"I have discovered an argument [...] which, if just, will, with the wise and learned, be an everlasting check to all kinds of superstitious delusion"
Miraculous events lie at the origins of most religions. Jesus’s resurrection from the dead. Moses’s parting of the Red Sea. Mohammed’s journeys on the back of a winged horse. Joseph Smith’s shenanigans with the Angel Moroni. All these events are, in common parlance, miraculous. If you wish to be a religious believer, you must accept the historical occurrence of at least some of these originating miraculous events. The problem is that you don’t get to observe them — to verify them with your own eyes. You must rely on the testimony of others, often handed down to you through religious texts or a lineage of oral histories. Is this testimonial evidence ever sufficient to warrant belief in the miraculous?
David Hume famously argued that it wasn’t. In his essay Of Miracles, which appears as section 10 of his larger work An Enquiry Concerning Human Understanding, Hume argues that testimonial evidence is unlikely to ever be sufficient to warrant belief in a miraculous event. Hume’s argument has been the subject of much interpretation and debate over the past 250 or so years. Much of that debate obscures or misrepresents what Hume actually argued. Fortunately, the philosopher Alexander George has recently published a beautiful exposition and analysis of Hume’s argument, titled The Everlasting Check: Hume on Miracles, which corrects the record on a number of key points.
George’s analysis is somewhat similar to that of Robert Fogelin, which I have written about previously. In essence, both authors argue that Hume’s argument is routinely misinterpreted as providing an a priori (or ‘in principle’) case against the possibility of testimonial proof of miracles. But this is emphatically not what Hume argues: Hume merely argues that testimonial proof of the miraculous is extremely unlikely. The mistake stems from the fact that Hume’s essay is broken into two parts, and many people assume that both parts present two separate arguments: an a priori argument and an a posteriori argument. They do not. Both parts must be read together as presenting one single argument.
Although Fogelin and George reach similar conclusions, they do so by subtly different means. George’s exposition has the advantage of being more thorough, more up to date, and ultimately more straightforward. On top of that, George makes some important points about how Hume defines miracles and how he relates the evidence for the occurrence of natural laws to the evidence for the reliability of testimony. When you understand these points, much of Hume’s argument falls neatly into place.
So what I want to do over the remainder of this post is to share George’s analysis of Hume’s argument. By the end, I’m hoping that the reader will appreciate the strengths (and limitations) of Hume’s analysis.
1. The Basic Structure of Hume’s Argument
Hume’s argument is about proof of miracles. But what is a ‘miracle’? Hume is pretty clear about this. He defines a miracle as:
Miracle = A violation of the laws of nature.
The problem is that this raises a further question: what is a law of nature? Some people argue that a law of nature is an exceptionless pattern in the natural world. The law of energy conservation or the second law of thermodynamics would be common examples. But to say that a particular law, L, is an exceptionless pattern is also somewhat ambiguous. Is the pattern truly exceptionless? In other words, is it some ontological necessity of the universe? Or is it simply that we have never observed an exception? In other words, is it a strong epistemic inference from our observations?
George argues that Hume favoured a strictly epistemic understanding of the laws of nature. He viewed laws of nature as well-confirmed regularities. We say that there is a law of conservation of energy because we have never observed energy coming into being from nothing; we have only ever observed it being changed from one form to another. But who knows, maybe some day we will observe an exception and this exception will itself be well-confirmed and hence we will have to revise our original conception of the law. That said, laws of nature are, for Hume, very very well-confirmed. We usually have the strongest possible evidence in their favour.
This epistemic understanding of miracles is consistent with Hume’s general empiricism, and it allows for miracles to come in degrees: an event can be more or less miraculous depending on how well-confirmed the regularity with which it is inconsistent really is. Indeed, Hume himself distinguished so-called ‘marvels’ from ‘miracles’ on the grounds that the former were inconsistent with less well-confirmed regularities than the latter.
