Is It Irrational to Believe the Testimony of the Miraculous?
A Critique of David Hume

By Jake Magee

In this article I examine and challenge David Hume's contention that it is in principle always irrational to believe reports of the miraculous.

David Hume argues that it is never rational to believe in the testimony of a miraculous event (Earman 141). He believes that this conclusion follows from a principle he holds to be true: the prior improbability of an event, if great, defeats the probability that the witness of the event is telling the truth. Put in general terms, testimony is always subservient to induction. As applied to miracles, since the probability of a miracle happening is by definition astronomical, we should never believe the report of someone who testifies of it. I endeavor to show that Hume's project has failed because the principle upon which his argument rests is flawed. In particular, there are contexts in which the improbability of an event happening, though great, is defeated by the reports of witnesses. Assuming my argument to be sound, I will offer a context in which the improbability of a miraculous event is defeated by the reports of eye-witnesses. I will then show that all I need in order to diffuse Hume's critique is the possibility that this context obtains.

From the start of his essay, Hume makes it clear that he is seeking for a general principle that would enable him to rule out testimony of the miraculous without having to examine each report case by case. As long as the world endures, "so long...will the accounts of miracles and prodigies be found in all history, sacred and profane" (141). To avoid the indefinite toil of examining each account, Hume claims that he has found an argument that will be "an everlasting check," effective against any claim of the miraculous (141). To further buttress this interpretation, Hume claims elsewhere that his principle, when applied to religions that are founded on the testimony of the miraculous, "amounts to an entire annihilation" (46). At the end of Part 1, Hume summarizes his position by stating that "no testimony is sufficient to establish a miracle" (144).

What is Hume's argument that testimony is always subservient to induction? Hume starts with the platitude that a rational person will believe something on the basis of good evidence (141). What counts as being good evidence is connected to our own observations and experiences. Of this Hume states "experience is our only guide in reasoning concerning matters of fact" (141). A rational person, then, will believe something insofar as it is confirmed by his or her experiences. By "experiences," Hume seems to mean the objects of our sense perceptions. Thus, rational beliefs are confirmed by a person's sense perceptions. Since testimony is a "matter of fact," then it must also be subject to the scrutiny of experience or sense perception. As Hume puts it, "the ultimate standard, by which we determine all disputes, that may arise concerning them (i.e. testimonies), is always derived from experience and observation" (142). Consequently, testimony is intrinsically subordinate to induction.

In order to establish his thesis, it is crucial that Hume demonstrate how probabilities are handled by a rational person. To this end, he emphasizes the various probabilities that attend to various beliefs. To adapt an example used by Hume (145), an Indian prince has observed that the sun, when visible, always travels across the sky uninterrupted. This phenomena is invariable. He has also observed that usually there is better weather in July than in January. However, this phenomena has exceptions. On the other hand, the prince has never observed water changing into ice, something reported to him by an emissary of a distant country. This phenomena is invariable. Since the prince is rational, he knows that there is a broad spectrum of assurance that can attend to his reasoning concerning matters of fact. Moreover, he always proportions his assurance for any belief to the degree of regularity with which he observes those things. According to Hume, what conclusions does he draw from these three phenomena?

As to the sun's uninterrupted course, the prince has the highest assurance that is allowed for in terms of probability. He concludes that this phenomena will never vary, for it is a law of nature. As to the weather, the prince observes that the number of instances of better weather in June is greater than the number of counter-instances. So, his assurance for this phenomena is proportionate to the difference between instances and counter instances. As to the phase change of water to ice, the prince scrutinizes the event reported and the report with the final arbiter, experience. With respect to the alleged event, the prince might conclude that the phenomena of water not turning into ice is like the sun's invariable course, for he has seen no counter instance. With respect to testimony, experience demonstrates that it is more like the varying weather patterns than the immutable course of the sun. Most of the time people tell the truth, but there are plenty of counter-instances. Since the prince proportions his beliefs to the degree of regularity with which he has observed any phenomena, and he has never seen water change into ice but has observed faulty testimony, he will reject the report.

