Miracles and the Laws of Nature

Humean Evidential Critique

Abstract

In Section X of the Enquiry, Hume claims an argument which, "if just, will, with the wise and learned, be an everlasting check to all kinds of superstitious delusion" (Hume 1748, §X pt. 1). Part 1 argues from the nature of the case: a miracle is by definition contrary to uniform experience, so the proof against it is maximal, and testimony — whose reliability is itself only an empirical regularity — can never outweigh it. Part 2 argues that no actual miracle report meets even ordinary standards of credibility, and that religious reports carry special defeaters. The critique is "far and away the most influential treatment of the topic" (Rockwood forthcoming, p.1), shaping critical biblical scholarship from Strauss through Troeltsch to Ehrman (SEP §3.3).

Formal statement

Hume's definitional core: "A miracle is a violation of the laws of nature; and as a firm and unalterable experience has established these laws, the proof against a miracle, from the very nature of the fact, is as entire as any argument from experience can possibly be imagined" (Hume 1748, §X pt. 1, ¶90) — or, in his footnoted refinement, "a transgression of a law of nature by a particular volition of the Deity, or by the interposition of some invisible agent." The argument issues in the famous maxim, quoted verbatim:

"The plain consequence is (and it is a general maxim worthy of our attention), 'That no testimony is sufficient to establish a miracle, unless the testimony be of such a kind, that its falsehood would be more miraculous, than the fact, which it endeavours to establish; and even in that case there is a mutual destruction of arguments, and the superior only gives us an assurance suitable to that degree of force, which remains, after deducting the inferior.'" (Hume 1748, §X pt. 1, ¶91)

Hájek's reconstruction of Part 1: (1) a miracle is a violation of the laws of nature; (2) a law is a regularity to which no exception has been experienced; hence (3) there is as compelling a "proof" from experience against a miracle as can be imagined; (4) the proof in favour of testimony cannot be more compelling; therefore (5) the falsehood of miracle testimony is always at least as probable as the event attested, and belief in a reported miracle is never justified (SEP §3.1.2). Part 2 adds four empirical arguments: no miracle has adequately qualified witnesses; the passions of surprise and wonder — worst when "the spirit of religion join[s] itself to the love of wonder" — corrupt telling and receiving alike; miracle reports "abound among ignorant and barbarous nations"; and rival religions' miracles destroy each other's credit as contrary evidence (Hume 1748, §X pt. 2, ¶92-95).

Key evidence / textual basis

The epistemic engine is Hume's proportionality principle: "A wise man, therefore, proportions his belief to the evidence" (Hume 1748, §X pt. 1, ¶87). When the reported fact is maximally contrary to experience, "I weigh the one miracle against the other... and always reject the greater miracle" (¶91). Part 2 presses that even the best-attested prodigies — Vespasian's cures, the Jansenist cures at the tomb of the Abbé Paris, "proved upon the spot, before judges of unquestioned integrity" — should be rejected on "the absolute impossibility or miraculous nature of the events" alone (¶96); hence the conclusion "that no human testimony can have such force as to prove a miracle, and make it a just foundation for any such system of religion" (¶98). The eight-days-of-darkness and Queen-Elizabeth cases (¶99) mark the boundary: massive testimony might establish an extraordinary natural event, never a resurrection — "I should only assert it to have been pretended."

Contemporary scholarship divides over the strength of Part 1. Rockwood distinguishes a probability reading (Fogelin, George: an enormous but defeasible presumption) from a proof reading (the traditional one: an inference from an established law of nature is decisive, leaving "no room for doubt or opposition"), and argues the texts favour the latter (Rockwood forthcoming, pp.8-18). Jacovides reconstructs the contrary-miracles argument as genuinely empirical: since rival religions implicitly call each other's miracle stories false, "the unreliability of testimony for rival miracles boomerangs back on the foundations of [the] initial religious system" (Jacovides forthcoming, 'The Empirical Character of the Argument'). And Ellis takes the closing irony of Section X — "Our most holy religion is founded on Faith, not on reason" (¶100) — seriously: Hume shows that Christianity, if based on testimonial evidence for miracles, would be a superstitious delusion, so believers must locate its basis elsewhere (Ellis 2024, p.253).

