How To Understand David Hume's Argument On Miracles

Hume's Philosophy of Skepticism, Probability, And Wisdom On Full Display

This post is too long for email. Please click on the title to read the entire thing in browser!

In my last post, I talked about how the internet gets Hume wrong. In this post, I’m going to teach you how to get Hume right. What is Hume trying to argue in Of Miracles? This is a question that many people have gotten wrong, going back to even Hume’s time.

You’re not going to find the answer just by reading the essay. Many of his points are implied, but not specifically stated (it was a chapter in a larger work, after all), and he also uses words differently than we do today.

To write this post I leaned heavily on two sources, but I have read others and am confident that what’s argued here will be in line with the consensus of Hume scholars. Sources will be cited by page number (which may be slightly inaccurate due to kindle formatting), but instead of writing Vanderburgh and Fogelin’s whole names within citations, I will use V and F.

As a statement of humility, this is not a peer reviewed article. I am not a Hume expert, though I consider myself a student of the Scotsman. As my interpretation is not a testament to the miraculous, I hope you’ll see me as a credible witness. And if a more credible witness testifies against me, consider it a proof.¹

The key points of this post, which can be found in each section with a major heading, are that:

1. Hume was not a Bayesian;

2. Hume understood probability and evaluated evidence more like a lawyer in a courtroom than a statistician in a lecture hall;

3. Hume used probabilistic terminology differently than we do today;

4. Hume was a mitigated skeptic;

5. His mitigated skepticism led him to understand probability in such a way that made him generally disbelieve all miracle claims;

6. To understand Of Miracles you must understand both parts one and two of his argument together, not as two separate arguments;

7. Because of the problematic nature of miracles and the Humean account of wisdom (apportioning beliefs based on evidence, being informed of the fact that religious motives degrades the quality of testimony), the wise person should be perhaps instinctively dismissive of miracle claims;

8. I think he’s basically right.

Before we dive in, I would appreciate your subscription, as this post took a lot of time and effort to research and write. Thank you!

Hume Was Not A (Proto) Bayesian

Hume’s probability calculation is not the same kind of mathematical probability we are accustomed to today, and it is certainly not a Bayesian one.

Bayes’s probability work was not widely known in 1748, when Hume published the first edition of the Enquiry, where the original essay Of Miracles appeared. Though, per some historical evidence, Hume read and admired Richard Price’s paper applying Bayesian methods to the evidence of miracles in 1767, Hume did not address these Bayesian arguments in his revised editions of the Enquiry in 1768 or 1777. Vanderburgh concludes:

This suggests that Hume ultimately did not view Bayes’s work as relevant to the argument against miracles…Given Hume’s familiarity with Pascalian probability, and his acquaintance through Price with Bayesian ideas, his non-numerical treatment of the evidential probability of miracles even in the latest editions of the Enquiry concerning Human Understanding must be seen as a deliberate philosophical position, not as a result of negligence or ignorance. (Vanderburgh 120)

Hume differed from the Bayesians in that “belief is determined by the purely psychological characteristic “force and vivacity” (V28).” Comparing pieces of evidence is a matter for judgment, not calculation (V29). In this way, for Hume, evidence is not measured, it is weighed.

One of the reasons this is the case is because, in Hume’s epistemology, mathematical calculations are relations of ideas, which are themselves things that can be known with absolute certainty, while empirical matters of fact cannot be known with certainty. Hume would have seen it as a category mistake to believe matters of fact in mathematical degrees (V30).

Though Hume’s reading of probability may be different from what we think of probability today, it’s worth mentioning that, per Vanderburgh, Hume’s method was the most dominant approach to probability throughout western history (V4). So as tempting as it is to view his reasoning as if he’s a frequent reader of Nate Silver or Matthew Adelstein, we must instead evaluate Hume on his own terms.

Hume’s Legal Language

Hume evaluated evidence like a classically-trained lawyer. Given Hume’s background as a law student, his experience as a librarian, and other historical evidence, it’s unlikely Hume would not be familiar with Roman and medieval legal texts.

