Three preliminary objections

A prima facie objection to an argument that proceeds from methodological naturalism to ontological naturalism would be that such a move seems patently question-begging: we rule out any consideration of supernatural activity as a starting assumption, and subsequently make this starting assumption the conclusion of our investigations. But the argument can be constructed in a way that could render it defensible against this critique. We might imagine it to be an inductive argument that takes naturalism to be a kind of provisional hypothesis, as Boudry et al. suggest, rather than a heuristic stance. That provisional assumption, while it may not be the primary hypothesis being tested in specific cases of scientific investigation, would nonetheless be a kind of accompanying hypothesis that in a sense is at issue in every scientific trial. A history of scientific successes in which this assumption was operative, then, might be thought to provide an indirect confirmation of its truth. Since the argument is an inductive one, a long history of scientific successes might be seen as constituting evidence in its favour.

A second possible objection concerns what counts as ‘success’ and what, precisely, can be inferred from it. There is no doubt that our best scientific theories have been remarkably successful, if we take success to refer to their capacity to provide a basis for retrodictions, novel predictions, and the enabling of working technologies. But is there a clear path from these successes to truths about how the world really is? If not, an argument from the success of science cannot help adjudicate the truth or otherwise of naturalism as an ontological commitment. Hilary Putnam famously argued that it would be miraculous if successful theories were not true (Putnam (Reference Putnam1975), 73; cf. Psillos (Reference Psillos1999), Lyons (Reference Lyons2003), Busch (Reference Busch2008) ). But the direct inference from success to truth faces equally well-known challenges. The history of science bears witness to countless discarded theories which in their day were successful, enabling novel predictions, the manipulation of nature, and design of specific technologies. Yet this succession of theories does not converge on a single picture of reality (Hesse (Reference Hesse1976), Laudan (Reference Laudan1981), Dear (Reference Dear2006), 1–8, Vickers (Reference Vickers2013); cf. Poincaré (Reference Poincaré1905), 161f.).Footnote ⁵ A straightforward inductive argument from the history of science would thus suggest that our present successful theories will similarly be discarded in time, along with their ontological furniture. Joining this ‘pessimistic induction (or metainduction)’ argument is the problem of underdetermination, according to which every theory has empirically equivalent alternatives that account for the same observations but posit different underlying realities (van Fraassen (Reference Van Fraassen1980), 46; Bonk (Reference Bonk2008); Stanford (Reference Stanford2002); Idem (2016) ). Van Fraassen (Reference Van Fraassen1980, 39–40; cf. Wray (Reference Wray2007), Idem (Reference Wray2010) ) has also proposed that there is nothing especially miraculous about the success of scientific theories, since from an evolutionary perspective only successful theories survive. They may do so by latching onto regularities in nature, but it does not follow that we have access to the ontological realities that underpin that regularity.

Defenders of realism have offered robust responses to each of these objections (Laudan (Reference Laudan and Savage1990); Psillos (Reference Psillos1999), chs 4–8; Fahrbach (Reference Fahrbach2011); Park (Reference Park2011); Mizrahi (Reference Mizrahi2013); Lyons (Reference Lyons2013); cf. Vickers (Reference Vickers2012) ), and there are, of course, different versions of realism and varying accounts of success. I will not argue for a definitive position on these issues in this article (although we will return to the issue of selective realism in the final section). Suffice it to say for now that the historical argument for naturalism based on the success of science would seem to require commitment to some version of scientific realism, a doctrine that is far from commanding universal acceptance.Footnote ⁶

A third misgiving about the historical argument for naturalism concerns the fact that a wide a range of commitments have been associated with the label ‘naturalism’, and versions of it are held by individuals with divergent and sometimes conflicting philosophical positions (Stich (Reference Stich1996), 197; Stroud (Reference Stroud1996) ). As van Fraassen (Reference Van Fraassen2002), Rea (Reference Rea2002, Reference Rea, Craig and Moreland2016), and Halvorson (Reference Halvorson and Clark2016) have pointed out, it follows that if ‘metaphysical naturalism’ cannot be identified with any precision then it is difficult to regard it as a unitary hypothesis for which we might gather supporting evidence. In the specific argument being critiqued here, however, the ‘naturalism’ under discussion seems to involve a simple opposition between ‘naturalism’ and ‘supernaturalism’. For the purposes of the present exercise, then, we can simply stipulate that ‘naturalism’ in the relevant sense means a rejection of the existence of supernatural or spiritual entities.

