How to Do Things with Theory: The Instrumental Role of Auxiliary Hypotheses in Testing

Abstract

Pierre Duhem's influential argument for holism relies on a view of the role that background theory plays in testing: according to this still common account of "auxiliary hypotheses," elements of background theory serve as truth-apt premises in arguments for or against a hypothesis. I argue that this view is mistaken. Rather than serving as truth-apt premises in arguments, auxiliary hypotheses are employed as (reliability-apt) "epistemic tools": instruments that perform specific tasks in connecting our theoretical questions with the world but that are not (or not usually) premises in arguments. On the resulting picture, the acceptability of an auxiliary hypothesis depends not on its truth but on contextual factors such as the task or purpose it is put to and the other tools employed alongside it. [I]n saying these words, we are doing something ... rather than reporting something. Austin (1962, 13) 0 Introduction Pierre Duhem's incredibly influential The Aim and Structure of Physical Theory (1914/1951) is usually and rightly remembered for its holism. Central to the argument for holism-and perhaps more influential than the holist conclusion itself-is a view of the role that background theory plays in testing. For Duhem, theories or models other than the hypothesis that we aim to test serve primarily as premises in the derivation of consequences. These consequences are then compared to the world; since all the premises have the same logical relationship to the consequence, either all of the premises pass as a conjunction, or they fail as the same. Hence holism. Call these background theories and models "auxiliary hypotheses." The central thesis of this paper is that the Duhemian view of the role that auxiliary hypotheses play in testing is mistaken. The theories, models, and analogies employed in testing a hypothesis rarely serve as premises. Instead, they serve as "epistemic tools": instruments that perform specific tasks in connecting our theoretical questions with the world but that are not (or not usually) premises in arguments. For the purposes of testing, therefore, it's largely irrelevant whether these theories or models represent truthfully. What we care about is whether they reliably perform their task.

Robustness and the Event Horizon Telescope: the case of the first image of M87*

“…Additionally, Dethier (2020Dethier ( , 2022 explicitly argues for a unified account of robustness across experimental and modeling contexts. This seems to be based on a view of models that treats them as 'epistemic tools' (Dethier 2021) and centers their reliability or adequacy-for-purpose over their 'truth' (see also, e.g., Morgan and Morrison (1999) and Parker (2020)). On this view, robustness across varied models can provide confirmation because model reports (like experimental results) are evidence whose production is apt to be appropriately varied.…”

Calibrating statistical tools: Improving the measure of Humanity's influence on the climate

“…as a control in an experiment testing the effects of some intervention. In these cases, as I put it in Dethier (2021), the model is less like a truth-apt proposition and more like a reliability-apt tool: the appropriate question is whether it reliably or adequately performs the role that it has been tasked with. (Or, better: whether it performs its role in a way that renders the relevant procedure as a whole reliable.)…”

“…20 In order to fulfil their function, therefore, the statistical models employed in attribution studies must accurately represent these likelihoods: a model that accurately represents 19 See the work of Bokulich and Parker in particular (e.g. Bokulich 2021;Bokulich and Parker 2021;Parker 2009Parker , 2020b as well as Currie (2017) and Dethier (2021). 20 There are sometimes other factors in both the classical and Bayesian context.…”

“…In measurement, the goal is to use the data to fill in or update the theoretical understanding; a successful measurement procedure relies on having good (enough) instruments and data, good (enough) prior theoretical understanding to determine what the data are telling us, and good (enough) statistical models to carry out the inference. On artefactual views, the quality of each of these elements cannot properly be judged separately from the others: whether the statistical model is "adequate" for the particular measurement procedure depends on the nature of the instruments and theory that are actually in place (compare Parker 2020b, 463;Dethier 2021Dethier , 1462. So while there's a conceptual distinction between these three elements, they are deeply interconnected both causally-our theoretical understanding partly determines which statistical models we use-and for evaluative purposes.…”

Climate Models and the Irrelevance of Chaos

“…No, though defending this point adequately here would take us too far afield. The central idea is that it's a mistake to read Q as a claim about the truth of the idealized model rather than as a claim about its (non-literal) accuracy(Frigg and Nguyen 2019), adequacy-for-purpose(Parker 2009), or reliability(Dethier 2019). 9 I'm borrowing this example from G. E Smith (2002)…”