A Systematic Look at the “Argument from Testimony”: David Hume’s Critique (Column 671)

With God’s help

Disclaimer: This post was translated from Hebrew using AI (ChatGPT 5 Thinking), so there may be inaccuracies or nuances lost. If something seems unclear, please refer to the Hebrew original or contact us for clarification.

In my debate with Yaron Yadan, and even more so at the end of my debate with Jeremy Fogel, we briefly touched on the Argument from Testimony, which deals with the reliability of tradition. (We later continued the discussion with Jeremy on WhatsApp.) As far as I recall, I have never treated the Argument from Testimony systematically here on the site. So I thought this would be a good opportunity to fill the gap. In this column I present Hume’s arguments; in the next one I will critique them.

“The Argument from Testimony” or “The Argument from Miracle”

Many religious thinkers—across traditions, not only in Judaism—rely on tradition that transmits to us various miracles and divine revelations. Some see tradition about miracles as evidence for the existence of God, while others point to tradition about revelation as a channel through which God’s demands upon us (conveyed in revelation) are transmitted. In The First Existent, in the fifth conversation, I presented a somewhat more nuanced picture, in which tradition plays a role on both planes. Nowadays this is called the “Argument from Testimony,” i.e., testimony that transmits information about miracles and revelation serves as evidence for their occurrence.

In Judaism, the Argument from Testimony begins in the Torah itself. Thus, for example, the Torah states (Deuteronomy 5:4):

“The LORD did not make this covenant with our fathers, but with us—even us—who are all of us here alive today. Face to face the LORD spoke with you on the mountain out of the midst of the fire.”

Or (ibid., 5:23):

“For who is there of all flesh who has heard the voice of the living God speaking out of the midst of the fire, as we have, and lived?”

Here it is not presented as an argument for the reliability of the testimony, but as an indication of God’s favor in leaving them alive after their encounter with Him.

The primary source is Deuteronomy 4:32–40:

“Indeed, ask now about the former days which were before you, from the day that God created man upon the earth, and from one end of the heavens to the other, whether anything so great as this has ever happened, or has anything like it been heard? Has a people heard the voice of God speaking out of the midst of the fire, as you have heard, and lived? Or has God ever attempted to go and take for Himself a nation from the midst of another nation, by trials, by signs, and by wonders, and by war, and by a mighty hand, and by an outstretched arm, and by great terrors—according to all that the LORD your God did for you in Egypt before your eyes? It was shown to you so that you would know that the LORD, He is God; there is no other besides Him. From heaven He made you hear His voice to discipline you, and on earth He showed you His great fire, and you heard His words out of the midst of the fire. And because He loved your fathers and chose their seed after them, He brought you out by His presence with His great power out of Egypt, to drive out nations greater and stronger than you from before you, to bring you in, to give you their land for an inheritance, as it is this day. Therefore know this day and lay it to your heart that the LORD, He is God in heaven above and on the earth beneath; there is no other. And you shall keep His statutes and His commandments which I command you today, that it may go well with you and with your children after you, and that you may prolong your days upon the land which the LORD your God gives you forever.”

Here the Argument from Testimony already appears in earnest. True, even here there is the addition “Has a nation… and lived?”, i.e., the emphasis is on the fact that we survived. But it seems from the whole passage that the Torah also intends the revelation and the miracles as testimony that the LORD is God.

The argument continues to be discussed throughout the history of Jewish thought (see here). Saadia Gaon presents it in his introduction to The Book of Beliefs and Opinions and in its third treatise; the better-known source is The Kuzari, in the first essay around section 20. In our day it is used extensively in outreach seminars. But the argument also appears in Christian thought and is discussed there at length (see here).

Many argue that it is implausible to persuade so many people of an event that never happened, and hence the power of the Argument from Testimony. There are differences among thinkers, especially Jewish ones, regarding which events the argument can be applied to. Some limit it to the Sinai revelation, while others use it for miracles generally (see here around notes 12–15).

