Worldviews and evidence

For a start, let us distinguish the following two general metaphysical worldviews (cf. Corner Reference Corner2005):

Naturalism: There are no entities and nothing causally effective ‘outside’ nature.

Supernaturalism: Entities and causally effective items ‘outside’ nature do exist.

This is admittedly very rough and vague as it stands, for it is not obvious how ‘nature’ in turn should be more precisely defined. But it will do for now.

If supernaturalism and a religious worldview (interpreted in a certain literal metaphysical way) are already assumed, miracles may be an integral part of this worldview. All kinds of events can then be easily seen as expressions of the will of a supernatural being, or effects of other supernatural causes. But in that case, miracles cannot, and there is no need for them to, serve as evidence for believing in the worldview. Moreover, within such a worldview, there is no need to draw a sharp boundary along the Humean lines between miracles and natural events which conform to the laws of nature. Miracles may then be taken to be simply events which result in more wonder and amazement than many others or have specific religious significance (cf. Holland Reference Holland1965).

Miracle stories have had many roles in various religions. However, in the philosophy of religion, one important role seems to have been the so-called apologetic use of miracles. That is, (alleged) miracles are often used, as it were, as neutral (i.e., without presupposing supernaturalism) decisive evidence for the truth of supernaturalism and specifically of a religious worldview such as Christianity and the existence of the Christian God. In such contexts, the crucial question is: Is it really possible to have cogent rational evidence for and sound knowledge about the reality of miracles without already presupposing supernaturalism?

It is important to note that there are many forms of religious standpoints for which miracles are not at all central. And even if they are, they need not be viewed primarily as persuasive neutral evidence for the truth of the religion at issue. Nevertheless, in the present paper, I want to focus primarily on the apologetic uses of miracles.

Here, an important qualification is in order: It was a part of the original Humean definition (the more specific one) – and this recurs in various recent definitions – that the violation of the laws of nature was especially brought about by a god or some other supernatural intentional agent. But surely, if it is a part of knowing that an event is a true miracle that it was brought about by a divine being, the occurrence of a miracle cannot then really be evidence in favour of the supernaturalist worldview and the existence of that god. Swinburne (Reference Swinburne1970, 7), for example, grants this: ‘there cannot be miracles which are evidence for his [God’s] existence because accepting a description of an event as a miracle commits a man to accept the existence of God’. Swinburne (Reference Swinburne1970, 21) nevertheless announces he assumes that ‘we have no evidence for or against the existence of a god other than historical and scientific evidence for or against the occurrence of miracles’.

Be that as it may, my principal focus below is on the Humean idea that it is at least a necessary condition for something to truly count as a miracle that it violates the laws of nature. (Many of my observations below are also relevant for the more general idea that miracles are something that natural sciences cannot explain even in principle.) If it is further required that the event must be caused by a god or some other supernatural being, we may rephrase the epistemological question about the apologetic use as whether we can have compelling evidence that we actually have a miracle before us. In other words, instead of viewing miracles as evidence for supernaturalism or the existence of a god, we may then ask what would count as persuasive evidence for something to be a genuine miracle. More precisely, we now face two epistemological questions: First, how can we know that the laws of nature have been, in fact, violated? And second, how can we know that the violation really is due to the intentional action of a divine being?

I shall now argue that various advances in the philosophy of science in recent decades are highly relevant for the whole theme of miracles (thus understood) and significantly change the big picture.

Can we possess the true laws of nature?

Behind many discussions of miracles, especially those which are somehow related to Hume’s definitions, is evidently a certain picture of science, the laws of nature, and scientific explanation. That general picture has indeed dominated discussions of science for a long time. Let us call it ‘the Once Received View’.Footnote ² At the heart of this picture is the idea that the natural sciences essentially aim to discover relevant laws of nature, and scientific theories are intended to describe such laws (cf. Swinburne Reference Swinburne1970, 23). According to this picture, an explanation of an event or a phenomenon in science amounts to deriving it from such theories. The laws of nature also enable, at least in principle, the prediction of future events.

Furthermore, it was long assumed that Newton had discovered the true laws of nature, at least within physics. As for the other fields, it was commonly expected that their laws could either be derived from those of Newton or at least were likewise discoverable for good with the help of some further observations and experiments.

It seems that a picture along these lines inspired Hume and much of the subsequent discussion of miracles. It makes reasonably clear sense, in such a context, to talk about miracles as violations of the laws of nature. However, the Once Received View has fallen into disfavour within the philosophy of science in recent decades. A closer look at the variety of natural sciences, their actual practices and history, demonstrates that the picture described is seriously flawed. Our understanding of what the natural sciences are and do has changed quite dramatically.

