquium
Where the truth lies: evaluating the scientific content of multiscale models in the physical sciences and engineering
Jeffrey Picka (University of New Brunswick)
Tuesday, April 15, 2014 at 3:30 pm
Tilley Hall Room 404
When modelling the climate, a flowing powder, or the emergence of unexpected patterns arising from complex chemical reaction dynamics, it is necessary to construct the model on a much smaller scale than the scale of the phenomenon of interest. For the model to be useful, a representation of the phenomenon of interest must emerge from the interactions of many strongly interacting elements on the smaller scale. Even if computer code can seem to represent this emergence, it is often an open question as to whether or not the code is consistent with the theoretical arguments that inspired it. If the model is consistent with theory, then it may be useful in predicting the future behaviour of the phenomenon and in managing it. If the model is not consistent with the theory, then the model is speculative and may have some scientific use but may be greatly misleading when used in management or planning. To understand the limitations on the scientific use of multiscale computer models of this type, it is necessary to use methods from the philosophy of physics to investigate the truth-content of claims that can be made from these models. This analysis, combined with an examination of the technical aspects of the models themselves, reveals that scientific information can only be created from these models if the models represent disorder, dynamic unpredictability, and model uncertainty separately from each other. The analysis also reveals that the ability of these models to represent Nature can only be assessed by statistical inference, and only under limited conditions. Some implications of these limitations for physics, for engineering, for research funding, and for the use of these models in making policy will be discussed. For scientists, this talk will emphasize the pragmatic use of applying philosophy to create better scientific understanding. For non-scientists, mathematical and technical details will be downplayed in favour of a discussion of the philosophical aspects of model assessment.
Jeffrey Picka is an Associate Professor in the Department of Mathematics and Statistics, University of New Brunswick, Fredericton.