Hume’s critique of the Argument from Testimony

Reliance on tradition has been attacked by secular and atheistic thinkers from several angles. Some point to contradictions within the relevant traditions. Others note that it is hard to accept a long-standing tradition as-is (the “telephone game” inevitably produces distortions), particularly when at least part of the chain involves ignorant and primitive generations. Above all, they argue that it is difficult to accept testimony about extraordinary phenomena, such as miracles or revelations. Traditions can be corrupted, and there may be traditions that began with the inadvertent or intentional incorporation of folktales, and the like. Sometimes positive evidence is even offered (from historical research of the Bible and beyond) that tradition is unreliable—certainly not precise. In general, the principal philosophical claim against tradition is that testimony to such phenomena requires very strong evidence, and for all the reasons above it is hard to see tradition as supplying evidence of that strength.

These are very common arguments in various versions, familiar to all of us. But the 18th-century Scottish philosopher David Hume took the argument a step further. He conceptualized and formalized it, presenting it—purportedly—as a probabilistic, formal choice between two alternatives. In doing so he sharpened the force of all the familiar arguments by setting them within a single organizing framework, which endowed the objections with tremendous persuasive power. It is no wonder that, to this day, atheist thinkers regard Hume’s formulation as a deathblow to arguments from tradition—what later came to be called the “Argument from Testimony,” or “Argument from Miracle.” Many maintain that after Hume, anyone who accepts arguments based on tradition is a fool, for there is no logical or philosophical possibility of justifying such a position. Thus the philosopher J. H. Randall wrote that Hume dealt the Argument from Testimony a fatal blow “by demonstrating the matter so clearly that since Hume scarcely has an intelligent person dared advance the argument.” Randall adds that “in the eighteenth century miracles were the principal supporting explanation of faith; [following Hume’s remarks they became] in the nineteenth century the principal problem requiring explanation.” True, other philosophers—such as Swinburne, Lewis, and Portill—have defended the Argument from Testimony; atheists, however, dismiss them as apologists and give their arguments little weight.

Needless to say, today’s atheists greatly enjoy repeating with absolute confidence (religious types, did I mention? See my conversation with Jeremy Fogel on our site’s podcast, Radical) such boastful declarations. This certainly gave me motivation to show the flaws in Hume’s arguments—of which there are several. I note that I have done so in my books as well, Emet ve-Lo Yatziv (chapters 15–16) and The First Existent (fifth conversation, chapter 4). There, however, I presented Hume briefly and my critiques addressed directly the fallacy at the heart of his argument, with less systematic analysis. Here I will describe Hume’s arguments in more detail as they appear in his An Enquiry Concerning Human Understanding (trans. from English: Guy Algat; Resling, 2008), chapter 10, and then I will critique them systematically, point by point. My thanks to Jeremy Fogel, who sent me a scan of the chapter as part of our WhatsApp discussion. Here I can present my detailed response.

This chapter is divided into two parts. The first part presents the schema by which arguments from testimony are to be examined. The second shows that the various traditions have not, and cannot, pass this test; hence the conclusion that such traditions cannot be accepted. In the next two sections I will present Hume’s arguments as he gives them, and only then proceed to my critiques.

The chapter’s first part: Hume’s test

The first part opens with the claim that experience is the only tool at our disposal when we examine matters of fact (a foundational assumption of Hume, a thoroughgoing empiricist). This tool, however, is not immune to error. His first example is a weather forecast that may fail. But that is not a good example, for the forecast is already given within a statistical framework. Experience itself tells us that weather forecasts do not always come true. From this he infers that there can be varying degrees of certainty regarding conclusions drawn from experience. That certainty depends on the number of cases underlying our inference, or on the nature of the facts and the inference itself. For instance, our trust in human testimony is based on experience that people generally do not lie (we have previously seen testimonies confirmed), unless they have reasons to do so. Without experiential corroboration, we would not place trust in such testimony.

This means that our trust in human testimony must be examined in light of experience. For example, we should consider the character of the witnesses, how many there are, contradictions in the testimonies, and the manner in which the testimony is conveyed.