To begin with, epistemological optimism encouraged by Newton’s laws did not last forever. Even Newton’s laws ran into trouble at the end of the 19th century and were eventually replaced by Einstein’s theory of relativity. The latter may not be the final word either. Somewhat similar incidents have recurred in various special sciences: theories have been found to be problematic and consequently revised or rejected.

In general, it has been standard already for some time to think that even our best well-developed scientific theories may turn out to be, strictly speaking, false. In favourable circumstances, we may still get closer to the truth and achieve in due course at least approximately true theories, which may quite often suffice, e.g., for practical applications. Still, our theories may never attain, even in their own area, the whole truth and nothing but the truth. As Popper liked to put it, the scientific project amounts to an interminable process of repeated trial and error.

Overall, the idea that we could have the absolutely true and comprehensive laws of nature in our possession and know definitely that this is so has vanished. The ideal of epistemic certainty in science has been replaced by a more modest fallibilist attitude: all human systems of belief, including even the most advanced scientific theories, are in principle fallible and may require continuous revision. We can never be completely certain that a theory at stake is absolutely true.Footnote ³

Accordingly, even if, for example, the observed behaviour of some particles in a particle accelerator does not fit the presently postulated ‘laws of nature’, scientists do not likely conclude that we now have a violation of the laws of nature, a supernatural miracle, before us. More probably, they simply go back to the drawing board and attempt to formulate more defensible hypothetical laws.

At this point, let us pause to note that, among the friends of miracles, at least Swinburne, for his part, shows a certain degree of awareness of the complications just discussed. (Whether he gives them and their consequences sufficient attention is less clear.) Apparently, at least partly for such reasons, and partly in order to avoid making miracles trivially impossible, Swinburne (Reference Swinburne1970, 26–27) adds to his definition of a ‘miracle’ the condition of being ‘non-repeatable’ (which he borrows from Smart Reference Smart1964, 37).

Some seem to think that this effectively removes all such worries. However, one may doubt whether it is, in reality, a persuasive response. Namely, no finite amount of observations can verify that an event is genuinely non-repeatable. It is always possible that it is just exceedingly rare, but nevertheless repeatable. There is also the epistemic problem of taking into account all potentially relevant causal factors (see below). Thus, the condition might make it, for its part, practically impossible to know that something really is a miracle.

Moreover, some common alleged miracles do not seem to be non-repeatable; think of, e.g., faith healing or exorcism, which are presumably quite repeatable, not to mention the real presence of Christ in the Eucharist and even more miraculous Eucharistic miracles. Therefore, the requirement that miracles must be non-repeatable seems also to violate the established understanding of miracles and to artificially redefine them and narrow their scope.

How can we know that the laws are being violated?

As to the laws of nature, there is even today little agreement on what defines their essential features. There is more than one quite different, competing interpretation in the market. There are even distinguished philosophers of science who contend that there exist no such things as the laws of nature. (For overviews of the variety of positions, see, e.g., Lange Reference Lange, Psillos and Curd2014; Roberts Reference Roberts and Humphreys2016; Carroll Reference Carroll, Zalta and Nodelman2025; in relation to the theme of miracles, see Luck Reference Luck2011.) Be that as it may, traditionally, the laws of nature have been assumed to be deterministic as well as general and universal, that is, exceptionless ‘strict laws’. Newton’s second law (F = ma) has been a classic, paradigmatic example.

One should then recognise that laws of nature, as traditionally interpreted, are entirely general and do not in themselves entail and predict any individual event or such. Only when combined with more detailed initial conditions can they have any more specific implications (Larmer Reference Larmer1985 acknowledges this). Moreover, even then, that may often require various auxiliary assumptions (e.g., assumptions about sophisticated instruments, which may sometimes be significantly theory-dependent and fallible).Footnote ⁴

But knowing even the sufficient initial conditions, except for untypically simple cases, may well be far beyond human capabilities. For example, a causal influence can travel at the speed of light and hence appear very quickly from an astronomical distance. Therefore, taking into account all potential factors may require knowing precisely, roughly, the entire state of the whole universe (more exactly, the entire cross section of the relevant light cone) at a given previous time (cf. Field Reference Field, Loux and Zimmerman2003).

