Most people today who start from the premise that miracles don’t or won’t happen knowingly or unknowingly depend on the influence of Scottish philosopher David Hume (1711-1776).

Hume did not originate the key ideas in his essay on miracles; most are recycled from arguments of some earlier deist writers, as Robert M. Burns has demonstrated (The Great Debate on Miracles: From Joseph Glanvill to David Hume; Lewisburg, PA: Bucknell University Press, 1981). It was Hume, however, whose influence mainstreamed these ideas so that some subsequent thinkers simply took for granted that he had established the case. Many thinkers from his own time forward offered strong responses to his case, including more sophisticated challenges based on mathematical probabilities, but Hume’s reputation in other areas lent credibility to his argument on this one.

Today scholars have published major academic critiques of Hume’s work. Oxford philosopher Richard Swinburne has been influential (The Concept of Miracle; London: Macmillan, 1970), and more recent critiques include J. Houston’s Reported Miracles: A Critique of Hume (Cambridge: Cambridge University Press, 1994), David Johnson, Hume, Holism, and Miracles (Ithaca, N.Y.: Cornell University Press, 1999), and John Earman’s Hume’s Abject Failure: The Argument Against Miracles (Oxford: Oxford University Press, 2000). Against criticism that Earman critiqued Hume’s argument because of Christian bias, Earman replied that he is not a Christian; he simply thought Hume’s argument was a poor one.

Violating Nature

Scholars reconstruct Hume’s argument in various ways, though Burns is probably right that we should fill the lacunae based on Hume’s assumptions of then-current deist debates. At the fundamental level, Hume’s argument is twofold: miracles violate natural law, and we lack credible eyewitnesses for miracles. In sum (acording to the most common understanding of Hume’s argument), miracles contradict uniform human experience.

The second part of his argument (the lack of credible eyewitness support for miracles) is probably meant to support the first part: lack of experience of miracles points to the ordinary course of nature (or, Hume would say, the uniform course of nature). Hume is trying to use induction to establish a negative, deductive argument—an argument that does not fit even his own normal approach. Hume normally did not believe that a finite number of examples could establish with certainty that something would always be the case—except when it came to miracles. (He could argue that it is improbable based on his circle of evidence, but his sample size proves too limited, as we shall see.)

Modern conceptions of natural law tend to be more descriptive than prescriptive, but Hume’s conception of natural law did not even fit the dominant paradigm of his day. Newton and his early followers were theists who affirmed biblical miracles; they did not regard God, the Legislator, as subject to his own laws. For Hume to argue that we cannot expect miracles because a God could not or would not “violate” natural law is an assumption, not an argument. It assumes without argument what no Christians believed anyway: a God subject to natural law. Defining miracles as “violations” of natural law lends the impression that God breaks such laws when he acts in nature; but this requires one to assume an uninvolved creator (as in deism) or no God at all.

A human who act in nature, by, for example, catching a falling object, does not “violate” the law of gravity; persons can act within nature without violating it. Why must God be less an actor than human persons? Moreover, most biblical miracles do not even fit a tamer definition of miracle that requires an action without nearer (as opposed to more distant) natural causes: when God used a strong east wind to blow back the sea in Exodus 14:21, the proximate cause was the east wind, and Moses and his rod functioned as agents, even though God was the ultimate cause.

No Credible Witnesses

The second part of Hume’s essay, probably meant to support the first half, is particularly problematic. To argue that uniform human experience absolutely excludes miracles, one must have comprehensive knowledge of uniform human experience. Instead, Hume argues that there are no credible eyewitnesses for miracles, but circularly uses the uniformity of human experience to challenge the credibility of witnesses. By almost everyone’s definition of miracles (as opposed to less conspicuous divine activity) they are not part of nature’s ordinary course; we don’t call them “miracles” when they are our common, easily predictable experience. But in some kinds of circumstances, what we consider ordinary is not ordinary: in black holes and cases of superconductivity, physical laws appear different than under many other conditions, inviting broadened definitions of overarching laws. If we do not a priori rule out the possibility of special divine activity, it would be rational to even expect special experiences during such activity.

Various subsidiary arguments inform Hume’s argument against reliable eyewitnesses. These arguments help him to narrow the field of evidence that should be acceptable, excluding testimony from nonwhite peoples and from antiquity. He excludes, for example, claims from non-Western and nonwhite civilizations. Hume considers such peoples “ignorant and barbarous,” fitting his ethnocentrism in his other work. One could elaborate at length on his ethnocentrism, e.g., his denial of any truly great achievements in Asian and African civilizations, his widely-used support for slavery, and so forth. See e.g., C. L. Ten, “Hume’s Racism and Miracles,” Journal of Values Inquiry 36 (2002): 101–7; Charles Taliaferro, and Anders Hendrickson, “Hume’s Racism and His Case against the Miraculous,” Philosophia Christi 4 (2, 2002): 427–41; and my “A Reassessment of Hume’s Case against Miracles in Light of Testimony from the Majority World Today,” Perspectives in Religious Studies 38 (3, Fall 2011): 289–310.

