Christopher Pincock - Explaining the success of models

Christopher Pincock (Ohio State University)

The Center for Philosophy of Science at the University of Pittsburgh invites you to join us for our Lunch Time Talk by a Center former fellow. Attend in person at 1117 Cathedral of Learning or visit our live stream on YouTube at https://www.youtube.com/channel/UCrRp47ZMXD7NXO3a9Gyh2sg.

LTT: Christopher Pincock

Tuesday, February 18th @ 12:00 pm - 1:30 pm EDT

Title: Explaining the success of models

Abstract:

The traditional defense of scientific realism uses inference to the best explanation (IBE) to infer some claims about unobservable entities from the predictive success of some theory. In this talk I consider how best to adapt this strategy to deal with the problem of idealized models. In most cases of predictive success, a theory is used to specify a model and only the model affords a prediction that can be checked. One problem is that models are typically also specified using non-theoretical claims, including idealizations that the users of the model believe are false. So, in what circumstances, if any, can a modeling success provide evidence that the theoretical claims are true? I argue that the realist should reconstruct this reasoning in two steps. First, a range of predictive successes for a model can support the conclusion that the model stands in what I call a robust representational relationship to its intended target. This conclusion is obtained through the use of IBE: the best explanation of the predictive successes of the model is that this kind of representational relation obtains. Second, given that the model is specified using some theoretical claims, the best explanation for why the representational relation obtains may be that these theoretical claims are true. When this additional condition is met, I argue that the modeling success can provide evidence for the truth of those claims. I conclude by engaging with two objections to this revised defense of realism. First, how should we generate and evaluate potential explanations that involve models and their relations to some target? Second, how can the best explanation of some modeling success involve the truth of some theoretical claims when that modeling success also involves idealizations that are believed to be false?

To follow along via Zoom, use this link:  https://pitt.zoom.us/j/93034356038