These objections, and the second in particular, already raise significant difficulties for the historical argument for naturalism. In each case, moreover, we encounter something of an equivocation between naturalism as a successful explanatory strategy and naturalism as an ontological commitment. But assuming for now that the argument can surmount these preliminary hurdles, what can be said about the more general witness of the history of science to which the argument appeals?

Natural versus supernatural?

Boudry et al. (Reference Boudry, Blancke and Braeckman2010) suggest that when we survey the history of science, what we see is ‘the consistent success of naturalistic explanations and the lack of success of supernaturalistic explanations’ (ibid., 227). This argument not only takes it for granted that there is an exclusive disjunction between naturalistic and supernaturalistic explanations, but also assumes that it makes sense to apply this dichotomy to the past. However, when we look to the history of science it quickly becomes apparent that conceptions of how natural and supernatural were related is much more complicated than this.

The natural science of Aristotle, for example, enjoyed a tenure that lasted into the seventeenth century (albeit with some modifications). As is well known, the Greek philosopher's explanations of change in nature invoke four causes: material, formal, efficient, and final. In this scheme of things, God is involved at the level of final cause. As the unmoved mover, God is the ultimate source of all motion in the cosmos, since he is the end to which all things are naturally directed (Metaphysics 1072a26–27; De caelo 279a20–30). Is Aristotle's explanation of motion naturalistic or supernaturalistic? The simple answer is that the question is misconceived, since Aristotle does not entertain any sense of naturalistic explanations that compete with supernaturalistic ones, and this is because he is not operating with anything like a sense of nature that opposes it to the supernatural. Given his understanding of nature (physis) as ‘what happens always or for the most part’, the relevant contrast cases for Aristotle are not ‘the supernatural’ but rather ‘chance’ or ‘the accidental’ (Physics 197a31–35, 196b, 15–16). It is these that threaten the regularity of nature's operations and thus represent a challenge to the possibility of scientific explanation. If anything, and in what will become a fairly consistent pattern in the history of science, God here looks more like a source of the regularity (and hence intelligibility) of nature's operations than an unhelpfully disruptive influence that will make scientific explanation impossible.

It is worth mentioning two other oppositions that are important for Aristotle: natural versus violent, and natural versus artificial. These distinctions also turn out to be important in the history of science, since on Aristotle's version of events any human interference in natural processes renders them unnatural and ‘violent’. This accounts for the resistance of early modern Aristotelian natural philosophers to the new idea that experiments could reveal anything about nature. By definition it was thought that any interference with nature would change the object of investigation. Similarly, the natural–artificial distinction seems to rule out the possibility of mechanics serving as a model for natural processes, since mechanics involves artificial human creations. All of this is consistent with the fact that Aristotle's deity is not a creative agent, since this would make nature an artefact and hence something ‘unnatural’. Again, this conception of the natural was challenged by Jewish and Christian thinkers for whom the world was created by God, and hence was in some sense an artefact. Indeed, ‘natural’ for these thinkers (and directly contrary to Aristotle's understanding) meant ‘created by God’. Arguably, the idea that the world was a product of a divine creative act assisted early modern thinkers to make the case that mechanics could be combined with a genuinely realist account of natural motions. Our second historical counterexample to a simple opposition between naturalist and supernaturalist involves a similar point.