Since all religions claim miracles at the beginning, he mistrusts miraculous claims from the beginning of religions. Hume’s target here is fellow Enlightenment thinkers, such as John Locke, who used early Christian miracles as evidence for Christian faith. But Hume is not correct that all religions claim miracles at their beginning, nor would the claims of some religions automatically cancel out those of others, any more than the discrediting of one witness for a case would discredit all the witnesses. (Moreover, Hume merely presupposes, with some of his contemporaries, that religions’ claims are mutually exclusive, so that genuine superhuman activity could not occur in more than one.) Excluding testimony in religious contexts presupposes what it would hope to prove.

To be credible, Hume believed, eyewitnesses must be educated, socially respectable Western white persons such as Hume and his circle; he avers that only such people have something to lose by lying. Today, of course, I can cite numerous witnesses who meet all his criteria, including medical doctors, philosophers, and plenty of fellow PhD’s. Not all of the witnesses began as Christians before the events they claim to witness, contrary to suspicion of religious bias (as if bias is endemic only to persons with religious convictions; as a former atheist, I can attest firsthand that bias is not limited to a single ideology).

A particular case allows us to understand more concretely how Hume might apply his criteria. Hume takes an example from then-recent history: Marguerite Perrier, niece of the famous mathematician and theologian Blaise Pascal, had a long-term, organic fistula in her eye that emitted a foul odour, seemed to accompany bone deterioration, and separated her from her peers because of the smell. She was instantly and publicly healed when touched by a relic (and few of us today would defend the relic’s authenticity), and the Queen Mother of France sent her own physician to examine this event. Hume points to this experience, noting that it was public, widely attested, even medically verified. It is far better verified than biblical miracles. Yet, he says, we do not believe this account; so why should we believe any other?

And then Hume moves on. He offers no argument; he simply takes for granted that no one will defend this account. Why? The setting in which Marguerite Perrier was healed was the early Jansenist movement, and nobody liked Jansenists; they were too Augustinian for French Jesuits, and, more to the point of Hume’s primary audience, they were too Catholic for Anglicans and Presbyterians. His Christian contemporaries who were accustomed to dismissing each others’ miracle claims without contrary evidence would not argue Hume’s point. But what if their sectarian dismissals were premature?

Challenging Hume’s Argument Today

David Hume was a smart man, and I do not believe that if he were around today, even he would argue his case the way he did in his day. (Admittedly, that is a postHUMous argument—sorry for injecting a bit of HUMour here.) It was one thing to deny credible eyewitness claims when the available sample size was so limited, and when most of his largely Protestant context relegated miracles to the distant past.

It would be a quite different to dismiss miracle claimants’ credibility a priori, or to make claims about uniform human experience, if there were millions of people who claimed to be witnesses. This is especially the case if one does not exclude witnesses based on sectarian or ethnic considerations.

Today, in fact, we have a fuller knowledge of global human experience, or at least the claims about such experience, and we can readily say that hundreds of millions of people claim to have witnessed or experienced divine healing. No one would argue that all of these claims represent genuine miracles, much less that they can be explained only in this way. But with hundreds of millions of claimants, it is simply not possible logically to start with an a priori claim about human experience on the matter being uniform.

In 2006, a Pew Forum survey of ten countries (representing most continents, including North America) interviewed Pentecostals, charismatics, and Christians who claimed to fit neither category. Given the percentages in the 231-page report’s executive summary, it appears that hundreds of millions of people in these ten countries alone (i.e., not even including other countries) claimed to have witnessed divine healing. Nor are such claims limited to one religion, although other surveys show millions of people with centuries of non-Christian background converting to Christianity, often despite great social pressures to the contrary, because of extraordinary miracles in Christian contexts. A 2004 survey of U.S. physicians reports that over half believed that they had witnessed miracles during their practice. (We can keep in mind here that those with scientific training tend to define miracles more narrowly and rigorously than do many others.)

My own sample size of hundreds of sources is more limited, but from written sources and my own interviews, I conclude that many of these cases are significant. They include most of the range of miracles reported in the New Testament, including instant disappearances of blindness, resuscitations from apparent (and sometimes clinically documented) death, the instant vanishing of goiters, and the like. Again, some of these are medically documented. Although I initially collected such accounts much more deliberately for my book on miracle accounts (Miracles: The Credibility of the New Testament Accounts [2 vols.; Grand Rapids: Baker Academic, 2011]), I have since come across many more, and often with fuller documentation than available at the time.

I will not digress from my main point here to elaborate examples, but I conclude by reinforcing the point of my brief response to David Hume here. Hume’s a priori dismissal of credible eyewitness support for miracles, and thus his argument from the uniformity of human experience and nature, does not work in a twenty-first century context. That is not to say that Hume might not have tried to argue against miracles from a different standpoint, or to seek other ways to counter his contemporaries’ apologetic use of biblical miracle claims. It is to say that the case that Hume argued, on which most modern assumptions that dismiss miracles are based, is no longer logically tenable.

About the Author


Dr Craig Keener

(PhD, Duke University) is F. M. and Ada Thompson Professor of the New Testament at Asbury Theological Seminary. He is especially known for his work as a New Testament scholar on Bible background (commentaries on the New Testament in its early Jewish and Greco-Roman settings). His award-winning, popular-level IVP Bible Background Commentary: New Testament (now in its second edition [2014], and available in a number of languages) has sold over half a million copies. This article on miracles first appeared on his blog in two parts, here and here and is used with his kind permission.