Non‐competitor Conditions in the Scientific Realism Debate

Abstract

A general insight of 20th‐century philosophy of science is that the acceptance of a scientific theory is grounded, not merely on a theory’s relation to data, but on its status as having no, or being superior to its, competitors. I explore the ways in which scientific realists might be thought to utilise this insight, have in fact utilised it, and can legitimately utilise it. In more detail, I point out that, barring a natural but mistaken characterisation of scientific realism, traditional realism has not utilised that insight regarding scientific theories, i.e., has not explicitly factored that insight into, and invoked it as justification for, what realists believe. Nonetheless, a new form of realism has. In response to a key historical threat, two of the most thoroughly developed contemporary versions of realism—one put forward by Jarrett Leplin, another by Stathis Psillos—are anchored on the sensible tactic of requiring that the theories to which realists commit themselves have no competitors. I argue, however, that the particular kind of non‐competitor condition they invoke is illegitimate in the context of the realism debate. I contend further that invoking a non‐competitor condition that is legitimate, sensible, and even, as it turns out, required in the context of the debate threatens to eliminate the possibility of scientific realism altogether.

Acknowledgements

For thoughtful commentary on earlier drafts of this paper I am indebted to David Harker, two anonymous referees, and the editor of this journal, James W. McAllister. For conversations and/or correspondence regarding various topics that pertain to this paper, I am grateful to Howard Sankey, Neil Thomason, Brian Ellis, John Worrall, Alan Musgrave, John Tilley, Gerald Doppelt, David Papineau, Peter Lipton, Stephen Ames, and Kristian Camilleri.

Notes

[1] It is important to remain mindful that this point is being made in an attempt to locate how, if at all, scientific realists utilise the insight regarding competing scientific theories. Notice that nothing here bars the realist from believing meta‐hypotheses that attribute truth simpliciter to certain parts of T (and which, at the same time, refrain from attributing truth to other parts of T): as a consequence of believing a meta‐hypothesis, realists may very well believe that some parts of T are true simpliciter. (Of course, to claim justification for believing those parts, realists must specify just which parts do and do not so qualify, and I will address a sophisticated proposal of a similar kind below.) Recognising this does nothing to hinder my claim that realists do not believe T per se. And, as we will see, the latter point brings clarification on how, if at all, the competition insight is utilised by (standard) scientific realism.

[2] While embracing the no‐miracles argument may ultimately entail holding that any Ts enjoying novel success can have no genuine competitors that also enjoy novel success, that success, in itself, does not pertain to competition.

[3] As should be clear here, at no stage will I be denying that it is legitimate for realists to employ a non‐competitor condition. I fully agree that they should and, for reasons just noted, even that they must.

[4] Notice furthermore that modifying the realist meta‐hypothesis to pertain to ‘most successful theories’ conflicts with the realist arguments themselves, conceding as it does to successes that are miraculous, or at least, for which there are better explanations. Moreover, the list of counterinstances directly challenges the common realist claim that the methods of science (such as inference to the best explanation) are reliable guides to truth. It is clear then that invoking the premise, ‘the methods of science are reliable guides to truth’, in any defence against the historical argument requires that realism be granted victory in advance.

[5] In accord with note 1, these points hold irrespective of whether some meta‐hypothesis (regarding say approximate truth) might as a consequence license belief in (even the truth of) some specific parts of theories.

[6] In the post‐Kuhnian era, we have come (or at least begun) to recognise as skewed the historical perspective offered by science textbooks, as well as many texts dedicated to the history of science. And, when it comes to the question of whether there were competitors at t, we cannot justifiably ignore our recognised tendency toward presentism or whiggism.

[7] Also, and crucially, noting here the extraordinary difficulty in identifying even one instance of a past theory that had no competitors at t, we see that a realist meta‐hypothesis containing such a requirement is rendered practically if not wholly immunised from the possibility of refutation. As a result, the legitimacy of such a meta‐hypothesis as a response to the historical challenge is dramatically diminished, if not negated.

[8] In the next section I will address the proposition that realist’s can justifiably deny a competitor’s (approximate) truth by appealing to criteria such as non–ad hocness.

[9] For Leplin, the alternatives need not, themselves, meet all the strict details of his independence condition (Leplin 1997, 75). So I am here considering a demand stronger than any Leplin imposes.

[10] Since Psillos also packs novelty into his definition of success, we now see something we might otherwise fail to notice, that, in Psillos’s desire to protect realism from the historical threat, he has packed a demand for novelty into both correlates of his realist hypothesis.

[11] It is important to recognise that the issue of concern here is not whether standard novelty can be employed to define T’s success in the realist’s meta‐hypothesis and no‐miracles argument. (I am not denying here that it can.) What is at issue here is quite distinct: whether, in the context of the realism debate, novelty can be legitimately invoked to eliminate alternative theories as genuine competitors (And I will contend that it cannot.)

[12] Put another way, we might construe the realist as asking, under what conditions would it be reasonable to believe or infer a theory to be approximately true? (I’m indebted to David Harker for suggesting this sort of phrasing.) We see, however, that, in asking this, realists are presupposing that such a belief or inference is reasonable and, hence, requiring that victory against non‐realism has been achieved prior to addressing the non‐realist’s challenge.

[13] From the standpoint of the distinction made in Section 1 and revisited in this paragraph, the ‘argument from underconsideration’ (Lipton 2004) can be seen (at least in its explicit form) as challenging a realism grounded on eliminative inferences at the level of scientific theories. Wray (2008) offers a recent defence of that non‐realist argument that also draws briefly on the historical threat.