Can We Rationally Believe in Miracles?

Miracles

“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.”

David Hume, An Enquiry Concerning Human Understanding

Hume’s sentence has done more work against miracles than every other argument combined, and it is worth noticing how economical it is. He does not claim that miracles are impossible, or that he has examined the reports and found them wanting. He offers a rule of evidence which, if sound, disposes of every miracle claim in advance and without inspection — because the falsehood of testimony is never as surprising as a suspension of the order of nature.

A good deal of what passes for the reply to Hume is not worth much. Most reported miracles really are false; credulity is real, and pious exaggeration is real, and the honest believer should say so before he says anything else. Any argument that requires taking every wonder story at face value is not going to survive.

But the rule itself has a difficulty, and it is not a theological one. It was pressed hardest in recent years by a philosopher of science who does not believe in miracles at all, in a book whose title states his verdict without ceremony: Hume’s Abject Failure. The question this page takes up is therefore narrower than it looks. Not whether any particular wonder happened, but whether we are entitled to decide that none did before looking at any of them.

Let’s examine the rule first, and then what would count as evidence, and ask …

  • If uniform experience is what rules miracles out, how do we know experience is uniform without first assuming every report of an exception is false?
  • And would the same rule, applied consistently, have let us believe the first reports of stones falling out of the sky?

I invite your engagement with me on these questions. You may leave comments below, but please be sure to read our policy on commenting before doing so.

Atheist/Skeptic View

The skeptic follows Hume. A miracle is a violation of a law of nature, and those laws rest on the entire uniform experience of mankind. Set against that weight there is only human testimony — and testimony fails constantly, from honest error, enthusiasm, faulty memory and outright fraud. So the wise man proportions his belief to the evidence, and the evidence for the regularity of nature will always be heavier than the evidence of a few witnesses to its breach.

What They Affirm

The balance can never come out the other way. To accept a miracle report, its falsehood would have to be more improbable than a suspension of natural law — and nothing is. That witnesses lie, misremember, or believe what they hope is among the most ordinary facts about human beings. That the dead rise is not ordinary at all.1

The circumstances of miracle reports are suspicious. They come chiefly from places and periods where checking was impossible, they multiply in conditions of religious excitement, and they attach to the founding of movements that benefit from them. People love the marvellous, and passing on a wonder story costs the teller nothing while gaining him an audience.2

Rival miracles cancel one another. Every religion has them, and they are offered as proof of contradictory doctrines. The miracles of Islam, of Hinduism, of pagan antiquity, and of Christianity cannot all be evidence for the systems they support. In crediting one and dismissing the rest, the believer is applying a standard he has not justified.

Extraordinary claims require extraordinary evidence. The principle is not hostility, it is proportion; we apply it to everything. Nobody accepts a new particle on the word of two witnesses. The evidence offered for the central Christian miracle is a handful of ancient documents by interested parties, which would not establish a disputed property boundary, let alone the suspension of death.3

And nothing survives investigation. Wherever a claim can be examined under controlled conditions, it dissolves. Studies of intercessory prayer show no effect. Lourdes has received millions of pilgrims across a century and a half and its own medical bureau has certified about seventy cures — a rate indistinguishable from spontaneous remission, which is what one would expect if nothing were happening. The miraculous is simply the not-yet-explained, retreating as explanation advances.4

Endnotes
  1. David Hume, An Enquiry Concerning Human Understanding (1748), §X, “Of Miracles,” part 1, where the maxim quoted at the head of this page appears. ↩
  2. Hume, §X, part 2, on the passion of surprise and wonder, on the concentration of such reports among what he calls ignorant and barbarous nations, and on the interest of religious parties in propagating them. ↩
  3. The maxim is Laplace’s in substance and Carl Sagan’s in the familiar wording; see Sagan, Cosmos (Random House, 1980), ch. 12. ↩
  4. On intercessory prayer, see H. Benson et al., “Study of the Therapeutic Effects of Intercessory Prayer,” American Heart Journal 151 (2006). The Lourdes figure is the Bureau des Constatations Médicales’ own count of cures recognized as medically inexplicable. ↩

Christian View

The Christian reply is not that Hume undervalued the evidence. It is that his rule, taken strictly, is not a rule of evidence at all but a way of never having to look at any — and that the objection has been pressed most sharply not by apologists but by philosophers of science who have no interest in defending miracles. What follows is about the argument, not about any particular wonder.