After all that, we must also factor in the content of the testimony: how unusual and implausible it is. A testimony whose content is highly wondrous and implausible will require very strong corroboration (many reliable witnesses, a trustworthy mode of transmission, etc.). Note that the reason is a competition between two findings of experience: on the one hand, such an extraordinary event is implausible in light of our experience; on the other hand stands the reliability of the testimony reporting it (again, per our experience). The more implausible the event, the higher the reliability we will demand of the testimony about it.

Now consider testimony whose content is not merely unusual and marvelous (i.e., of low prior probability) but a miracle proper—say, an iron bar that remains suspended in midair and does not fall to the ground. Such a miracle is a violation of the laws of nature and thus an event never observed. Our confidence that such an event does not occur is very high, and to overturn it we would need testimony with a degree of certainty greater than our confidence in the natural law purportedly violated. Only testimony whose falsity would be more miraculous than the event described could succeed. Even then, the confidence left us in the occurrence of the miracle would be relatively low, since its strength is the result of subtracting our confidence in the natural law from our confidence in the testimony.

So much for the chapter’s first part, which presents the test itself. In the second part, as noted, Hume proceeds to examine testimonies to miracles and religious revelations in light of this test.

The chapter’s second part: Critique of the Argument from Testimony

The second part opens with the assertion that there has never been testimony to a miracle that met this test—i.e., whose reliability exceeded our confidence in the natural law allegedly violated. This is not, of course, a factual determination but an a priori claim. He assumes it. The reason: there has never been testimony from a large number of very reliable and educated people of sound judgment who testified to the occurrence of a miracle.

Against the absence of such reliable testimony stands induction: in light of our experience—miracles simply do not occur.

In addition, people have a tendency to accept events precisely because of their preposterousness. He claims we take pleasure in believing in miracles; even those who do not believe in them take great pleasure in basking in the aura of one who reports a miracle. Therefore the reliability of a report of a miracle should be discounted further. In particular this holds when the hearers are of low critical judgment and vivid imagination—in short, primitive thinking. This is evidenced by the many cases in which reports of miracles among such primitive populations were debunked (or at least failed to receive corroboration from any other source).

Reports of miracles are common primarily among barbarous nations; and even when found among a civilized nation, one finds that it received them from more primitive forebears. Almost every nation passes down wondrous mythic tales about its origins and earliest days. For some reason this never happens in our day and in regions close to us in space and time, but rather in regions where we have no direct contact by which to confirm or refute the story. The nearer one approaches the age of enlightenment, the fewer the reports of miracles and supernatural events.

Another fact is that people do occasionally lie to one another. Declaring some story to be fiction is thus not an extraordinary or wildly implausible claim; it certainly does not contradict the laws of nature.

Moreover, there is not a single testimony to a miracle that is not contradicted by other testimonies. Thus two factors undermine the reliability of such testimonies: their miraculous content and contradiction within the testimonies themselves. This is not meant as a universal claim, though he presents it thus. His intent is likely directed mainly at religious traditions, which by their nature contradict one another. If you accept one, you thereby reject all the others, and vice versa. Hence such testimonies contain built-in contradiction and denial. He then offers as an example a non-religious miracle that, according to Tacitus, the Roman emperor Vespasian wrought when he healed a blind man with his saliva. The miracle is presented in a very credible and eloquent manner by a renowned and reliable historian of proven judgment, from people who themselves witnessed the miracle which occurred publicly. Additionally, the subject is a very intelligent, honest, and educated emperor known not to exaggerate or present himself as a divine figure with special powers, but to speak plainly and straightforwardly. The miracle was told after the family had already been stripped of its standing, so it had nothing to lose by such a lie. Thus, anyone who accepts this testimony has every reason to adopt it. Hume concludes this description with the following sentence: “And if we add to this the very fact itself… the impression will be that it is impossible to conceive of stronger evidence than this in favor of such a gross and palpable falsehood.” This conclusion is surprising, for he has just presented all the reasons to adopt the testimony, without laying out any counter-considerations, and yet he ends with the emphatic claim that it is a gross and palpable falsehood. He proceeds to further examples that I will not enter here. Finally he reveals his contention: what leads us all to reject such testimonies despite their strength is solely the miraculous nature of the events themselves. For him, this proves that a miraculous event is never admissible, whatever the evidence for it. He adds the difficulty of falsifying such a story, which generally occurred long ago, and the fact that in many cases the story supports the views and interests of those who testify to it (as is usually the case with religious testimonies).