What’s more, Of Miracles is filled to the brim with references to legal concepts. Some of these are less obvious, given the relationship they have with today’s adversarial method of philosophy and argumentation. These include:

• Gathering and critiquing evidence,

• Presenting a case,

• Considering the counter-case,

• Answering the counter-case,

But some are more obvious, such as Hume discussing:

• The concept of “full proof” (a medieval law concept meaning proof sufficient for conviction in a capital case),

• Internal and external evidence (more medieval legal concepts),

• Principles and concepts related to the weighing of evidence (such as that a weaker evidence can never destroy a stronger),

• That more unusual claims require more evidence,

• The contrariety of evidence on either side,

• Competing witnesses,

• And factors affecting witness credibility,

On top of that (all of this is from Vanderburgh’s book if you couldn’t tell, you should read it!):

He explicitly uses analogies of courts and judges deciding a case—he even mentions a legal example of a judge comparing the competing testimonies of two pairs of witnesses and deciding that the evidence on each side cancels each other out, leaving no basis for conviction (EHU 10.24)....There is even a suggestion in Hume (EHU 10.32) of something we today associate with legal statutes of limitations: it is difficult enough for courts to reasonably establish recent facts, and nearly impossible with regard to facts in the distant past. (V87)

All of this to say, in reading Of Miracles and attempting to understand Hume’s argument, you must not read it as a contemporary statistician or analytic philosopher, but as a judge evaluating evidence in the courtroom. For Hume, probability isn’t something calculated, but something judged.

Important Terminology

Hume did not use probabilistic language in the same mathematical way that we’re used to as 21st century philosophy enthusiasts, and there are wide-spanning implications. Specifically, Hume used words like proof, probability, demonstration, certainty, zero probability, impossibility, and others different from the way we do today.

Zero Probability And Full Proof

Today, we think of probability in what Vanderburgh calls Pascalian terms: The fact that a proposition has “zero” probability” means that the proposition is a logical impossibility, while a proposition that has 100 percent probability is a certainty. Hume would not express probability in this way.

Instead, he would use language in what some scholars would call a Baconian way, but what Vanderburgh argues is better understood as Pre-Pascalian. That is, “the fact that a proposition has no probability means that there is no positive evidence in favor of accepting it” (V77).

What’s more:

When a proposition has maximum probability for a Pascalian, this means it is a logical truth (its falsity is logically impossible); for a non-Pascalian, “full proof” regarding an empirical fact is less than perfect certainty, even though full proof is the highest degree of probability attainable for empirical propositions (V77).

Thus, Hume affirms admissible propositions or evidence much like a judge affirms admissible evidence. Saying a proposition has “no probability” is not the same as saying it is impossible, nor is saying one amounting to a “full proof” is saying that it is certain.

Rather, a proposition having “no probability” is one with insufficient evidence to be used or weighed against other evidence, while one having “full proof” amounts to the best quality evidence that can be weighed against other evidence.

Moral Certainty and Moral Evidence

In this way, for Hume, the highest degree of certainty we could achieve for empirical proofs is moral certainty or moral evidence. This is “a degree of assurance sufficient for action and belief but short of certainty” (V30). However, we “never achieve absolute certainty regarding matters of fact simply because of the kind of knowledge in question” because matters of fact are not relations of ideas (V96).

To repeat, for the Humean:

• Relations of ideas are propositions we can be certain of because they are by definition true (such as a triangle having three sides). These are often called a priori propositions.

• Matters of fact are propositions we cannot be certain of. We know them from experience, not from definitions or pure logic. These are often called a posteriori propositions.

Remember: Hume would believe it’s a category error to talk about probability in mathematical ways.

More importantly: when he uses the language of probability, proof, and possibility in Of Miracles, he’s using the terms in an a posteriori or matter of fact way.

Demonstration, Probability, And Proof.

For Hume, demonstrations were certain propositions because they described relations of ideas. Confusingly enough, today we would refer to a demonstration as a proof, but that’s not what Hume means by the term.

For Hume, a proof is a matter of fact. A proof is the accumulation of probabilities in favor of a generalization that we can safely assume is the case. In this regard, a proof is “the zenith of evidence and belief with regard to an empirical matter of fact” (V96).

Probabilities can become proofs, but they cannot become demonstrations, nor can proofs become demonstrations. In this regard, Hume’s moral certainty amounts to a proof (V96). From that proof, we “end up with well-grounded expectations about past uniformities continuing into the future” (V96).

Impossibility

Thus, when Hume is talking about a matter of fact being impossible, it’s important to not conflate him talking about logical or demonstrative or a priori impossibility.

More explicitly, and in line with my last post, Hume does not believe that these things we call miracles cannot happen out of principle. In his Enquiry, he says: “The contrary of every matter of fact is still possible; because it can never imply a contradiction.”