What Is Granted

Most reported miracles are not miracles. Credulity is real, pious fraud is real, and the appetite for a marvel is one of the most reliable things about human nature. Whole categories of claim — weeping statues, faces in plaster, healings announced from a stage — deserve the flat skepticism they get, and a Christian who defends them indiscriminately has made the skeptic’s argument for him.

The Church is in fact stricter than her critics assume. Lourdes has recorded roughly seven thousand claimed cures and certified about seventy, which is a rejection rate of ninety-nine per cent, imposed by a medical bureau that has included non-Catholic and unbelieving physicians. That figure is quoted against the Church as evidence that nothing happens. It is at least as good evidence that the filtering is severe.

A Miracle Is Not a Broken Law

The definition is doing hidden work. If a miracle is by definition a violation of a law of nature, then laws of nature are by definition exceptionless, and the question is closed by the dictionary before the evidence arrives.

But that is not what a law of nature states. A law says what happens in a system left to itself; it does not say the system is never interfered with, and no amount of observation could establish that. C. S. Lewis put it as a piece of arithmetic: if I leave money in a drawer and come back to find it gone, the laws of arithmetic have not failed. They told me what would be there if nothing intervened, and something did. The laws are what make the interference detectable in the first place. On this account a miracle is not a suspension of the order of nature but a new event fed into it, after which nature carries on exactly as before — which is why the Christian expects the world to be regular, and would have nothing to report if it were not.

The Circle in “Uniform Experience”

The heart of Hume’s case is that uniform experience stands against every miracle. Ask how the uniformity was established. Not by examining all human testimony and finding it false, because he explicitly declines to do that; the argument is offered as a shortcut that makes the examination unnecessary. So the experience counts as uniform only if the reports of exceptions are set aside in advance — and they are set aside because experience is uniform.

This is not a Christian complaint about an unfriendly conclusion. John Earman, a philosopher of science and no believer, devoted a book to the argument and titled it Hume’s Abject Failure, holding that the reasoning is confused, that Hume lacked the probability theory the case would need, and that his eighteenth-century admirers have been quoted ever since without being checked. When the most thorough modern critic of the argument is an unbeliever, its authority is doing more work than its logic.

The Rule Proves Too Much

Test the maxim on a case where we know the answer. For most of the eighteenth century, educated opinion held that stones do not fall from the sky. Peasants reported them; the reports were dismissed as superstition, and the French Academy of Sciences declared against them, one committee concluding that the stones had been struck by lightning rather than fallen. The testimony came from uneducated people in rural places, it described something contrary to uniform experience, and it was rejected on exactly Hume’s grounds.

The peasants were right. After the fall at L’Aigle in 1803, investigated by Jean-Baptiste Biot, meteorites entered science. Nothing about the reports had improved; what changed was that somebody examined them instead of ruling them out. A principle that would have kept us wrong about meteorites is not a principle for distinguishing truth from falsehood. It is a principle for protecting whatever we already think.

Rival Miracles Do Not Cancel

The cancellation argument assumes that all miracle claims are of the same kind and equally supported, and they are not. Many are not public events at all; many are attested centuries after the fact by a single source; a great many are the sort of thing this page has already conceded. Claims have to be weighed individually, and weighing them individually is precisely what the Humean rule forbids.

The logic also fails on its own terms. False sightings of a person do not show the person never existed, and rival claims to have found a cure do not show that nobody is ever cured. That two witnesses contradict each other is a reason to examine both, not to dismiss the pair unheard.

