The Argument from Underconsideration and Relative Realism

Abstract

In this article, through a critical examination of K. Brad Wray's version of the argument from underconsideration against scientific realism, I articulate a modest version of scientific realism. This modest realist position, which I call ‘relative realism’, preserves the scientific realist's optimism about science's ability to get closer to the truth while, at the same time, taking on board the antirealist's premise that theory evaluation is comparative, and thus that there are no good reasons to think that science's best theories are close to the truth.

Acknowledgements

An earlier version of this paper was presented at the 23rd Biennial Meeting of the Philosophy of Science Association, San Diego, 15–17 November 2012. I would like to thank the PSA and the History of Science Society for a generous travel grant. I would also like to thank Paul Hoyningen-Huene and P. Kyle Stanford for their useful feedback on my talk. I am very grateful to K. Brad Wray and two anonymous referees for International Studies in the Philosophy of Science for helpful comments on earlier drafts. Finally, special thanks are due to the editor, James McAllister.

Notes

1 See also Boyd (1983) and Psillos (1999). Since the main argument for (E) is an argument known as the ‘no-miracles argument’, which is an instance of inference to the best explanation (IBE), any version of scientific realism that endorses (E) is sometimes referred to as ‘IBE realism’. See, e.g. Psillos (1999) and Doppelt (2002). For critiques of the no-miracles argument, see Frost-Arnold (2010) and Mizrahi (2012).

2 As is well known, explicating the notion of approximate truth is notoriously difficult. Popper's (1972) attempt to formalize the notion of approximate truth or verisimilitude was shown to be problematic (Miller 1974; Tichý 1974). Other formal approaches, such as the similarity approach (e.g. Niiniluoto 1987, 1998) and the type hierarchy approach (e.g. Aronson, Harré, and Way 1994) also suffer from technical problems (e.g. Aronson 1990 and Psillos 1999, respectively). For these reasons, realists have tried to explicate approximate truth in non-formal, qualitative terms (e.g. Leplin 1981; Boyd 1990; Weston 1992; Smith 1998). For example, it has been suggested that T₂ is more approximately true than its predecessor T₁ if T₁ can be described as a ‘limiting case’ of T₂ (e.g. Post 1971; French and Kamminga 1993). But there are problems with these informal approaches as well (Chakravartty 2010).

3 My attempt to articulate a modest realist position is motivated by my work on the pessimistic induction and the no-miracles argument. As I show in Mizrahi (2012, 2013), both arguments fail. If these two arguments are the best arguments on both sides of the realism–antirealism debate, then realism and antirealism lack adequate support.

4 Rescher once used the label ‘relativistic realism’ to refer to

a realism that is relativistic in that its insistence on the multi-faceted nature of the real means that any science will reflect its deviser's particular ‘slant’ on reality (in line with the investigator-characteristic modes of interaction with nature). On such a view, knowledge of reality is always (in some crucial respect) cast in terms of reference that reflect its possessor's cognitive proceedings. There is, no doubt, a mind-independent reality, but cognitive access to it is always mind-conditioned. (Rescher 1990, 102)

5 Kuipers (2009) defends a view that he calls ‘comparative realism’ (CR), whose main claim is that ‘truth approximation provides the default explanation and prediction of empirical and aesthetic progress’. My own RR shares CR's main epistemic claim that we are only justified in making comparative judgements about the truth (or, as I call it, ‘comparative truth’) of competing theories, not absolute judgements. As I see it, the main difference between CR and RR lies in the way they are argued for. Kuipers's argument for CR is essentially an IBE, whereas my argument for RR is an argument from the nature of theory evaluation itself, since I do not think that one can infer realism from explanatory and/or aesthetic considerations (Mizrahi 2012). In other words, Kuipers's CR is wedded to IBE in a way that my RR is not.

6 There is some debate as to whether, and to what extent, van Fraassen's constructive empiricism is informed by his voluntarist epistemology (e.g. Psillos 2007; cf. Dicken 2009).