Take the proposition (to use Vanderburgh’s example) “A human can raw bench press 1,500 pounds.”

It’s not logically impossible, just impossible-given-what-we-know. That is, the world record for raw (unassisted) bench press in 2025 was 782 pounds, the progress of the record is decelerated in spite of training improvements², and human physiology and the laws of physics (the breaking strength of bones) make it utterly unbelievable that human could complete a 1,500 pound raw bench press.³

Again, that’s not because it’s logically impossible, but given the facts we know about related matters, it’s impossible in Humean terms.

Understanding Mitigated Skepticism:

To understand how all of these concepts come together for Hume and miracles, we must understand three more Humean concepts: The problem of induction, causality, and mitigated skepticism.

First, Hume famously pioneered the problem of induction. According to this problem, the uniformity of nature is not a logical necessity (a demonstration), but a matter of fact (proof). In this way, there is nothing in reason that says that the world will be the same tomorrow as it is today. In human experience, the sun has risen every morning for as long as humans can remember and attest, but tomorrow could be the first day that it doesn’t happen.

Secondly, Hume understood causality to not be something we observe. This is more complicated, and there is disagreement among Hume scholars about what specifically he believed.⁴ In short, Hume believed that causality wasn’t observable, as we only really observe conjunctive events, not conjunctive events plus this magical thing that reveals itself as “causality.”

For example, when a billiard ball hits another billiard ball, we don’t see this “thing” called “causality” happen, we just see one object move, make contact with another object, and then they both subsequently act in a certain way.

Due to this observational limitation, we can imagine the billiard balls doing things we have no experience observing them do, such as colliding at a slow speed, and then conjunctively flying vertically in the air at the speed of sound. What’s more, because we do not observe causality, just viewing the conjunctive events will not give us an understanding of what event caused what event (especially if we only do it once).

For many people, this is pretty bad! How can we make sense of the world at all, if there’s no logical basis for induction and if we don’t actually observe causality?!

Though Hume unearthed these problems, he believed there were practical solutions that would work for most people in their daily lives. In this regard, Hume was a mitigated skeptic. This means that, though Hume may have rejected the idea that we could have deep “knowledge” or perception of how the world works (just as we couldn’t know or perceive causation or rationally establish induction), he thought that we could still make some sense of the world using other tools of our cognition, namely probability.

From Mitigated Skepticism To Proofs of Natural Laws

As a mitigated skeptic, Hume assumes The Principle of the Uniformity of Nature, which simply says “future cases will resemble past cases” (43). According to Vanderburgh, the principle of the uniformity of nature, is “the principle on which all inductive reasoning depends,” but also “merely a regulative ideal for investigation of the world, rather than a claim regarding a supposed truth about the world” (43). From this principle and induction:

The depth and breadth of the exceptionless regularity of past experience gives the strongest kind of warrant possible to the belief that the law will continue to hold in the same way in the future. It is not that the evidence demonstrates with certainty that the law is true, it is just that no empirical claim can possibly have stronger evidence than what we have with regard to those things we call laws of nature. (50)

Let’s illustrate this with the billiard ball example. To the extent that we can make any causal inference at all about what follows from two billiard balls making contact with each other, it will be based on previous experience. Though it is logically possible that their interaction could have them spontaneously combust or transform into butterflies or blast Bohemian Rhapsody at 120 decibels, we have no experience of them doing those things.

The more experience you have watching billiard balls interact, the more data you have in support of the probability that they will act in specific ways in the future (namely: move in geometric ways on the billiard table). In the absence of these balls acting in alternative ways, the less evidence you have that they will act in alternative ways. In this sense, you have no probability that the balls will act in alternative ways.⁵

Thus, based on your previous experience of the movements and interactions of billiard balls, you can form probabilities about how they will act in different circumstances. The more data and experiences you collect, the more accurate your probabilities. As your probabilities accumulate and remain uncontested, they eventually become moral proofs or laws of nature.

To recap: Though there is no reasonable foundation (i.e inherent in logic) for the reliability of induction or the uniformity of nature, for Hume, if we merely assume the principle of the uniformity of nature, we can formulate probabilities about how the world works. Over time, certain probabilities reach the status of moral proofs or laws of nature.

The Problem of Miracles Pre-Testimony

And this is where miracles and testimonies for miracles become problematic. As a matter of fact, a miracle is a one-time event.⁶ Though the miracle apologist is correct in saying we cannot call a miracle as such if there isn’t this background law or proof to discern it, from Hume’s perspective, that does not save it from being rendered impossible.