Given the relationship that Hume believes holds between experience and testimony, he proceeds to argue that it would be irrational for a person to believe any report of a miracle. A rational person believes someone's testimony insofar as the facts reported conform to his or her experience. Some kinds of experiences are unalterable, and as such are regarded as laws of nature (143). Miracles, however, are by definition violations of the laws of nature. By definition, then, the prior improbability of a miraculous event happening always swamps the probability that the person is reporting correctly. Therefore, a rational person would never believe the report of a miracle.
To make his case iron clad, Hume grants that it may be the case that someone's testimony of a miracle is of the highest quality, so that it might be considered an entire proof (143). In this instance we have the proof of testimony pinned against the proof of induction. But, as Hume has argued, the evidence of testimony is a species of the evidence of experience. Since uniform experience amounts to a "full proof," no report about an exception to the uniform experience should induce belief in that report. Just as a river never rises higher than its source, so testimony can never lead rational people to jettison uniform experience for it.

It is important to note that Hume believes that there is crucial difference between the miraculous and the marvelous. The sun altering its course would be miraculous, while the phase change of water to ice is regarded a marvel. The difference is that marvels are repeatable and covered under the laws of nature. Consequently, the prince could test the report, in principle, by traveling to the emissaries' land to see whether the phenomena happens. Hume thinks that miracles permit no conditions under which the event could be observed. Because of this difference, Hume suggests that there is testimony strong enough to warrant belief in the face of uniform experience when it comes to the marvelous (143). It's not that the probabilities are overwhelmed by testimony, it's just that we are ignorant of the true probabilities. When it comes to the miraculous, however, even testimony of the highest quality is insufficient. We might say that in the case of the phase change, the prince has a default-defeater. With the claim that the sun has altered its course, we have a clear-defeater.

While Hume seems to say that the probabilities of events matter in all contexts, I want to suggest that the context determines whether or not the probabilities matter. What I see, then, are two extremes that have been taken in this debate. One extreme is taken by Hume, in which the probabilities associated with an event's occurrence should always be taken into account. Price, reacting to Hume, errors on the other extreme: "that improbabilities as such do not lessen the capacity of testimony to report truth" (165). If my assessment is true, it may be the case that Hume has failed in his endeavor to provide a general principle which would settle all debates about particular claims to miracles.

Contrary to Price, it is clear that in some contexts we are rational to discard testimony given the improbability of the event reported. Let's say that Rick reports to his best friend Martin that he had been visited by aliens. With all sincerity and passion, he relates a story in which while traveling on a lonely desert road, he sees a strange luminescent object streaking erratically through the night sky, only to land next to his vehicle. Creatures with small bodies, big heads, and large dark eyes proceed from the ship, telepathically communicating that they have come from a distant galaxy for the sake of studying humanity. With this brief conversation, the aliens leave Rick alone on the highway.
Martin sees the passion and sincerity of Rick. He has good reason to think that Rick wouldn't lie, wouldn't jest, and wouldn't believe fanciful stories. Despite all of this, Martin concludes that the report is false. In this case, his background information makes the judgment sound. For example, being an astronomer, he has good grounds for the belief that the universe is the kind of place in which life-permitting planets are nearly impossible. But, even if two life-permitting planets existed, he knows the prospects that intelligent life would evolve on two planets in the universe is also highly unlikely. Furthermore, even if life did evolve on each planet, the probability that a species, separated by millions of light years, could both locate this other planet and travel this distance is astronomical. Considerations such as these justify Martin's refusal to believe Rick. Consequently, this justifies the notion that in some contexts, the improbability of an event happening should defeat the probability that the person reporting the event is telling the truth.

