ONLINE: PhD Student Session: Margherita Harris and Dmitry Ananyev

In science, obtaining the same result through different means (i.e. obtaining a ‘robust’ result) is often seen as a valid way to further confirm that result. The Bayesian should of course have something to say about  the logic underpinning this method of confirmation. But, as Schupbach (2018) persuasively argues, Bayesian accounts of robustness analysis (RA) which rely on probabilistic independence to explicate the notion of RA diversity are in many cases woefully inadequate.  Given this, it seems evident that in order to capture those cases we must depart from independence-based accounts of RA diversity. Schupbach’s explanatory account is arguably a promising  step in the right direction. Indeed, by having ‘as its central notions explanation and elimination’, this account seems to fit very nicely with many empirically driven cases of RA in science, while at the same time providing important normative implications.  In this talk, however, I will assess Schupbach’s further claim that his account of RA ‘applies to model-based RAs just as well as it does to empirically driven RAs’, since when we arrive at this claim, he and I decisively part ways. And Schupbach is not alone in  making this claim. Winsberg (2018), for instance, has also argued that Shupbach’s account can successfully be applied to climate model-based RAs. But in this talk, I will argue, contrary to both Schupbach and Winsberg, that this explanatory account of RA cannot  be applied to model-based RAs in the way they suggest. Finally, I will comment on what lessons we might be able to learn from this fact, lessons about the viability of model-based robustness analysis as a method of confirmation.

I will discuss the possibility of cases in which several agents performing acts of a certain type cause some morally significant outcome, but no individual act of this type seems to make a morally significant  difference to the outcome in question (e.g. Parfit’s (1984) “Harlmess Torturers” case). My aim is to argue that cases in which a single act of the relevant type never makes a morally significant difference while many such acts do are not possible. I proceed  in three steps. First, following other authors, I show that the cases in question generate a phenomenal sorites series. If they cannot avoid the absurd implications of the paradox, then there is a strong reason to think that they are impossible. Second, drawing on empirical work by Diana Raffman (2012), I suggest that two claims in conjunction explain how the paradox is avoided and why it is intuitive to think that the cases in question are possible while they are not. The first claim is that no two adjacent objects  in a sorites series can be categorised differently if considered pairwise and the second is that individual objects in a phenomenal sorites series may change the way they appear to or are experienced by an observer. Third, I show that the strongest argument  in defence of the cases rests on a rejection of the second claim and is for this reason unsuccessful.