Miracles Are A Sample Of One

To use mathematical phrasing, for Hume, you need more than one data point to be confident that a law of nature has been violated or that an established proof is not, indeed, a proof. Yet by both common definition and as matter of fact, miracles are only one data point.

Per Vanderburgh: “Singular events such as miracles of the sort with which Hume is concerned could never supply the evidence needed to put aside the regularity” (V52).

Another way of expressing this is asking two rhetorical questions:

1. How do you know that you observed a miracle, and not just a poorly understood natural event?

2. How do you know that the thing you attribute as the cause of the purported miracle was indeed the cause?

Though he does not touch upon this directly or explicitly in Of Miracles, to the extent Hume believes that we can establish an understanding about the relationship between conjunctive events, we need more than one set of conjunctive events to reliably draw a probable inference.

This means that miracles, generally understood as singular events (either by definition or as matter of fact), have a probability problem. Namely, to draw a reliable inference from an observed events, humans need more than one sample of that event; something that miracles don’t typically provide.

Supernatural Causes Are Hard To Prove

To illustrate it in a more humorous way, if you transplanted a pre-industrial human to the present, explained to him what a video was, and showed him a video of a car blowing up after its ignition is turned, he may assume that car ignitions cause cars to explode, and subsequently violently stop you from going to work, in an attempt to save your life. In this way, to have any inferential idea about how an observed event functions, we need more than one data point, which miracles do not provide.

The ambiguity doesn’t just stop with one not knowing what exactly happened with conjunctive events, but also in one being able to attribute the cause of the conjunctive events in the first place. Put more concretely, it’s easier to demonstrate that the sun turned dark than it is to demonstrate that the sun turned dark and that Odium caused the sun to turn dark.

Back to our pre-industrial human. Let’s assume we show him automobiles that don’t blow up for a day. He thinks they’re cool! What inference can he draw about them, knowing very little about them,⁷ when one randomly explodes on day two? Very little, if he knows nothing about combustion. For this man, a supernatural explanation is just as plausible as a natural one.

When we understand a causal phenomenon sufficiently, the pre-industrial man attributing a car explosion to a supernatural event seems obviously wrongheaded. Yet still, miracle apologists think we should attribute unknown, weird causal events to the supernatural, and it’s not clear how the situation is much different.

Yes, Miracles Should Be Replicated (Seriously!)

In this way, even if you see a one-time event that Hume would call marvelous, it’s not quite rational, given the human limits of observing causation, to call it miraculous (i.e. having supernatural or religious causal origins).

Just as you need more than one data point to discern what causally happened in your observation, you similarly need more than one data point to assess the cause of that event as well, to determine if it was a magic trick, an illusion, a marvel, or a miracle. Put in other terms:

So, for Hume, the warrant of the causal inference is founded on the long course of uniform experience. This suggests that on the first occurrence of an event of any event-type (even something that later turns out to be an instance of a perfect regularity) we are not warranted in inferring the future occurrence of a similar consequence from a similar antecedent. This in turn suggests that no (other) matter of fact can properly be inferred from a truly singular event. By itself this provides grounds for doubts about miracles…If the conditions under which the event took place are reproduced and a similar event comes about, that is the beginning of evidence for a new law of nature or revision of an old one. As the evidence of this kind builds up to a high enough level (in which case it becomes increasingly unlikely that there was and continues to be misperception, mistransmission, or deception involved in the testimony and increasingly likely that the event type really occurs), there are then grounds for a rational belief that the original event (although it was singular when it first occurred) really did occur. Degree of belief is for Hume a function of evidence: It will change as the evidence changes. (V55-56)

Miracle apologists often say that skeptics wouldn’t believe a miracle if they saw one, and thus the skeptics are just dogmatically anti-miracle. This is not true. Rather, skeptics such as Hume recognize the necessity of needing multiple, repeated observations before drawing any probabilistic causal inference for any observed novel phenomena, not just miracles.

In this regard, the Humean skeptic doesn’t have a bias against religion or miracles, but understands how the science and psychology of causality works.

Experience tells us that when marvelous observed events are repeated, studied, and tested, a natural explanation will better fit the data, by virtue of being more austere in its assumptions. Though marvels and miracles may happen, experience leads us to doubt their reports, given both their poor track record and that human minds begin to decode marvelous events when they can be observed multiple times.