Secondly, there are other contexts in which we seem rational to believe the reports of people, despite the fact that the probability that the event happened is astronomical. Price asks us to consider a lottery. The chance that any particular individual will win "exceeds all conception" (164). Modifying Price's example, imagine that while at Starbucks, a customer reports to Conin the clerk that Betty Gamble from Redlands California won the jackpot. Though he has never known a lottery winner, and doesn't know the person reporting this information, he believes. Because he has never known a lottery winner, and doesn't know the reporter, he might think that there is a vast conspiracy orchestrated by the state to fool the populous into believing that there is a winner for the sake of generating revenue from the sale of lottery tickets. But, he doesn't. And though the chance that this patron of Starbucks is reporting falsely is far greater than the chance that Betty (of all people) won the lottery, Conin seems rational in believing the report. Hume seems to argue that the clerk is irrational.

Modifying this example further, Hume would argue that the clerk would not be rational to believe the report, even when there are good reasons for trusting the reporter. In this version, Conin is watching the news after his shift. The news anchors announce that Betty Gamble from Redlands California won the lottery. Being a part-time statistician, Conin knows that the odds that Betty would win is mind-boggling. But, he also knows that the accuracy of this news organization is 99.99%. Hume would argue that the improbability of the event happening overwhelms the probability that the news anchors are telling the truth, and so one must discard the report. Yet, in this case it looks rational for Conin to believe the reports (Craig 324).

If my examples are sound, then they seem to suggest that the context will somehow determine whether or not the probabilities of events should be taken into consideration. The question then is this: From these two examples, what is the feature in each context in virtue of which a belief is labeled rational or irrational? The answer seems to be the expectation born from one's background knowledge. In the case of the astronomer Martin, his background knowledge gives him grounds for thinking that aliens did not visit his friend, for the odds are against it. In the case of Conin, his background knowledge gives him assurance that someone will win despite the odds being against Betty winning. It's sort of common knowledge that people have won, and hence someone will win. But, certainly in the case of alien visitation there is no common knowledge that aliens do exist, and hence we have no expectation that we will be visited.

Since a certain kind of context determines whether or not the probabilities of events matter, what kind of context obtains when reports of miracles are given? Are reports of miracles like reports of alien visitations, or like reports of lottery winners?

Paley offers a context in which reports of miracles are like reports of lottery winners (41-42). He asks us to assume that there is a God who is quite interested in our well-being, not only for this life, but ultimately for the afterlife. Furthermore, we are asked to assume that we are free both to obey and disobey God's will. However, due to some flaw in humanity, we don't have the knowledge that we need in order to attain the well-being that God intends. Given all this, Paley thinks that it is reasonable that God would give revelation about this most important subject. And since a revelation requires miracles for its confirmation, then it is reasonable that we should expect miracles from God.

One might object that this does little to provide an actual context in which belief in miracles is rational. It might be the case that all evidence goes against Paley's assumptions. Imagine, after Martin gives Rick all the reasons why he doesn't believe his report of the alien visitation, Rick employs Paley's strategy: "Suppose that the universe is arranged in such a way that life-permitting planets are more probable than life-prohibiting planets. Suppose further that macro-evolution occurs all the time. Also, suppose that aliens could travel through wormholes to any place in the universe, and that they really want to talk to us. Therefore, it is reasonable to believe my report." Of course, this tactic would be futile, since it is in a way begging the question.

It is an objection like this that leads J.L. Mackie to distinguish between two different contexts in which the debate about miracles could take place. The first context might be one in which the individuals involved already accept some notion of theism. It may very well be rational for someone in this context to believe the reports of miracles. As Mackie puts it, "In this context supernatural intervention, though prima facie unlikely on any particular occasion, is, generally speaking, on the cards: it is not altogether outside of the range of reasonable expectation for these parties" (94). Put differently, Mackie is saying that reports about miracles are like reports about lotteries just in case the person is a theist.

The second context that Mackie addresses in one in which the truth of theism is at issue. Since it is only the existence of a particular kind of God that gives one an expectation of the miraculous, the miraculous can never serve as evidence for this deity in a context where the existence of this deity is at issue. As he puts the matter, "it is pretty well impossible that reported miracles should provide a worthwhile argument for theism addressed to those who are initially inclined to atheism or even to agnosticism" (95).