Consequently, establishing that one genuinely has a violation of the laws of nature in front of one may sometimes require a prior knowledge of – not only the precisely true and comprehensive general laws of nature but also – the exactly correct, detailed, and comprehensive initial conditions. And that, as I have just underlined, can in fact be far more difficult than some have naïvely supposed. It may require a virtually infinite amount of exact knowledge of every elementary particle. Add the many auxiliary hypotheses that the concrete setting might require, and that increases further the risk of error, and it is likely to be humanly impossible to provide compelling evidence for an actual violation of the laws of nature. Instead, the source of trouble may well be an imprecise assumption about the initial conditions or an inaccurate auxiliary hypothesis.

It is for such reasons that Popper’s falsificationism has been generally discarded (for overviews, see, e.g., Godfrey-Smith Reference Godfrey-Smith2003, ch. 4; Ladyman Reference Ladyman2002, ch. 3). Largely the same reasons make, for their part, the idea that we could know indubitably that we have a genuine violation of the laws of nature before us highly dubious.

As was noted, traditionally the laws of nature have been assumed to be deterministic and universal. Obviously, one could expand the notion of a ‘law’ and allow probabilistic laws and/or so-called ceteris paribus laws (laws which include a ceteris paribus clause, such as ‘under normal conditions’, and can thus have exceptions). Indeed, some philosophers of science have done exactly that. However, in our present context, those would be problematic moves. For it seems that it would be even more difficult to establish that we have a violation of the laws before us. If a law states that a certain outcome is merely very probable, and/or occurs only under normal circumstances, given the condition, the possibility of a sporadic exception is built into the law, and the occurrence of such would in no way violate the law.Footnote ⁵ (This is not to say that a putative ceteris paribus law could never be rationally rejected in science. But in such a case, it was presumably false to begin with. However, it seems extremely difficult to know conclusively both that we possess an absolutely true ceteris paribus law and that a singular event from everyday life violates it.) All things considered, it seems that within the broadly Humean definitions of miracles, it would be better to stick to the understanding of the laws of nature as universal and deterministic.

From laws to mechanisms and models

Whatever one may have assumed in, say, the 18th century, it seems quite clear today that genuine laws of nature, as far as there are any, concern primarily elementary particles and not directly observable macro-level phenomena. The garden-variety examples of (alleged) miracles, by contrast, do not engage with the phenomena of fundamental physics and elementary particles, or even atoms or molecules, but typically with higher macro-level issues, e.g., events of ordinary life, which are mainly the business of, e.g., biological and medical sciences.

In recent years, it has become common in the philosophy of science to distinguish more clearly physics (or, particularly, fundamental physics) from various so-called special sciences (e.g., biological and medical sciences). And once one looks more carefully, and without preconceptions, at the various special sciences outside physics, there seem to be hardly any genuine laws of nature within them. Special sciences describe and explain phenomena, but frequently this does not seem to involve ‘laws of nature’ in any traditional sense of the term.

So, what do these sciences do if they do not search for laws? It seems that at least quite often scientists discover different mechanisms; this appears to be the case especially in biological sciences and neuroscience, for example. That is, there is a phenomenon which is in need of an explanation. The aspiration is then to discover and reveal the mechanism that produces, underlies, or maintains the phenomenon, and the latter is thereby explained mechanistically. Schematically, a mechanism is something that has parts with specific functions in the mechanism, and the parts are organised in a certain way to form a complex entity capable of producing a phenomenon. Examples of such mechanisms in science include, e.g., protein synthesis, photosynthesis, action potential, working memory, the heart, body temperature homeostasis, and metabolism.

This new viewpoint in the philosophy of science is commonly called ‘the New Mechanistic Philosophy’.Footnote ⁶ It should not be conflated with the traditional mechanistic philosophy of the 17th century (e.g., Descartes, Boyle). For unlike the latter tended to do, the explanations using mechanisms are now not assumed to necessarily be reductionistic, deterministic, or atomistic.

To be sure, the operation of such mechanisms and their parts is somewhat regular, but not without exceptions: There may well be a malfunction, or the mechanism may break down, such as, e.g., with digestion and its disorders. As their regularities are not assumed to be anything like laws, exceptions are perfectly possible and sometimes even quite commonplace, and all that is a part of the very framework. They are emphatically not anything like violations of the laws of nature. Glennan (Reference Glennan1996) suggests that, in fact, even many of the traditional so-called ‘laws’ of science are more like this: e.g., Mendel’s laws, Kepler’s laws, or Hooke’s law.

The new mechanistic philosophy does not contend that scientific research is always just about uncovering mechanisms or that all explanation in science is explanation in terms of mechanisms, only that mechanisms are a common and important theme in science, which has been somewhat neglected in the earlier philosophy of science literature.