Conversely, the Humean can be confident that a natural causal explanation can be unearthed with enough observation and experimentation. The scientific method has a strong track record.⁸

Part Two Is Important For The Argument’s Structure

In line with my previous post, according to Robert J. Fogelin’s A Defense of Hume On Miracles, it’s better not to break down Of Miracles into two separate arguments because the two sections fit together into a coherent whole:

Stated broadly, the task of part 1 is to establish the appropriate standards for evaluating testimony in behalf of a miracle of any kind; the task of part 2 is to show that reports of religious miracles have not in the past met these standards. Taking experience as his guide, Hume further concludes that there is no likelihood that they will ever do so. (F9-10)⁹

In this way, Part 2 is important to the essay because Hume is arguing as a matter of fact that no testimony has met the standards of being a miracle. It’s an essential step to his conclusion, which (as stated in my previous post) was about the rationality of miracle testimony in establishing religion.

Just as the specific wording about testimony is important to understand Hume’s argument on miracles, so too is his wording on the foundation of religion. Namely, Hume states the maxim “that no human testimony can have such force to prove a miracle, and make it a just foundation for any such system of religion.”

Fogelin asserts:

The phrasing here is important. As stated, it does not completely rule out the possibility of testimony establishing the occurrence of a miracle, for Hume’s maxim is limited in its scope to miracles intended to serve as the foundation for a system of religion. (F24)¹⁰

This post mainly deals with the first part of Hume’s argument, and not the conclusion, which I feel I argued sufficiently in my last post. But, still, keep the overall structure in mind! Part 2 and the conclusion are just as important as part 1. Hume has more to say than just “miracles are typically unlikely!”

Religious Testimony For Miracles Is Low Quality

Without full context, some commenters view some of Hume’s dismissals of testimony as arbitrary. Namely, that he wouldn’t believe a religious testimony about a queen resurrecting, but he would believe credible a non-religious account attesting to a period of darkness that lasts days, seems inconsistent.

But if you read his whole essay, Hume’s skepticism toward the queen scenario is anything but arbitrary. Rather, it’s informed by the fact that religious motivations often make testimony more unreliable:

As violations of truth are more common in the testimony concerning religious miracles, than in that concerning any other matters of fact; they must diminish very much the authority of the former testimony, and make us form a general resolution, never to lend any attention to it, with whatever specious pretence it may be covered (Enquiry 10.38)

Or, as Fogelin summarizes:

In short, for Hume, it is an empirical fact, amply illustrated by history, that testimony concerning religious miracles is notoriously unreliable. On the basis of this general fact about the quality of such testimony, the wise reasoner has ample grounds for rejecting it. (F29)

For Hume, if you can imagine a situation where testimony could rise to the level of credibility to establish a religious miracle, he claims it has not actually happened in the course of history.

Hume’s Maxim and Dismissing Improbable Events

Let’s think of some examples of testimony of improbable events. Say I tell you that:

• A load of tile fell off a semi on the Pennsylvania Turnpike and spilled onto the road in a configuration that replicated Leonardo’s Last Supper, except that the figure of Jesus looked astonishingly like Elvis.¹¹

• I have lived an interesting life. I went to middle school with Jadeveon Clowney. I ran into Jill Biden at my local Barnes & Noble when she was First Lady. I once rear-ended Josh Turner in Atlanta traffic. Lance Armstrong once injured me on a multi-use trail in Texas. While visiting the Pacific Northwest, I once was at a coffee shop in Seattle at the same time as Jeff Bezos. While partying in DC, I did psychedelics with Matthew Yglesias. My wife just let me make out with [super hot celebrity, take your pick].

Should you, as a wise person, believe either of these claims? Should you feel compelled to investigate them? The Humean says no!¹²

On the example of some random schmuck bumping elbows with famous people all the time, Folegin says: “The sheer improbability that all (or most) of these things happened is sufficient for discounting his testimony, whatever his reasons for presenting it.” (F12)¹³

We understand this when it comes to events that are merely very improbable, but why not share the same standards for the miracle claim? Why are we more defensive of the spectacular?

Hume’s Moral Probability and Wisdom

Hume’s maxim in Of Miracles is understood as 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.

In this way, the extreme improbability of miraculous events is the reason why we should be skeptical from the outset, not the source of the miracles (testimony).