If I have understood Mackie correctly, it looks as if reports of the miraculous are like alien visitations for atheists and agnostics, and like lotteries for theists. Neither Hume, nor I, would want to accept this conclusion. It appears as strange as saying that Rick and Martin are both rational because they are operating with different background assumptions. Intuitively, I want to say that Rick can never be rational about this belief precisely because he operating from the wrong context. If the context is in error, so are the beliefs grounded in it. Resisting this notion, I want to find the context in which it would be irrational for an atheist or agnostic to view miracles like alien visitations. So the question is this: what must the theist provide, by way of argumentation, to illicit an admission from the atheist or agnostic that it may be the case that miracles are like lotteries (so that theist isn't unreasonable in believing a particular miracle report)?

At this point, Paley's reply to Hume gives us direction. As we have seen, Paley argues that Hume erred in not seeing that there might be a context in which miracles might be expected. The context Paley provides in one in which God exists and has certain attributes which would incline him to perform "wonders" on earth. Of this Paley says, "We do not assume the attributes of the Deity, or the existence of a future state, in order to prove the reality of miracles. That reality must be proved by evidence" (42). What "evidence" is Paley alluding to? It is plausible to read him as alluding to the evidence provided by natural theology (where natural theology refers to knowledge of God derived from consulting nature). For Paley, then, the context in which it would be irrational for an atheist or agnostic to view miracles like alien visitations in one in which natural theology is successful.

I would like to provide a schema in which natural theology might be successful. First of all, arguments provided from natural theology need not be demonstrable. All that is needed is plausibility, where this refers to an argument whose premises are more plausible than their negations (Craig 50). Secondly, the case from natural theology should be cumulative (180). The Kalam Cosmological Argument might make it plausible to believe that the universe was caused by a powerful agent. By itself, however, it is unable to hold the weight of miraculous expectation. By adding other plausible arguments, we might have a sufficient ground for believing that miracles are like lotteries. So, the argument from design might be offered in such a way that the powerful agent is seen to be really smart. Likewise, the ontological argument might make plausible the notion that this being is one, and perfect. The argument from morality might make it reasonable to think that this being is moral. In sum, natural theology is successful when a cumulative case, comprised of plausible arguments for theism, makes the expectation of the miraculous reasonable.

If there is such a natural theology available, it might in turn find support or confirmation by reports of miracles. If the theist has mounted a plausible case, the burden in the theist's case might shift from natural theology to historical investigation. She is able to look over the vast amount of data provided by history and determine whether there are any reports of miracles that stand up to reasonable criteria used for judging testimony. Among other things, this criteria includes the following: (1) the coherence and consistency of testimony; (2) the character, number, and independence of witnesses (Mackie 93); (3) the manner in which the testimony is delivered; (4) the implausibility of alternative explanations (Craig 125). If the theist were to find testimony of a miracle that met these criteria, and the miracle confirms revelation that attests to Paley's initial assumptions, the testimony would have the effect of reinforcing her conclusions about natural theology.

To illustrate how natural theology and testimony of the miraculous can mutually support each other, I appeal to an example given in the history of science. At the turn of the last century, two models of physics were in competition for acceptance. The long standing tradition of Newtonian Mechanics viewed space, time, and motion as absolute. The fixed frame was thought to be provided by an invisible medium called ether, through which light traveled and stars wandered (Barnett 41). Another tradition which was to emerge viewed space, time, and motion as relative, and light as absolute. On this view, there is no fixed frame, and hence no ether. Michelson and Morely, two American physicists, hypothesized the following: If ether exists and is a substance, then it can be measured. Since the earth moves through ether, there should be a disturbance as the earth moves through it. Using experiments with light, one should detect a variation in the speed of light given our motion relative to the ether. When Michelson and Morely performed their experiment, the speed of light remained the same. This was thought to be the critical experiment that forever sealed the case against Newtonian Mechanics and was later thought to secure Relativity Theory (41-44). The crucial experiment, however, was thought to be anything but decisive for the Newtonians. In fact, they simply theorized that the reason we don't detect variation in the speed of light is because the earth drags some fraction of ether with it. This portion of ether isn't moving, relative to the earth (Kuhn 73).