This is all by way of saying that Hume’s target is best understood in the following manner (note: this is my gloss on Hume, not George’s):
Miracle* = A violation of a very well confirmed regularity that is observed in the natural world.
What then of Hume’s argument? That argument has a very simple structure, consisting of two premises and a conclusion. George uses formal mathematical terminology to describe the two premises of the argument, referring to them as the ‘first lemma’ and ‘second lemma’, respectively. He also refers to the conclusion as a ‘theorem’. I will follow suit though I don’t think it is strictly necessary:
- (1) If the falsehood of testimony on behalf of an alleged miraculous religious event is not “more miraculous” than the event itself, then it is not rational to believe in the occurrence of that event on the basis of that testimony. [First Lemma]
- (2) The falsehood of the testimony we have on behalf of alleged religious miraculous events is not more miraculous than those events themselves. [Second Lemma]
- (3) Therefore, it is not rational to believe that those miraculous religious events have occurred. [Hume’s Theorem]
I have modified the wording of the lemmas and the theorem slightly from how they appear in George’s original text. I did this in order to make the argument more logically coherent and more consistent with Hume’s aims. As we will see, Hume doesn’t try to argue against all testimonial proofs of the miraculous; merely against the historical testimonial proof provided on behalf of the major religions. Indeed, Hume specifies under what conditions it might be acceptable to believe in a miracle.
For those who care about how this argument maps onto the structure of Hume’s essay, the first lemma is defended in the first part and the second lemma is defended in the second part. Again, to reiterate what was said above, it is important that we don’t disconnect the two parts. You cannot derive Hume’s Theorem from the first part alone.
2. Establishing the First Lemma
The first lemma of Hume’s argument is what has generated most of the debate and controversy. It is significant because it states a general principle or test that should apply to the evaluation of testimonial evidence. As such, it has significance beyond the debate about the occurrence of religious miracles. Here’s how Hume introduces it:
[N]o 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. And even in that case, there is a mutual destruction of arguments, and the superior only gives us an assurance suitable to that degree of force, which remains, after deducting the inferior.
Here’s the basic idea: the evidence we have in favour of the laws of nature is strong. We have observed the same regularities over and over again. So our epistemic confidence in their status is very very high. A miraculous event, by definition, contradicts these regularities. To believe in the miracle we would need to have even stronger evidence in favour of its occurrence than we do in favour of the laws of nature. When the evidence supplied comes in the shape of testimony, then we will only reach this standard when the probability of the testimony being false is lower (i.e. more miraculous) than the probability of the miracle itself.
The major philosophical problem with this analysis is that it relies on the commensurability of the testimonial evidence in favour of the miraculous and evidence in favour of the laws of nature. In order to establish the respective improbabilities of the two propositions (testimony is false; miracle is true) we need to establish some common metric along which those probabilities can be measured. This is where Hume made his most important philosophical contribution to the debate. He argued that all evidence ultimately stems from our observations of the world around us. Any particular proposition that we hold to be true about the world (e.g. that dead men stay dead; that energy is conserved; that there is a universal tendency toward increased entropy) is ultimately warranted by observations we have made about the world. This is clearly true for the laws of nature, but it is also true for testimonial proofs as well. The reason why we are usually confident of testimony is because because there is a well-confirmed regularity to the effect that when someone says that an event occurred it usually did occur. To put it another way, Hume is saying that there is a law of testimony. This law of testimony is grounded in our experiences of the relationship between testimony and events in the real world.
Drawing out this equivalence is critical to Hume’s project, and it is a point that interpreters of Hume often miss, according to George. It is only through establishing this equivalence that Hume is entitled to say that we can meaningfully compare the relative probabilities of testimonial proof and a putative law of nature. As he puts it:
Given Hume’s central commensurating analysis in Part 1 of his essay, it is indeed meaningful to evaluate whether the relation is greater than holds between an event’s occurrence and a testimony’s falsehood (because we can now appreciate that there is a lawlike claim about nature with which the falsehood of that testimony conflicts).