Per Vanderburgh:

What makes [Hume’s Maxim] rational, and hence not really paradoxical, is a point about epistemic reliability. Miracles either occur or they do not. If miracles do not occur, testimony about miracles will always be false. If miracles do occur, testimony about the occurrence of a miracle will still (given what we know about humanity) almost always be false (by misperception, deception, or mistransmission) and only very rarely true. Considering the set of all cases of testimony, then, it will universally or almost universally be better to believe (the belief will always or almost always be true) that testimony to violations of laws of nature is false. If we follow Hume’s maxim we will be in error only very rarely, if at all. (V 65)

The probabilistic calculus of the maxim has built-in constraints to advance various principles beyond just measurable precision and certainty, but also actionability, minimizing error rate, and wisdom. In this regard, the probability is not calculable like Bayesian probability, but more like legal probability, which has principles such as “proof beyond reasonable doubt,” “presumption of innocence,” and so on, that advances values beyond just confidence.

Wisdom is perhaps the most important constraint. For Hume, a wise thinker does not believe wholesale every idea that becomes enlivened in her brain, nor does she believe everything she is told, especially if it’s extremely improbable. A careful, deliberate, informed, and probabilistic thinker will understand that this epistemic behavior would be less effective than proportioning her belief to the evidence she has experienced and is familiar. It’s from that experience that she judges not to believe a miracle testimony. Per Morris:

I don’t need to investigate [the Last supper claim] further, even though the alleged event violates no natural laws…The fact that the truth is out there doesn’t mean that we have to be out there investigating everything we’re told. Part of being wise is knowing when not to waste your time. (V58)

Tying It All Together

When evaluating miracles, Hume implores a non-mathematical, non-Pascalian understanding of probability that is more like one you find in a court room than in a science lab. He is a skeptic about induction and causality; he is not a Pyrrhonian skeptic, but a mitigated skeptic, which (in the most simplified way) means he believes we can have practical “knowledge” about the world around us, even if we can’t have “deeper knowledge” about how the world works.

From this, Hume assumes the principle of the uniformity of nature, and from that he argues for the reliability of probability at drawing inferences for the future, including establishing moral proofs and laws of nature. From the outset of this system, miracles are considered very improbable, not out of definition, but out of the facts of their case weighed against the proofs of the laws of nature, per Hume’s non-Pascalian probability.

But Hume’s argument against miracles does not end there, as for him, the evidence for miracles doesn’t even amount to a probability. This is the case for two reasons:

1. To make any sense of any novel observation, we need more than one occurrence of the event to formulate a probability. As miracles detail events that only happened once, they cannot amount to a probability, and therefore cannot even weigh against the proofs of the laws of nature.¹⁴

2. All religious miracle testimony falls radically short of credibility because of how religious motivation corrupts witness credibility, the additional attribution of causation to divine agents, and the limited data available to draw an inference.¹⁵ This corrupting influence makes religious miracle testimony inferior to witness attestation of the merely novel and extraordinary.

For Hume, this means miracles are a posteriori impossible, not because they are logically impossible, but because they lack the necessary evidence to weigh their probability against the “morally certain” laws of nature. Therefore, miracle testimony does not even amount to a Humean probability, let alone a proof, and so there’s no contest between the two.

Final Thoughts

Just as a raw 1,500 pound bench press is considered impossible, as we have various proofs against it and no evidence for it, so too are miracles impossible. For Hume, this is a matter of fact, not a logical necessity. This assertion may be demonstrated as false, but the effort of disproving it would be a waste of time for wise people.

Although we could chase down and debate the credibility of every specific miracle and testimony, a wise person does not do this: We have enough data showing the failure to establish religious miracles and the failures of their species of supporting testimony, itself amounting to more than just a probability. In this regard, the accumulated probability for the failure of miracle arguments is arguably a proof in itself.¹⁶

Until witness testimony for religious miracles radically improves, and thus more miracles can be credibly established, amounting to a probability and eventually a competing proof against the proof of the laws of nature, skeptics will continue to be confident in their dismissal of miracles as impossible.

If you’ve read this far and enjoyed this post, please consider clicking the button above, or right here to buy me a coffee. My content is free, but I spent well over $100 researching this post, so a little “tip” would be appreciated.

