Models of real-world systems are widely used in science. It is often suggested that these models are tested or confirmed when their results are compared with observational data. I contend that this way of thinking is misguided; what we can sensibly aim to test or confirm via such comparisons are not scientific models themselves, but rather their adequacy for particular purposes. I then argue that testing a model’s adequacy-for-purpose involves challenges beyond those faced when testing whether a model embodies a true hypothesis about the workings of a target system, and I illustrate with some examples, including the case of climate modeling. Finally, I offer some exploratory remarks on how the notion of adequacy-for-purpose could figure in our understanding of the aims of science more generally.

Speaker: Wendy Parker, Ohio State University

http://events.stanford.edu/events/322/32205/