The apparent standstill facing the physicists of both camps was that the facts could be made to fit either theory. So, it looks as if each side could in principle preserve their theories by interpreting the data differently. However, there comes a point in which reinterpretation becomes ad hoc. At this juncture, one theory is implicated, while the other is vindicated. This is exactly what happened in the struggle between Newtonian Mechanics and Einsteinian Mechanics.

When it comes to the testimony given of the miraculous, there are two opposed background theories that conclude different things. One theory is that there is no God, or at least we can't know that there's one. As such, any evidence provided by historical investigation is interpreted within this framework. So, for any reported testimony, a proponent of this theory will argue that the data fits within his own schema and concludes that the testimony is in error. Another theory is that God exists. The same data is thought, by the theists, to fit far better with her theory. Presuming that both theism and atheism are plausible positions, it may be the case that one's interpretation of historical data is ad hoc. Now, if the testimony is the kind that fits the criteria mentioned heretofore, I want to suggest that the explanation given by the atheist or agnostic is ad hoc. If this is true, theism receives support for that which it in some way expects.

One of the biggest obstacles that Hume could raise against my proposal is to argue that I can never get the natural theology that I need without begging the question. To get all the attributes listed by Paley, and which are necessary for my proposal, I must rely upon what is already assumed as being revelation. It's from Scripture that I derive that there is a certain kind of God who intends for us to experience a blissful eternity, not from nature. In short, miracles are like lotteries only in the context in which Scripture is already assumed true, which presumes the validity of the miracles that attests to its truth.

In reply, I suggest that this is irrelevant to my case against Hume. As we have seen, Hume offers a general principle which he asserts will always doom reports of the miraculous to the realms of irrationality. His assertion is that, in all contexts, the improbability of an event defeats the probability that the person reporting is not deceived. Against Hume, I have argued that there are some contexts in which the improbability of an event is defeated by testimony. Furthermore, I suggested a context of theism, supported by natural theology, in which the same is true for the testimony of the miraculous. It isn't necessary for my thesis to show that this context actually obtains, but only that it could obtain.

Following what I think Paley was endeavoring to do, I'm arguing that all that is needed to defeat Hume is the possibility of a persuasive natural theology. This in turn would make our expectations for miracles legitimate. It may be the case that all the proofs for God's existence fail miserably. In this, all Hume could charge is that it is irrational to believe in the reports of miracles because the context needed to ground these reports hasn't obtained. Yet, I could imagine a world in which the argument from design is persuasive given that there is relatively little dysteleology in the world. I could imagine a world in which the moral argument is persuasive, given there is only small pockets of evils in the world. If such a world is possible, then Hume's principle is flawed.

If my assessment is right, then in reality Hume has begged the question in the presentation of his argument. Given his failure to recognize that the context determines whether the prior improbabilities of an event are to be regarded, he hasn't shown that natural theology doesn't provide the context needed by the theists. What he needs to make part of his case work is what is at issue. He seeks a way out of the toils of examining claims of the miraculous case by case, but this is precisely what he might need in order to say that it is irrational to believe the testimony of the miraculous (that is if natural theology turns out to be plausible).

Perhaps Hume might have recourse to the principled distinction that he makes between miracles and marvels. He suggests that a person may be rational to believe the reports of the marvelous, so long as the testimony was strong enough (143). The reason why Hume allows for testimony to defeat uniform experience in cases of the marvelous is because a subject could place herself in the situation to see whether or not the testimony is true. So, the Indian prince might be justified in believing that water changes into ice because he could put himself in the circumstance and observe if the contrary happens. In the case of the miraculous, Hume thinks that we have a clear-defeater, in that placing ourselves in the circumstance will yield only the uniform experience of the laws of nature.