Leading proponents

• David Hume — the canonical statement; in Part 2, the SEP notes, "Hume's credit... is not that of originating the arguments but rather that of stating them clearly and forcefully" (SEP §3.2) — Morgan and Toland had canvassed most of them in the deist controversy.
• David Friedrich Strauss, F. H. Bradley, Ernst Troeltsch, Bart Ehrman — the Humean principle rechristened as the "principle of analogy" and the "presuppositions of critical history" in biblical scholarship (SEP §3.3).
• Robert Fogelin, Peter Millican, William Vanderburgh, Lawrence Shapiro — contemporary defenders against Earman-style critiques (via SEP §3.3; not in corpus).

Strongest counter-arguments

First, the "uniform experience" premise begs the question. Paley: to state "that no such thing was ever experienced, or that universal experience is against it, is to assume the subject of the controversy" (Paley 1794, Preparatory Considerations). Richard Price: "a miracle is more properly an event different from experience than contrary to it" — there is not "a single piece of direct testimonial evidence to Jesus' non-resurrection" (SEP §3.1.2).

Second, the Bayesian rebuttal. Babbage proved that "if independent witnesses can be found, who speak truth more frequently than falsehood, it is ALWAYS possible to assign a number of independent witnesses, the improbability of the falsehood of whose concurring testimony shall be greater than the improbability of the alleged miracle" — a result endorsed by Holder and by John Earman, whose Hume's Abject Failure (2000) argues Part 1 is trivial where true and false where substantive (SEP §2.2.4; SEP §3.3).

Third, the parody objection. Whately's satire "establishes" that Napoleon never existed: Humean scruples, consistently applied, are "a standard that cannot be applied without absurdity in any other field of historical investigation" (SEP §3.3). Hájek likewise: if lack of analogy made an event maximally improbable, we should be comparably skeptical of all spectacular scientific discoveries — "And that is absurd" (SEP §3.1.2).

Fourth, the relevant-similarity objection. Inductive inference requires observed cases relevantly similar to disputed ones; but a law "tells us only what happens absent divine intervention," so evidence that As are Bs "is not relevant to those cases in which God intervenes" — where there is independent reason to expect intervention, "Hume's objection in Part 1 has no force whatever" (Rockwood forthcoming, pp.21-22).

Fifth, against contrary miracles: Richard Swinburne, Gaskin, and Langtry note the argument "only works in situations where there are rival religious traditions who reject miracle stories from other religions" (Jacovides forthcoming, 'The Empirical Character of the Argument'); Campbell had already observed that a hostile audience's conversion is a presumption for, not against, founding miracles (SEP §3.2.2).

Responses

Humeans reply (i) that Ahmed showed the Babbage-Earman result depends on a conditional-independence assumption "plausibly always violated" — though Ahmed concedes his arguments "do not after all realize the 'everlasting check'... that Hume envisaged" (SEP §2.2.4); (ii) that Fogelin, Millican, and Vanderburgh read Hume non-mathematically, as showing "how very high the bar is set" rather than as a failed Bayesian theorem (SEP §3.1.2); (iii) that the religious context of miracle reports remains an empirically documented defeater: "the many instances of forged miracles... which, in all ages, have either been detected by contrary evidence, or which detect themselves by their absurdity... ought reasonably to beget a suspicion against all relations of this kind" (Hume 1748, §X pt. 2, ¶93).

Assessment

Assessment: Live — the critique remains the operating assumption of much critical historiography, but its in-principle form is under the heaviest attack it has faced since the eighteenth century: "Hume's critique of reported miracles has itself come under heavy fire and is now viewed in some quarters as requiring defense" (SEP §3.3).

Evidentialist Defense

Abstract

The classical Anglican response — Butler before Hume, Paley after him — concedes the regularity of nature and the fallibility of testimony, and argues that the presumption against a miracle is nonetheless finite and defeasible. Butler's Analogy supplies the probabilistic framework; Paley's Evidences mounts the direct rebuttal of Hume's maxim and the positive case that the original witnesses staked their lives on their testimony. The contemporary Bayesian literature (Babbage, Holder, Earman, Swinburne) is in effect a formalization of this tradition.

