Title: Predictivism and Model Selection
By Alireza Fathollahi (Princeton, Philosophy)
Date: Tuesday March 29, 2022
Zoom: go to www.phil.bilkent.edu.tr
Abstract: There has been an intense debate in the philosophy of science over predictivism – that prediction has epistemic advantage over mere accommodation of data. Despite predictivism’s strong intuitive appeal, its opponents argue that whether the data was used in the process by which the hypothesis was designed belongs to the context of the discovery of the hypothesis and is irrelevant to how well-supported the hypothesis is. Using statistical results in the ‘model selection’ literature, I argue that the amount of support a hypothesis, H, receives from a body of data, D, is inversely related to the number of adjustable parameters of the family of hypotheses (model) from which H was selected. Crucially, when D is not essential to the design of H (i.e., when it is predicted), the model to which H belongs has fewer adjustable parameters than when D is essential to the design of H (i.e., when it is accommodated). This, I argue, gives us an argument for the following (very) strong version of predictivism: predicting a set of data provides significantly stronger support for a given hypothesis than merely accommodating it, even if we are fully aware of the contents of the hypothesis and the data and the method by which the theorizer has constructed it.
About the speaker: He got his BSc in physics from Sharif University of Technology. He then earned his MA in philosophy from University of Houston and his PhD in philosophy from Princeton. His research is on philosophy of science and early modern philosophy (especially Leibniz and Newton).