In late antiquity, the Christian philosopher John Philoponus (c.490–c.570) argued against a number of prevailing scientific doctrines, rejecting Aristotelian dynamics and proposing in its place a theory of impetus that resembles the modern theory of inertia. Directly relevant to the present discussion about naturalistic explanation, he contended that because the world had been created by a single deity – the Christian God – stellar bodies were not divine and, further, that the heavens and the earth must have similar properties and principles of motion. Combined with his theory of impetus, the first contention thus represents a theological, yet naturalistic, account of the motions of the stars. The second – a unified theory of dynamics – was similarly informed by his assumption of a ‘supernatural’ creation of the heavens and the earth as a single act (Lindberg (Reference Lindberg1978), 11; Sorabji (Reference Sorabji1987) ). These ideas played a role in the later Renaissance rejection of Aristotelian ideas of motion and offer some vindication of historian Pierre Duhem's claim that Philoponus should be regarded as ‘one of the main precursors of modern science’ (Duhem (Reference Duhem1913), 398). In sum, the monotheistic commitments of Philoponus led him to offer both a ‘naturalistic’ account of heavenly motions (the stars are not self-moving celestial intelligences) and a unified dynamics (the same principles of motion obtain on the earth and in the heavens). Again, in this instance, it makes little sense to speak of competition between naturalistic and supernaturalistic explanations, with the former being ‘successful’. To a degree, there are competing theological conceptions at play, with the idea of single, creative Deity providing the foundation for a more naturalistic explanation of celestial mechanics.

Our third example involves the emergence, for the first time, of a formal, terminological distinction between natural and supernatural. As we have seen, Aristotle had contrasted the natural with the accidental, the artificial, the violent, and the conventional. Earlier Christian writers had understood an alternative distinction between created (natural) and uncreated. But the specific terminology of a natural–supernatural distinction did not become common until the high Middle Ages. Peter Lombard (1096–1164), explained that some events come about through the joint operation of divine causes and causes inherent in things. These events are said to come about ‘naturally’. Other events, however, are brought about by God alone, and these are said to be ‘beside’ or ‘above’ nature.Footnote ⁷ These latter events are ‘praeternatural’ or ‘supernatural’. Crucially, though, God was still causally implicated in natural events, which could not take place without his ongoing conservation of natural agents, or his concurrence in the exercise of their causal powers.Footnote ⁸ On this view, then, God's ongoing creative agency was always operative in natural events. The distinction between natural and supernatural became widespread from the thirteenth century onwards and this is reflected in the changing frequency of relevant terminology (de Lubac (Reference De Lubac1946), Bartlett (Reference Bartlett2008) ).

Interestingly, this new distinction afforded scholastic philosophers the opportunity to adopt an approach that resembles, from our modern perspective, a naturalistic approach to scientific investigation. Albert the Great (c.1200–1280), for example, declared in connection with his discussion of scientific topics that ‘since I will discuss natural things, I am not concerned about the miracles of God’ (Albertus Magnus (Reference Albertus and Borgnet1890), 363). Siger of Brabant (c.1240–1280) advocated a similar approach to natural philosophy: ‘But the miracles of God do not now concern us, since we will discuss natural things naturalistically.’Footnote ⁹ This looks very much like our present methodological naturalism but, again, with the proviso that this ‘naturalistic’ explanation assumed an indispensable divine involvement.

A fourth and final example of why a simple natural/supernatural distinction does not map well onto the past concerns the modern development of the idea of laws of nature. The idea that nature is governed by universal, mathematical laws first appears in the seventeenth century, displacing the older Aristotelian conception of a nature ordered by the inherent virtues or powers of natural objects which direct them towards particular ends (Milton (Reference Milton, Garber and Ayers1998), Henry (Reference Henry2004), Harrison (Reference Harrison and Watkins2013) ). Descartes made the key move, insisting that natural philosophy dispense with final causes and focus instead on efficient causes. If, for Aristotle, God had been the source of all cosmic motion as an ultimate final cause, for Descartes, God was the cause of motion as an immediate efficient cause. This did not result in a chaotically unpredictable world since divine immutability guaranteed both the immutability of the laws and the reliability of inductions (Descartes (Reference Descartes, Cottinghman, Stoothoff and Murdoch1985), I, 202). Indeed, if anything, the world becomes more predictable since the Aristotelian allowance for exceptions, based on his ‘for the most part’ condition, is now dispensed with. The laws of motion, says Descartes, ‘follow manifestly from the mere fact that God is immutable and that, acting always in the same way, he always produces the same effect’ (ibid., 90–91, 93–96; cf. 240–249).