Formal statement

1. If there is a God who wishes to authenticate a revelation, miracles are the only conceivable means: "Now in what way can a revelation be made, but by miracles? In none which we are able to conceive" (Paley 1794, Preparatory Considerations).
2. Hence the prior probability of miracles-in-attestation-of-revelation is no lower than the prior that God would give a revelation at all — "in miracles adduced in support of revelation there is not any such antecedent improbability as no testimony can surmount."
3. There is satisfactory evidence that many professing original witnesses of the Christian miracles "passed their lives in labours, dangers, and sufferings, voluntarily undergone in attestation of the accounts which they delivered" (Paley 1794, Part I).
4. There is not satisfactory evidence of the same for the witnesses of rival miracle claims (Paley's second proposition).
5. Therefore the testimony is of the kind whose falsehood would be "more miraculous" than the facts attested; by Hume's own maxim, belief is warranted.

Key evidence / textual basis

Butler supplies the epistemology: "to us, probability is the very guide of life" (Butler 1736, Introduction) — and he had already deployed the frost example later made famous by Hume's Indian prince, drawing the opposite moral. In Part II chapter ii he dismantles the "peculiar presumption" against miracles: "There is a presumption of millions to one, against the story of Cæsar, or of any other man" — any particular fact is antecedently improbable, yet overcome "by almost any proof"; miracles should be compared not to daily events but "to the extraordinary phenomena of nature" — comets, magnetism, electricity; hence "there certainly is no such presumption against miracles, as to render them in any wise incredible" (Butler 1736, II.ii). His cumulative-case principle — the truth of religion "is to be judged of by all the evidence taken together" — is read by the SEP as "a verbal explication of the categorical form of the Bayesian argument" (SEP §2.2.4).

Paley engages Hume by name. He exposes an ambiguity in "contrary to experience": strictly, a fact is contrary to experience only when we were present and perceived it not to happen; in the looser sense the objection is merely a want of experience — and a miracle, by its nature and purpose, is precisely not expected to recur on demand (Paley 1794, Preparatory Considerations). He indicts Hume for "a want of argumentative justice" in suppressing the theistic background: on Hume's framing, "miracles are alike incredible to him who is previously assured of the constant agency of a Divine Being, and to him who believes that no such Being exists" — whereas "once believe that there is a God, and miracles are not incredible." And he tests "Mr. Hume's theorem" on a simple case: twelve men of known probity, each separately choosing execution rather than recant their circumstantial account — "still if Mr. Hume's rule be my guide, I am not to believe them. Now I undertake to say that there exists not a sceptic in the world who would not believe them" (Paley 1794, Preparatory Considerations).

Scripturally, miracles-as-credentials is the New Testament's own category: Nicodemus reasons "no man can do these miracles that thou doest, except God be with him" (John 3:2 (bib)), and Elijah at Carmel stages a public evidential contest resolved when "the fire of the LORD fell" (1 Kings 18:38-39 (bib)).

Leading proponents

• Joseph Butler — probabilistic framework, cumulative-case method, deflation of the "peculiar presumption" (in corpus).
• William Paley — direct anti-Humean argument plus the positive historical case (in corpus).
• Adams, Leland, Campbell, Douglas, Price — Hume's first wave of critics, who anticipated most modern points (SEP §3.3; via SEP only).
• John Earman {{PROFILE-PENDING}} — agnostic philosopher of physics; sharpest contemporary formal critic of Part 1 (via SEP).
• Richard Swinburne — "by far the most sophisticated and elaborate development" of the Bayesian cumulative argument (SEP §2.2.4).

Strongest counter-arguments

First, the independence problem. Babbage-style results assume conditionally independent witnesses; Ahmed argues this assumption "is plausibly always violated," so stacking witnesses does not multiply evidential force as advertised (SEP §2.2.4).

Second, sincerity is not accuracy. The martyrdom argument "proves at best only the sincerity of the witnesses" (SEP §3.1.2); Hume's second Part-2 argument explains how enthusiasm, vanity, and holy-cause perseverance generate sincere false testimony (Hume 1748, §X pt. 2, ¶93).

Third, the minimal-facts base is softer than advertised. The SEP finds that scholars cited by Habermas as conceding the disciples' experiences (Perrin, Sanders, Dodd) grant "something a good deal more minimal" than the argument needs — Dodd calls tangibility and eating "apologetic expansions" (SEP §2.2.3).

