Naturalism, Functionalism and Chance: Not a Best Fit for the Humean

Abstract

How should we give accounts of scientific modal relations, such as laws and chances? According the Humean, we should do so by reducing these relations to parts of non-modal actuality: typically, patterns in actual events, where the relevant events do not metaphysically necessitate each other or ‘build in’ facts about modality. 2 Modal relations are nothing ‘over and above’ the non-modal. Here are three motivations for being Humean. Firstly, one might be worried about admitting ‘mysterious’ elements into one’s ontology (Loewer 2012, p. 121). Modal relations are strange. Humean accounts reduce modal relations to the non-modal. They provide a straightforward account of what modal relations are and their relation to the non-modal. If we take modal relations as primitive, by contrast, we seem saddled with strange entities, and have to explain their connection to actual events. Secondly, one might be motivated by a kind of functionalism. One may wish to account for modal relations by considering the role such relations play in our lives and scientific theorizing. Perhaps chance, for example, should be accounted for by identifying something that plays its role of guiding credences, and so forth. It might seem that Humean accounts are particularly well suited to meet this aim, since they can use the role of modal relations to specify what non-modal relations scientific modal relations reduce to. We’ll see some examples below. The association between Humeanism and functionalism has become so strong that even non-Humeans take it that Humeans are uniquely interested in showing why modal relations are fit to play their roles (Hall 2015), and that Humean explanations are of a kind that even non-Humeans should adopt (Ismael 2015). Recent Humean accounts have been particularly explicit in their functionalist motivations. While Lewis (1983, 1994) was largely content to appeal to broad criteria such as generality and simplicity in his analysis of laws, recent accounts (Hicks 2018, Dorst 2019, Jaag and Loew 2020) have focused on refining these criteria and justifying their relevance by arguing that we need laws to be simple, general, and satisfy other criteria, if they’re to be useful to us. If one has functionalist motivations, and Humeanism is required to meet those, one has strong reason to be Humean. Thirdly, one might be motivated towards Humeanism by naturalism of a kind of that envisages a close connection between science and metaphysics. The general thought is that metaphysics shouldn’t attempt to replace science or revise it in its image—‘[metaphysics] should wherever possible prefer scientific explanations over metaphysical postulation’ (Loewer 2012, p. 136). Instead, metaphysics is constrained by science and provides accounts of the kinds of relations and entities that science is concerned with. I’ll develop this idea further below. It may seem that, by being functionalist, Humean accounts can deliver the modal relations used in science—those the feature in scientific derivations and explanation. If so, Humeanism may seem like a good choice for those wanting a naturalistic metaphysics. The argument of this paper is that two of these motivations do not count in favour of Humeanism. If one is motivated by functionalism, one has no reason to adopt Humeanism over its rivals. If one is motivated by a naturalist connection between science and metaphysics, one has reason to reject Humeanism. Motivations of the first kind (finding modal entities strange) may still lead one to Humeanism. But for those more concerned with the function of modal relations and the fit between science and metaphysics, one should look elsewhere. To make this argument, I will focus on the case of chance: objective probabilities that apply in the single case. In Section 2, I discuss the positive claims made by Humeans: that Humean accounts fit well with science, and that only Humean chances can be shown to play the role of chance. In Section 3, I examine recent attempts to show Humean chances satisfy the chance role and show how they rely on indifference reasoning. In Section 4, I argue that, notwithstanding this concern, Humeans have no special advantage when it comes to showing that chances are fit to play the chance role. In Section 5 I consider whether the Humean can respond by advocating a revision of science—and argue this response fits ill with naturalism. In Section 6, I argue that there is a deeper tension: Humeanism implies a disunity between science and metaphysics of a kind that naturalists should reject. In Section 7 I offer a brief sketch of an alternative naturalist justification.