English natural philosophers also adopted the idea that the regularities of nature were divine decrees (although they differed from Descartes on the issue of how the laws were to be derived). Robert Boyle maintained that at the moment of creation God established the laws of nature and continues to uphold them at every moment (Boyle (Reference Boyle1674), 4; Idem (Reference Boyle1688), 91; Anstey (Reference Anstey1999) ). Newton, and the Newtonian natural philosophers who came after him, adopted a similar understanding. Gravity could be understood as ‘the immediate fiat and finger of God, and the execution of divine law’ (Bentley (Reference Bentley and Dyce1838), 74–75; Newton (Reference Newton and Cohen1978), 279–312). The effects of nature, on this understanding, were nothing other than divine power ‘acting according to fixt and certain laws’ (Whiston (Reference Whiston1696), 6, 211). Samuel Clarke, Newton's spokesman in the Leibniz–Clarke controversies, wrote similarly that:

Clarke insisted that there was really no such thing as ‘the course of nature’ or ‘the power of nature’. Rather, ‘nature’ as commonly conceived, was an imprecise shorthand or, to use Robert Boyle's descriptor, a ‘vulgar’ notion (Clarke (Reference Clarke1738), II, 697–698). In our own terms, ‘nature’ was an illicit reification or a deliverance of folk metaphysics. For these seventeenth-century thinkers, ‘nature’ was an idle conception that added nothing to an understanding of the regularities of the world upon which science was predicated. This view of laws of nature persisted into the nineteenth century, and was defended by leading scientific figures such as John Herschel (Reference Herschel1830, 137) and William Whewell (Reference Whewell1836, 362).Footnote ¹⁰

Two features of this new conception of laws of nature are worth noting. First, the laws of nature derive their necessity from the fact that it is an immutable God who wills them into existence. At the same time they are also in a sense contingent, since the specific laws were chosen arbitrarily by God. The first condition provides a divine guarantee of the regularity of nature's operations; the second provides a motivation for experimentation since only by experience can we discover what laws God has chosen (Newton [1713] (Reference Newton, Cohen and Whitman1999), 397). The second noteworthy feature of this new approach is its susceptibility to redescription in purely naturalistic terms. This occurs when the seventeenth century's ‘laws of nature’, which are in effect regular divine volitions, become, over the course of the eighteenth and nineteenth centuries, brute and inherent features of nature itself. Somewhat ironically, the possibility of describing the totality of nature's operations as ‘natural’ is predicated upon this early modern flattening of all causation into a single layer of supernaturally originating efficient causes. What for Boyle was a vulgar error was subsequently to become an almost universal presupposition of modern scientific investigation.

More could be said about the examples set out above. But one thing should be clear: there is no simple story, drawn from the history of science, of an enduring and consistent pattern in which ‘successful’ naturalistic explanations compete with and displace ‘unsuccessful’ supernaturalistic explanations.