Fourth, underdetermination of the cause. A certified anomaly does not certify divine agency: Baden Powell distinguished the extraordinary fact (matter for testimony) from divine causation (matter of opinion) (SEP §4.1). Migliorini sharpens it: an inexplicable healing wrought by hidden, benevolent, technologically superior aliens would satisfy the same explanatory constraints, so "from an incomprehensible event we can only infer the existence of an unknown cause," and "belief in miracles seems reasonable if one already believes in God, not vice versa" (Migliorini 2025, pp.3-4).

Fifth, the fideist dissolution. Ellis's Hume-Wittgenstein reading contends the evidentialist enterprise mistakes what religious belief is: "if there were evidence, this would in fact destroy the whole business [of religion]" — miracles change persons, they do not tip probability scales (Ellis 2024, pp.258-260). A cousin of this objection stands inside the Christian camp: Reformed epistemology denies that warranted Christian belief must rest on evidence — Alvin Plantinga, as Ellis notes, is willing to affirm a supernatural instigation of belief (Plantinga 2000: 258, cited at Ellis 2024, p.259).

Responses

Evidentialists reply (i) that even Ahmed grants a Babbage-style critique survives the failure of strict independence "provided that the ratio of the likelihoods does not converge to 1 too quickly" (SEP §2.2.4); (ii) that Adams answered the sincerity objection — in the apostles' case "the nature of the facts attested precludes the possibility that the witnesses are themselves deceived": empty tombs and shared meals are not inner impressions (SEP §3.1.2); (iii) that against parity, Adams's asymmetry holds: "There is a wide difference betwixt establishing false miracles, by the help of a false religion, and establishing a false religion by the help of false miracles" (SEP §3.2.4); (iv) that underdetermination is met by context — the "very description of the event — and even more, of the context in which it occurs — might render any naturalistic alternatives non-starters" (SEP §4.1) — and the theistic background Migliorini demands is what the wider natural-theology cumulative case supplies; the argument from miracles was never meant to run alone (SEP §4.2).

Assessment

Assessment: Strong — the in-principle Humean bar is now widely judged to fail, and the Butler-Paley probabilistic framework has been vindicated in its Bayesian modernization; but the positive case for any particular miracle remains contested at the level of historical detail, and its force is openly conditional on background theism.

Law-Conception Revisionism

Abstract

A third family of positions attacks the presupposition shared by the first two: that a miracle is a violation of a law of nature. On older and newer conceptions alike — Thomas Aquinas's "event exceeding the productive power of nature," Swinburne's "non-repeatable counter-instance," J. L. Mackie's "world not left to itself," and the dispositionalist metaphysics revived in contemporary philosophy of science — nothing law-like is violated when God acts in nature. The payoff is deflationary in both directions: miracles cease to be conceptually at war with science, but they also cease to be automatic trump cards, since what counts as a miracle becomes hostage to one's metaphysics of laws.

Formal statement

1. A miracle, classically, is an event that exceeds the productive power of nature (Aquinas, SCG 3.103; ST 1.110 art. 4, as cited at SEP §1.1) — the concept "predates any modern concept of a natural law by many centuries" (SEP §1.2).
2. On every serious account of laws — regularity, necessitarian, dispositionalist, ceteris paribus — the notion of a "violation" is either incoherent or dispensable.
3. Laws describe what nature does absent outside interference: "The laws of nature... describe the ways in which the world — including, of course, human beings — works when left to itself, when not interfered with. A miracle occurs when the world is not left to itself" (Mackie 1982: 19-20, quoted at SEP §1.2).
4. Therefore divine action in nature contradicts no law, and the Humean "proof from the nature of the fact" never gets started; the credibility of miracle reports becomes an ordinary, case-by-case evidential question.

Key evidence / textual basis

The SEP catalogues the trouble with violation-talk: it is hard to say which law a resurrection violates ("it is not, in the ordinary scientific use of the term, a law of nature that dead men stay dead" — the operative laws sit at the biochemical level); on Hume's own regularity view an apparent violation would show only "that what we had thought was a natural law was, in fact, not one"; and on necessitarian views violations of necessary relations are metaphysically problematic (SEP §1.2). Swinburne's repair defines a miracle as a non-repeatable counter-instance: where a putative law has broad scope and simplicity, "it may be more reasonable... to retain the law... and to accept that the event in question is a non-repeatable counter-instance of that law than to throw out the law" (SEP §1.2).