1. This is a self-deprecating joke that will make more sense after you read the post!

2. In 2015, it was 735.5 pounds, in 1972 it was 675 pounds, and in 1953 it was 500 pounds.

3. Students of the history of fitness and performance may point out that it was thought impossible that a human could run a mile in less than four minutes, and so to say we couldn’t raw bench 1500 pounds seems similarly fallacious. I don’t think this is a good comparison for a few reasons. First, one of the reasons it was thought impossible was because there would be a mental barrier stopping people from accomplishing it. Second, there was radical progress toward this goal. The lack of progress was a result of World War II stopping the progress, but from what I’ve read, some athletes were dangerously close to breaking four minutes anyway. These trends don’t hold for the raw bench pressing 1,500 pounds. That much weight will probably break most people’s bones and we are only about half way to that goal, after decades of progress. Say that it’s actually possible to bench 1,500 raw. Okay. What about 3,000? This isn’t moving the goal posts, so much as it’s showing that there are physical limitations on human performance (most people intuitively understand this), and that we can know those limits - and thus the limits of possibility, given our “proofs” about the laws of physics, human physiology, etc.

4. Was Hume a skeptical realist? Or something else? I’m going to oversimplify, and perhaps say something that an actual Hume scholar may disagree with, but this oversimplification and possible nuanced error won’t mislead you about Hume’s perspective on miracles.

5. One vulnerability of Hume’s method is that, relative to Bayesian and other forms of probability, it is more resistant to updating its prior. The archetypical example is the existence of meteorites. For a long time, rural people would testify to seeing meteorites falling out of the sky, to which the educated authorities simply didn’t believe. This is because they had their own proofs about the laws of nature. Eventually, the consensus was overturned with some spectacular examples of meteor showers that the educated could not ignore. I don’t think this is a defeater for Hume’s method, so much that those who disbelieve specific miracle claims on “law of nature” grounds need to be very specific about what laws are their proofs. Because the more specific we are about what our proofs are, I suspect the less confident we’ll be in the specific laws that we have or affirm. But as we will see, unlike miracle claims, these novel and spectacular claims about nature can be affirmed by virtue of not being one-time events and by the testimony not being poisoned by zeal. There’s an entire literature on Hume for the role of testimony for non-spectacular events, believing things because we read about them, etc. But! This post is long already and I don’t think we have the time or space.

6. To the extent the occurrence of a miracle can be called a fact, of course.

7. Other than going vroom vroommmmm!

8. One reviewer (

Katelynn Bennett

) of this post said this about this section: “I feel as if there is something missing. Are miracles by nature one off events? Would they no longer be miraculous if they were replicable? What does any of this have to do with faith, i.e. isn’t faith needed to believe in miracles despite the science and psychology of causality? I’m curious whether Hume says anything about faith” I too agree that there’s more to be said here, and I’ll probably write a post about it at some point. What’s important here, in the context of Of Miracles, is that Hume is trying to defeat what’s called “Natural Theology” - the idea that “rational” people can come to theism or Christianity through rationality and the observation of nature. For Hume, saying “well if you presuppose faith” is to cede defeat to his argument. Remember: Of Miracles is a work of epistemology, not metaphysics or ontology. I’ll maybe say more about this in a future post.

9. I am aware that Fogelin says “in behalf” here instead of “on behalf.” It’s weird! But trust me, it’s not a typo

10. Emphasis is mine. Also, I focus in this piece less on Hume’s conclusion about establishing religion because I believe I covered it sufficiently in the first piece.

11. This example is from Ted Morris in 2022, in a commentary on Vanderburgh’s conference presentation of his defense of Hume.

12. The Clowney and Biden stories are true, though I did walk past Matthew Yglesias on U Street once

13. It’s worth noting that Folegin’s imagination of celebrity reports was much more extreme than mine, including the likes of Osama Bin Laden and Margaret Thatcher. I think it’s also worth noting here that, yes, celebrities are more likely to run into each other by virtue of being celebrities. What’s more, if you are proximate to a city like Washington, DC, as I am, you’re more likely to randomly see famous people in public. I’ve seen Jill Biden, Sean Spicer, Matthew Yglesias, and Pete Buttigieg randomly in public, for instance, but I didn’t hang out with them. One’s experience with running into famous people can be conceivably more likely with mitigating considerations (such as geographic proximity). But the likelihood of a “nobody” interacting with multiple world famous people is decidedly low.

14. Again, Hume does not explicitly state this in Of Miracles, but it’s an implication of his philosophy of causation.

15. You can perhaps replicate a novel or extraordinary experience, but likely not a miracle

16. Though as far as I know, Hume never said that specifically