Science and supernatural explanation

In view of the preceding discussion, talk of supernaturalistic explanation in past historical contexts is problematic, since the historical actors did not recognize the natural–supernatural distinction in the way we presently do. Yet it is still worth attempting to resolve the question of whether there is anything in the history of science that might point to either successes or consistent failures of something like ‘supernaturalistic explanations’. One way around the difficulty of anachronistic uses of ‘supernatural’ would be to speak instead of how in the history of science theological conceptions involving divine activity might have featured in particular explanations of natural phenomena. Broadly speaking then, we might ask if ‘legitimate’ or ‘successful’ science even involved the invocation of God's activity. As we have already seen, virtually all natural philosophical enterprises up to the nineteenth century relied upon theological assumptions or presuppositions, which in turn made possible certain kinds of scientific explanations. The notion of laws of nature is an obvious example. A subsequent, and related, instance was the shared conviction of James Joule, Michael Faraday, James Clerk Maxwell, and Lord Kelvin that God's creative activity underpinned the principle of the conservation of energy (Wilson (Reference Wilson, Knox and Noakes2003), 322). In this section I want to propose two other ways in which theological considerations contributed positively to scientific investigation. First, theological ideas could provide a heuristic which could facilitate new avenues of theorizing and investigation. Second, on some occasions, God was invoked as the direct explanation of some natural phenomenon, and in what look like scientifically reputable contexts.

First, to theological heuristics. In 1277, Stephen Tempier, the Bishop of Paris, issued a condemnation prohibiting the teaching of 219 propositions that had been topics of discussion at the University's Faculty of Arts. The condemnations pick out a variety of targets, but there is an interesting set of propositions that concern divine omnipotence and what God counterfactually might will if he so chose. One such proposition (these are propositions that are being condemned) reads: ‘That God could not move the heaven in a straight line, the reason being that He would then leave a vacuum’ (Grant (Reference Grant1974), 48). The article references the traditional conception of the horror vacui – nature's abhorrence of a vacuum – and the possibility of the motions of the heavens as a whole, both of which contravened Aristotelian physics. It has been argued that prohibitions such as this liberated scholastic thinkers from an undue deference to Aristotle, which enabled them to entertain theoretical possibilities that were either counterfactual or at odds with the prevailing scientific views. It was on this basis that Pierre Duhem contended that we see in the Condemnations of 1277 the beginning of modern science (Duhem (Reference Duhem1906–1913), I, 412; cf. Grant (Reference Grant1969), 59).Footnote ¹¹ Employing the idea of divine omnipotence as a heuristic could, in this way, promote scientific advance.

More generally, and adverting to the brief discussion in the first section of this article, the supposition of divine omnipotence also underpinned late medieval conceptions of what we now call the underdetermination of hypotheses by data. Aquinas thus remarked in his commentary on Aristotle's De caelo that astronomical hypotheses might be empirically adequate – which is to say that they fit with observations – and yet not be true, ‘because the astronomical phenomena can perhaps be saved in some other way not yet understood by men’ (Aquinas (Reference Aquinas1886), 186f.). Similarly Galileo's Simplicio – the character in the Dialogue concerning the two Chief World Systems (1632) who represented conservative Aristotelianism – notoriously refused to acknowledge that the Copernican hypothesis was ‘true and conclusive’, since ‘God in his infinite power and wisdom’ could have organized the motions of heavenly bodies ‘in many ways which are unthinkable to our minds’ (Galileo [1632] (Reference Galileo and Drake1962), 464). While Simplicio is rarely regarded as an icon of scientific sophistication, his stance bears a remarkable similarity to P. Kyle Stanford's (Reference Stanford2006) recent defence of underdetermination, which is similarly based upon the possibility of unconceived alternatives to our present scientific theories. Stanford remarks that we are not ‘cognitive supercreatures who are adept at conceiving of all possible theoretical explanations for a given set of phenomena’ (ibid., 45).Footnote ¹² For this reason, Stanford maintains, all of our scientific theories are going to be underdetermined. Stanford's ‘cognitive supercreature’ is not too far away from Simplicio's omnipotent and omniscient deity. Whether all of this counts for or against the use of divine omnipotence as a heuristic may depend on one's commitments to scientific realism.

A second theological heuristic is design. In a post-Darwinian world, notions of design have understandably been anathema to evolutionary biologists, and it is clear that the proponents of Intelligent Design have fallen well short of making a case for its return as a scientific principle. But given that our interest is the history of science, rather than the present, it is worth noting how on occasion the assumption of divine design could promote scientific successes. One instance is William Harvey's discovery of the circulation of the blood. Harvey argued that from the positions of the valves in the veins, and on the assumption that God has designed them for a purpose, he was led to hypothesize that blood was sent from the heart through the arteries and circulated back via the veins.Footnote ¹³ For Harvey, the assumption of divine design provided a true insight into how blood circulated through the human body.