What Would Actually Count

The right question is not whether miracles happen in general but whether any particular one did, and that is answerable in the ordinary way: is the event public, is it datable, is it attested early and by people with something to lose, is there a natural explanation that fits the whole of the evidence rather than part of it? These are historical questions with historical answers.

Which is why the Christian case does not rest on wonders in general. It rests on one event, in a known place, under a named Roman governor, proclaimed within a few years in the city where it was said to have happened, to people in a position to contradict it. Whether that evidence holds is the subject of the companion page on the Passion and of the case for Christianity in particular. The point here is narrower: nothing in Hume entitles anyone to refuse to look.

The Reply in Brief

The definition begs the question, since a law of nature tells us what happens when nothing interferes and cannot tell us that nothing ever does. The appeal to uniform experience is circular, because the uniformity is obtained by discarding the very reports at issue. The rule proves too much, since applied consistently it would have kept the Academy of Sciences wrong about meteorites for another generation. And the cancellation of rival miracles works only if claims are never weighed individually, which is what the rule was designed to avoid.

None of that shows a miracle has occurred. It shows only that the question is open, and that it is a question about evidence rather than about the dictionary. If there is a God, the order of nature is His and He is not trespassing in it. Whether He has ever done anything unusual there is not settled by insisting in advance that He cannot have.

The full essay takes Hume’s section apart in order, sets out Earman’s Bayesian reconstruction of why the argument fails, examines Craig Keener’s two-volume survey of modern eyewitness reports and what can and cannot be made of them, works through the Lourdes medical bureau’s criteria and several certified cases, and takes up the question of what a genuine miracle would be for — why signs are given at all.

Summary

Hume’s argument is not that the evidence for miracles has been examined and found wanting. It is a rule which, if sound, makes examination unnecessary. That is its appeal, and it is also the difficulty with it.

Most reported miracles are false, and the Church agrees. Credulity and pious fraud are real, and a defense that takes every wonder story at face value deserves to lose. Lourdes has recorded some seven thousand claimed cures and certified about seventy — a rejection rate of ninety-nine per cent, applied by a medical bureau that has included unbelieving physicians. That figure is cited as proof nothing happens; it is at least as good proof that the filtering is severe.

A law of nature is not a claim that nothing ever interferes. Define a miracle as a violation of natural law and the case is closed by the dictionary. But laws describe what happens in a system left to itself. As C. S. Lewis put it, if money vanishes from a drawer, arithmetic has not failed — it told you what would be there if nothing intervened, and something did. On that account nature is not suspended; a new event is fed in, and nature carries on.

“Uniform experience” is circular. Experience counts as uniform only if the reports of exceptions are set aside first, and they are set aside because experience is uniform. Hume explicitly offers the argument as a shortcut that spares us examining the testimony — which is exactly the step that would be needed to establish the uniformity.

The sharpest critic of the argument is not a believer. John Earman, a philosopher of science, devoted a book to it and called it Hume’s Abject Failure, holding the reasoning confused and the probability theory absent. When the most thorough modern demolition comes from an unbeliever, the argument’s standing is resting on its reputation.

The rule would have kept us wrong about meteorites. Eighteenth-century science held that stones do not fall from the sky. Peasants said otherwise; the reports were dismissed as superstition and the French Academy declared against them — uneducated witnesses, rural places, testimony contrary to uniform experience. After L’Aigle in 1803 the peasants turned out to be right. Nothing about the testimony had improved; someone examined it instead of ruling it out.

And rival miracles do not cancel. The claims are not all of one kind or equally supported, and false sightings of a man do not prove he never existed. Contradictory witnesses are a reason to examine both, not to dismiss the pair unheard — which is what weighing claims individually means, and what the Humean rule was built to avoid.

In short: none of this shows that a miracle has occurred. It shows that the question is open, and that it is a question about evidence rather than about definitions. If there is a God, the order of nature is His and He is not trespassing in it. Whether He has ever done anything unusual there is a matter for looking, and looking is the one thing the argument was designed to make unnecessary.

For a more complete essay on this topic, download Can We Rationally Believe in Miracles? (.docx)