The epistemic value of independent lies: false analogies and equivocations

Margherita Harris

doi:10.1007/s11229-021-03434-8

Abstract

Here I critically assess an argument put forward by Kuorikoski et al. (Br J Philos Sci, 61(3):541–567, 2010) for the epistemic import of model-based robustness analysis. I show that this argument is not sound since the sort of probabilistic independence on which it relies is unfeasible. By revising the notion of probabilistic independence imposed on the models’ results, I introduce a prima-facie more plausible argument. However, despite this prima-facie plausibility, I show that even this new argument is unsound in most if not all cases of model-based robustness analysis. This I do to demonstrate that the epistemic import of model-based robust analysis cannot be satisfactorily defended on the basis of probabilistic independence.

Highlights

Any model of a real world phenomenon is bound to include idealizations of some sort
By revising the notion of probabilistic independence imposed on the models’ results, I introduce a prima-facie more plausible argument. Despite this prima-facie plausibility, I show that even this new argument is unsound in most if not all cases of model-based robustness analysis. This I do to demonstrate that the epistemic import of model-based robust analysis cannot be satisfactorily defended on the basis of probabilistic independence
A question arises: why can we use models to learn about the world despite their idealizing assumptions? If no model is ever a complete and veridical representation of its target system, why do we think of them as ‘vehicles for learning about the world’ (Frigg and Hartmann 2020)?

Summary

Introduction

Any model of a real world phenomenon is bound to include idealizations of some sort (by disregarding some variables, or ignoring or simplifying interactions amongst variables, etc.). The aim of this paper is to critically assess an argument put forward by Kuorikoski et al (2010) for the epistemic import of model-based robustness analysis. This assessment is important for two reasons. I believe Kuorikoski et al.’s argument is a formal expression of a widely held, but what I believe to be an misleading intuition This intuition is the following: a model’s conclusion is more likely to hold in the target system if several models lead to that conclusion because it would be a remarkable coincidence if that were not the case. I hope to show more forcefully that the assumptions that underscore Kuorikoski et al.’s argument are untenable

Setting the scene

An argument from coincidence?

Conclusion