His conclusion is that there has never been a report of a miracle accepted by an educated and judicious public; therefore, in light of our experience, it suffices that the content of a testimony be miraculous for us not to bother examining its reliability. No degree of reliability will suffice to persuade us of the occurrence of a miracle.

He further claims that the common argument—that it is difficult to deceive a broad public with such a fabrication—is also unconvincing. The reason is that while it may be difficult, it is not impossible. By contrast, the occurrence of a miracle that departs from the laws of nature is, by its very nature, of lower probability.

Attributing the miracle to an omnipotent being likewise changes nothing, since the very existence and capacities of such a being do not themselves derive from experience.

Finally, he summarizes the argument as follows:

Here the circle is closed. He applies the criterion presented in the first part to traditions of miracles and revelations, and his conclusion is that to accept such testimonies we would have to suppose that the probability the testimony is false is lower than the probability that the events occurred. But the probability of the events is 0, since they contravene the laws of nature; whereas the probability that a testimony has been corrupted—by error or deceit—is never 0, since such an occurrence can and has happened throughout history. Therefore we must categorically and a priori reject every tradition of a miracle or revelation, even without any further examination (which is usually impossible anyway). In chapter 15 of The Science of Freedom I offered several examples of the fallacies involved in rejecting one alternative merely because of its low probability without considering the probability of the other (see there the discussion of Munchausen syndrome by proxy, and more).

In the end, Hume even appeals to our integrity, claiming that if we examine ourselves we will find that, as critical and rational people in all other contexts (apart from religious tradition), we would not be prepared to assume that something supernatural occurred. Integrity therefore obliges us to do the same here as well.

A note on two components of Hume’s argument

As noted, Hume’s argument came up briefly at the end of my debate with Jeremy Fogel, where a hermeneutic dispute arose between us as to whether Hume attacks only the chain of transmission or the very encounter with a miracle. Jeremy argued, for some reason, that Hume attacks only the chain; I disagreed (see here Yaron’s question about this).

Hume does address, in several places, the reliability of the traditional chain of transmission about a miracle. But Jeremy also agreed that the root of the difficulty lies in the very encounter with a miracle, not in the transmission chain. Consider a person standing before an event that appears to him as a miracle. He faces two possibilities: assume that the event occurred and that he is indeed witnessing a miracle, or assume that he is mistaken or has been deceived in some way. Hume’s argument is meant to lead this person to the conclusion that what he himself is seeing is not a miracle, because errors exist in the world but miracles do not exist at all. Note that the problem’s focus is not in the chain of transmission that carries the miracle from that person to us, but in that person’s own conclusion. One can then add the possibility of error introduced by the chain of transmission, further reducing the probability that this is reliable testimony—but that is not the crux of the issue. In my view, even in Hume’s text he attacks the very encounter with the miracle, not only the reports about it. As noted, the report merely further diminishes the testimony’s credibility and strengthens the argument, but that is not its essence. Moreover, Hume likely assumes that the original people who experienced the miracle would themselves reject it out of hand (recall, he is an empiricist), and therefore he focuses on attacking the chain from them onward.

In any case, from here on I will focus mainly on the logic of the argument itself and only briefly address the reliability of the chain of transmission. I do not divide the critiques according to the chapter’s two parts, since it is hard to separate a logical critique of the argument from critiques of its applications. The reader can discern when I am attacking the logic of the argument itself and when I am addressing more practical, secondary points.

So much for Hume’s arguments. In the next column I will detail my critiques of them.