The more straightforward traditional picture of science also typically suggests that there are theories and there is reality (or its part or its aspect) which theories at least aspire to describe. Another important development in the recent philosophy of science has been the realisation that this is in many cases too simplistic a picture. Namely, in various developed sciences, different (mathematical, computer, and other) models habitually play an essential role. Indeed, scientists construct and manipulate models continually. Models in science, however, are deliberately idealised and simplifying. (For overviews of the role of models in science and their philosophy, see, e.g., Bailer-Jones Reference Bailer-Jones, Machamer and Silberstein2002; Contessa Reference Contessa, French and Saatsi2011; Morrison Reference Morrison and Humphreys2016; Portides Reference Portides, Psillos and Curd2014; Frigg and Hartmann Reference Frigg, Hartmann, Zalta and Nodelman2025).

At the very least, this further complicates the straightforward picture that scientific theories simply describe or represent (for better or worse) the external reality. It is often the case that models, so to speak, mediate between theories and reality. A key feature of all this is that models regularly misrepresent their subjects; they are significantly oversimplified and idealised – and scientists know this very well. Indeed, that is intentional for their part. Deviations from the models are thus assumed to be possible from the start. If an actual observation then conflicts with the prediction of such a model, that is definitely not considered as proof of a supernatural intervention.

Whatever the status of determinism is with respect to the hypothetical complete true theory of physics,Footnote ⁷ which is in any case still, and probably forever, beyond our capabilities, it is quite uncontroversial today that at the macro-level and in various special sciences, all we can very often do is identify weaker regularities and even at best more local, probabilistic laws with the built-in possibility of exceptions.

Such a more up-to-date and sophisticated picture of natural sciences poses serious critical questions for the whole theme of miracles: The broadly Humean definitions of miracles and the idea of knowing about miracles so understood become quite problematic. Moreover, it is not at all clear what a more sustainable and yet natural and non-arbitrary definition that would still harmonise with the traditional conception of a ‘miracle’ would look like. How exactly could we reliably recognise a miracle, have compelling evidence that we have indeed a miracle before us?

More concrete examples

Let us briefly consider a few more concrete examples in order to put some flesh on the bones: To begin with, there are some recognised phenomena in nature, such as ball lightning and St. Elmo’s fire, which are exceedingly rare and for the present not completely understood. They occur irregularly, and there are no laws of nature, at least any known ones, governing their occurrence. Nevertheless, scientists do not consider them as supernatural interventions but view them as perfectly natural phenomena in need of a scientific explanation. In the biological realm, there are hardly any proper laws: species evolve, sometimes even quite rapidly, mutations occur, and overall development may be highly unpredictable (though we know that natural selection is at work).

Stories involving faith healing are a quite common type of miracle narrative. One may wonder, though, what exactly are the laws of nature that are violated in such cases? It is nowadays an established fact in medical sciences that, e.g., one can seldom recover from some forms of cancer, even from an advanced stage; this is called ‘spontaneous remission’. Analogously, a blind person may occasionally gain the ability to see. Such cases may be anomalous, but they do occur. Even the so-called ‘Lazarus syndrome’Footnote ⁸ is well documented. It refers to the extraordinary cases where a patient has been declared dead, but after some time, the heart activity spontaneously returns. A few have even fully recovered. The phenomenon is extremely rare, and, at present, it lacks a satisfactory explanation. Although tremendously unusual and still incompletely or even poorly understood, these are standardly taken as entirely natural phenomena, and not as proof of the intentional intervention by some supernatural personal being.

In the pre-scientific era, when the supernaturalist worldview reigned, such erratic events were no doubt easily interpreted as expressions of the will of a supernatural personal agent. However, if supernaturalism is not presupposed in advance, such cases hardly count as compelling evidence for supernaturalism. Today, such incidents, even if agreeably exceptional, are established facts within the naturalistic worldview. To be sure, some events may be extraordinary and cause amazement. However, it may be all but impossible to establish convincingly that a violation of the laws of nature has occurred, and even more difficult to show that an abnormal event results from the intentions of a personal supernatural being.

In general, just how law-governed, regular, and predictable our macroscopic reality actually is appears to be a contingent empirical question. And it now strongly seems that it is less so than the naïve mind may have earlier assumed: the world we live in is often quite messy, irregular, and unpredictable – sometimes even random and chaotic. Many incidents apparently are largely matters of chance. For example, a hurricane or an earthquake can occur utterly unpredictably – as do the above-mentioned ball lightning and St. Elmo’s fire. That in no way forces the conclusion that there must be, after all, some supernatural causes responsible for such episodes.