Another, less obvious, role for design is the way in which William Paley's now notorious arguments for the ‘contrivance’ of living things set up the explanandum for Darwin's theory of evolution and provided an analogy for the model of natural selection. Paley's watchmaker argument drew a parallel between human and divine design. Darwin was led to enquire whether nature might ‘select’ particular creatures on analogy with the way in which human breeders select particular stock lines (Darwin (Reference Darwin1859), 84; Livingstone (Reference Livingstone, Lindberg and Numbers2003), 188). There is no doubt that, for Darwin, God was not directly involved in the design of creatures. But the ubiquitous idea of design formed an important part of the context of the discovery of his principle of natural selection.Footnote ¹⁴

Finally, it is worth making reference to Isaac Newton's resort to both the principle of design and direct divine action. Newton did not think that the specific arrangement of the planets of the solar system, or the means by which the matter of the universe forms into the distinct bodies of suns and planetary masses, was susceptible to explanation ‘by mere natural causes’. He attributes the causes of these phenomena, respectively, to ‘the counsel and contrivance of a voluntary Agent’ and ‘an intelligent Agent’.Footnote ¹⁵ Newton also recognized two potentially fatal threats to the stability of the cosmos on his understandings of the operations of gravity. One was that, given mutual attractions of the celestial bodies, all of the masses in the universe would naturally tend to convene at its centre. In order to overcome this difficulty Newton postulated an infinite universe, with stellar masses carefully and evenly distributed to ensure no gravitational collapses to a central point. This required some fairly delicate, divine fine-tuning, with Newton conceding that ‘this frame of things could not always subsist without a divine power to conserve it’.Footnote ¹⁶

At a local level, Newton was again acutely conscious of the inherent fragility of our solar system. The stability of planetary orbits was explicable if one assumed gravitational attraction between two bodies balanced by an equal and opposite force owing to the curvilinear motion of the planetary body. But the solar system is much more complicated than this because the sun, planets, moons, and comets all exert gravitational forces on each other. While most of these forces will be quite small in comparison to the gravitational attraction of the sun, cumulatively and given enough time, there will be perturbations of planetary orbits which could ultimately threaten to wreak havoc with the solar system. To address this problem Newton infamously spoke of the need for a ‘reformation’ of the planetary orbits. This reformation would be brought about by divine action (Newton (Reference Newton1706), query 23).

This bold step is often read as a blot on Newton's copybook, an unfortunate resort to the ‘god of the gaps’ that was scientifically illicit and theologically unhelpful. Leibniz certainly thought so. Significantly, though, Leibniz's primary objection was theological not scientific. The usual sequel to the Leibniz–Newton controversy is the story according to which Pierre-Simon Laplace eventually solved the problem of the stability of the solar system with his own mathematical modelling. This was deemed superior on account of its purely naturalistic account of events and Laplace, like a good naturalist, was famously said to have remarked that he had no need for the God hypothesis. But in fact Laplace's touted solution failed to solve the problem, and his ambitiously deterministic view about the predictability of all future events in the universe was mistaken. We now know that the solar system is a partly chaotic system and that precise prediction of the future orbits of the planets is in principle impossible. There was a sense in which Newton was correct to see the problem as insoluble, at least in terms of being able to offer assurances about the stability of the solar system on the basis of gravitational theory. (Recent computer simulations reassuringly postpone disastrous planetary collisions for several billion years, but this only on the basis of probabilistic calculations (Laughlin (Reference Laughlin2009), Tremaine (Reference Tremaine2011) ).)