In opposition to Hume, I want to argue that since there is no principled difference between miracles and marvels, Hume must allow for testimony of the miraculous that is strong enough to induce rational belief. Firstly, it's not quite clear what counts as being able "to place ourselves in the circumstances." Let's say, for example, that the prince has a horrible reaction to cold climates, in that his sensory faculties cease to operate. Conceptually, he is able to position himself in the circumstances, but because of his ailment it's of no use. So, instead of going himself, he sends his most trusted adviser to determine whether the emissary is telling the truth. Upon his return, the prince's adviser testifies with zeal and fervor that the water does turn to ice. Intuitively, it looks as if the Prince would be justified in believing the report.

What would Hume say?
Hume might say that the prince wouldn't be justified. But that seems overly restrictive. In science, for example, people depend upon instruments to "observe" what humans are unable to "observe." In the sub-atomic world, our sensory faculties cease to operate. Conceptually, however, it seems possible to see the microscopic world. To add to our knowledge of this realm, scientists use reliable and precise instruments to get a view of that world. Their knowledge is justified despite the fact that it is mediated by something else. In the same way, the prince seems to be justified in believing the report given by his most trusted and reliable adviser, even though the knowledge obtained is mediated. In Campbell's words, testimony looks like perception second hand (166).

Hume might argue that the prince would be justified because it is conceptually possible for him to put himself in the circumstance where he would observe the phenomena reported. In this move, however, the defender of miracles could argue that it is conceptually possible for us to go back in time and place ourselves in the context in which the miracles were alleged to take place. Furthermore, he might argue that just as scientific knowledge is mediated by reliable instruments, so historical knowledge is mediated by reliable procedures. In the rigorous application of the procedures available to historians, we are "placing ourselves in the circumstances" in which the miracles were alleged to take place.

If my assessment is correct, then Hume can have no clear-defeater for miracles. He argues that attending to all the dead bodies that remain dead in our life time is a clear-defeater that the resurrection of any man has not taken place (143). This conclusion follows, he thinks, because he has placed himself in the circumstance, and no resurrections occur. But I have charged that he hasn't placed himself in the circumstance, and thus can have no clear-defeater. Consequently, if a person can be rational in believing strong testimony of the marvelous, and there is no principled difference between the miraculous and the marvelous, so a person can be rational in believing the strong testimony of the miraculous.

In conclusion, I have argued that Hume has failed in his endeavor to find a general principle that would serve to always defeat the reports given of miracles. The principle that he offers suffers from counter-instances. In some contexts, it is rational to believe the testimony of events despite the events' inherent improbability. In other contexts, it is irrational to believe. This implies that context determines the rationality of belief in testimony. I have sought to sketch a context in which it would be rational to believe the reports of miracles. All that I need to make my case against Hume is the conceptual possibility that this context could obtain, not that it does. This then forces Hume to deal with whether such a context obtains, which contravenes his attempt to devise a principle that would avoid this dirty work.

Works Cited

Barnett, Lincoln. The Universe and Dr. Einstein. New York: Bantam Books, 1968.
Cowan, Steven.ed. Five Views On Apologetics. Grand Rapids: Zondervan Publishing House, 2000.
--Craig, William Lane. "Classical Apologetics" Cowen.
Earman, John.,ed. Hume's Abject Failure: The Argument Against Miracles. Oxford: Oxford University Press, 2000.
--Campbell, George. "A Dissertation on Miracles" Earman.
--Earman, John. "Hume on Miracles."
--Hume, David. "Enquiry Concerning Human Understanding" Earman.
--Price, Richard. "Four Dissertations" Earman
Kuhn, T. S. The Structure of Scientific Revolutions. Chicago: The University of Chicago Press,
Swinburne, Richard.,ed. Miracles. New York: Macmillan Publishing Company, 1989.
--Mackie, J.L. "Miracles and Testimony-Hume's Argument" Swinburne.
--Paley, William. "Evidences of Christianity-Preparatory Considerations" Swinburne.