The dispositionalist program gives the view contemporary metaphysical depth. Page argues that "laws of nature are not external to the objects they govern, but instead should be thought of as reducible to internal features of properties": particulars are "internally powerful rather than being governed by external laws of nature, making external laws in effect ontologically otiose" (Page 2015, p.1). God then governs not through decreed regularities but through concurrence with creaturely powers. The test case is the fiery furnace of Dan 3:27 (bib) — "the fire had not had any power over the bodies of those men": on concurrentism God need not overpower or rewire the fire; he withholds the cooperation without which no created cause acts, so that, quoting Molina, "it was only because God did not concur with the fire in its action that the young men were not incinerated by it" (Page 2015, pp.8-9). Miracles contra naturam thus require no "violation" at all — a line Page traces through Aquinas and Suárez (Page 2015, pp.7-10).

Rockwood supplies the epistemological bridge: on the Mackie-style non-intervention conception, "we could consistently maintain both that a miracle occurred and that it is an exception to a genuine law of nature," and miracles "can no longer be ruled out by an a priori argument" (Rockwood forthcoming, p.16).

Leading proponents

• Thomas Aquinas — miracles as events exceeding nature's productive power, within a concurrentist account of divine action (via SEP §1.1 and Page).
• Richard Swinburne — non-repeatable counter-instance definition (via SEP §1.2).
• Ben Page {{PROFILE-PENDING}} — dispositionalist metaphysics of laws applied to divine governance and miracle (in corpus).
• J. L. Mackie — atheist critic whose "world not left to itself" definition is nonetheless the standard non-violationist formulation (via SEP §1.2).

Strongest counter-arguments

First, redefinition changes little evidentially. With laws Mackie-style redefined, the violation definition "becomes virtually equivalent to the earlier definition of a miracle as an event that exceeds the productive power of nature" (SEP §1.2) — and Hume's Part-2 arguments about religious testimony survive untouched, since they never depended on the word "violation."

Second, the Coleman-Vanderburgh line: an event with no ready natural explanation is not thereby an event with no natural cause; "past experience shows that what are at one time considered violations of natural laws are frequently found at some later time not to be so," so the reasonable course is to revise the law or suspend judgment rather than infer supernatural agency (SEP §3.1.2). A view that softens laws to admit exceptions makes every anomaly a candidate for future naturalization.

Third, necessity cuts back. Dispositional essentialism makes laws metaphysically necessary given the properties instantiated ("laws of nature are the laws of natures" — Oderberg, quoted at Page 2015, p.14); critics such as Jaeger press that a God who cannot alter the causal roles of the properties he creates looks like a diminished sovereign.

Fourth, the agency-attribution gap remains: Chryssides and Dawes argue that assigning an anomaly to a particular agent presupposes predictability that miracle claims lack (SEP §4.1); Migliorini's alien scenario applies with full force to non-violationist miracles (Migliorini 2025, p.3).

Responses

Revisionists reply (i) that dissolving the in-principle conflict is the point: the question becomes empirical and contextual, and Larmer's survey of in-principle objections finds "all of them fail to deliver the promised conclusion" (SEP §3.1.2); (ii) that on a ceteris paribus conception, apparent counterevidence may leave a law's probability nearly intact while raising the probability that "all else is not, in the present case, equal" — "everything depends on the details of specific cases" (SEP §3.1.2); (iii) that sovereignty is enhanced, not diminished: "a created world with genuine secondary causation... rather displays [God's power] more fully than would a universe of impotent beings" (Shanley, quoted at Page 2015, p.10); (iv) that Paley had already answered the theological objection that a wise God would not "violate" his own laws — authenticating a revelation makes discriminating divine action not impious but inevitable (SEP §3.1.1).

Assessment

Assessment: Live — the violation definition is in retreat in the philosophical literature and dispositionalist/concurrentist accounts are gaining ground; but the view redistributes rather than settles the evidential question, and its apologetic force depends on background metaphysics few disputants share.