Does Newton's enlisting of God in these cases represent the ‘successful’ use supernaturalistic explanation in the history of science? Again, it really depends on what counts as successful. Arguably, it was Newton's apparently ad hoc invocation of God on a number of occasions that enabled him to persist with a cosmological theory which, with the benefit of hindsight, was the most scientifically sophisticated at the time and which might otherwise have been falsified by fundamental and apparently insoluble difficulties. Putting it crudely, reference to the supernatural worked to support a scientific model that was basically correct – or certainly more correct than the contemporary alternatives. This strategy would certainly not do in the present, when the trend is to invoke occult entities such as ‘dark energy’ and ‘dark matter’ to preserve theories in the face of potentially falsifying anomalies. But like Newtonian divine action, these are ad hoc place-holders for incomplete, yet otherwise well-attested, theories.

Naturalistic reconstructions and selective realism

Sceptical readers who have persisted to this point will no doubt have significant misgivings about some of the directions of my argument – perhaps foremost among these a worry that in the historical cases described above the supernaturalistic commitments do not do any of the real work and hence play no substantive role in the postulated progress of science. The episodes in question might thus be susceptible to reconstruction in purely naturalistic terms. This worry is redolent of Imre Lakatos's contentious notion of ‘rational reconstruction’ – aptly characterized by Thomas Kuhn as ‘not history at all, but philosophy fabricating examples’ (Lakatos (Reference Lakatos1971); Kuhn (Reference Kuhn1971), 143). But a charitable way of articulating the concern might be to draw parallels to the ‘selective realism’ response to pessimistic induction. This response grants that past theories now regarded as false did in fact provide a basis for successful novel predictions, but insists that only specific parts of those theories – parts that were true or approximately true – were responsible for the relevant successes. The success of the theories in question, according to this ‘divide and conquer’ strategy, did not depend on components of the theory that were false and now abandoned (Kitcher (Reference Kitcher1993), 140–149, Psillos (Reference Psillos1999), 96–139). It could be argued analogously that the supernaturalist components of the theories I have described were not the elements that contributed to their success. We can imagine, for example, laws of nature yielding true predictions without thinking of them as divine volitions.Footnote ¹⁷ It might then be possible for the supernatural to be retrospectively written out of the story of science's success.

A direct rejoinder to this objection would involve the identification of specific instances where the supernaturalistic elements of the theory did in fact seem to do some of the heavy lifting: Newton's insistence on absolute space, against the arguments of Leibniz, might be an example. This counter-response would mimic responses to selective realism, which carefully rehearse historical case studies in which it seems clear that the ‘false’ elements of theories (from our present perspective) contributed to their success (for examples see Lyons (Reference Lyons2006), Rossetter (Reference Rossetter2018) ).Footnote ¹⁸ But my suspicion is that for the already committed naturalist there are neither actual nor even conceivable developments in the history of science that could not in principle be redescribed in purely naturalistic terms. (If there are such instances, it would be interesting to see them specified.) If that suspicion is well founded, then it is not clear how actual history, as opposed to ‘history’ that never happened, can be relevant to the discussion at all, apart from offering challenges to the reconstructive ingenuity of contemporary naturalistic philosophers. The reconstruction move, in short, tells us nothing about the past (or future) and everything about the present.

To put this another way, like the selective realism response to pessimistic induction, this strategy begs the question (cf. Stanford (Reference Stanford2003), 914). Our choice of which elements of past theories are successful and which are true is based on our present understanding of what is true. Yet it is precisely this that the argument from pessimistic induction places in doubt, since it suggests these contemporary truth commitments, as represented in our best scientific theories, are likely to be rendered false in the future, mutatis mutandis for our present commitments to naturalism.

Finally, not only does this kind of response beg the question, it misses the point. The argument we are testing is not whether it is possible to contrive naturalistic redescriptions of past non-naturalistic science. What is at issue is whether there has been a long-standing deployment of methodological naturalism, as evidenced by the history of science, that has led to successful outcomes in a way that alternative ‘supernaturalistic’ approaches have not. The historical examples show that this is not the case.