This equivalence is what justifies the general principle stated in Hume’s first lemma. We are only warranted in believing a miracle on the basis of testimony if the warrant for the law of testimony in that instance outweighs the warrant for the law of nature. This only happens when the falsehood of the testimony is more improbable than the falsehood of the law of nature.
2. Establishing the Second Lemma
The first lemma is interesting in and of itself, but it doesn’t get us anywhere close to establishing that it is irrational to believe in historically testified miracles. Obviously, the suspicion underlying the first lemma is that the warrant we have for the law of testimony is much weaker than the warrant we have for any candidate law of nature. After all, even though testimony is usually accurate, there are plenty of times when it isn’t. People misunderstand what they have seen. They fabulate and overinterpret. They suffer from a confirmation bias, whereby they assume that what they saw is consistent with their prior beliefs/desires. They also, sometimes, lie outright for their own gain.
The purpose of the second lemma is to establish that the law of testimony breaks down in the case of historically testified miracles: that the falsehood of the testimony in favour of those miraculous events is not more miraculous than the events themselves. Hume does this by setting out four main lines of argument in favour of the second lemma. These can be briefly summarised as:
(A) The Reliability Argument: Testimonial evidence is strongest when it meets certain conditions of reliability. Specifically, when it comes from (i) many witnesses; (ii) of good sense, education and integrity; (iii) who have reputations that could be tarnished by the evidence they are presenting; and (iv) they testify to an even that occurred in public and so would have enabled to detection of fraud. Hume’s contention is that the evidence for religious miracles doesn’t meet these conditions.
(B) The Propensity to the Marvellous Argument: People in general have a propensity to the marvellous. As Hume puts it ‘the passion of suprize and wonder, arising from miracles, being an agreeable emotion, gives a sensible tendency towards the belief of those events, from which it is derived…[we] love to partake of satisfaction at second-hand or by rebound, and place a pride and delight in exciting the admiration of others.’ In other words, we have an emotional tendency to accept and repeat claims to the miraculous. What’s more, this tendency is even higher in the case of religious miracles because of the authority and power that is often granted to accepted religious prophets and missionaries. This means there is a tendency to trade emotional satisfaction for accuracy. This may not be done deliberately; it may be entirely subconscious; but it still undermines credibility.
(C) The ‘Ignorant and Barbarous’ Peoples Argument: The people from whom testimony of historical religious miracles emanates are ‘ignorant and barbarous’. They were likely to be predisposed to credulity and misunderstanding; they were not sceptical and disinterested observers. Obviously, Hume’s language in expressing this argument is antiquated and un-PC.
(D) The ‘End to Commonsense’ Argument: When it comes to religious affairs in general, Hume argues that there is an ‘end of commonsense’. In other words, people don’t follow commonsense rules of reasoning and inference when it comes to religious matters. Systems of religious thought tend to blind people to the truth and we should consequently weigh religious claims accordingly. Hume uses a thought experiment to make his point here. He asks us to imagine a group of people who testified to the resurrection of Elizabeth I in order to found a new system of religion. How much weight would we accord their testimony? Very little. Hume suggests that all ‘men of sense’ would ‘reject the fact…without farther examination’.
The first and third of these arguments have to do with the credibility and reliability of the witnesses we have for religious miracles. The strength of these arguments depends on how accurately they represent the historical record. The second and fourth arguments are more general, and focus on people’s tendency to be misled when it comes to the marvellous and the miraculous, particularly when it is religious in nature.
It is important to emphasise that these arguments do not undermine all testimony in favour of miracles. Hume is very clear about this. He thinks it would be possible to have proof of a miracle that overcomes the test established by the first lemma. Indeed, he uses a thought experiment — ‘the eight days of darkness’-thought experiment — to illustrate the conditions under which testimony would provide proof of the miraculous. The thought experiment asks us to imagine that “all authors, in all languages, agree, that, from the first of January 1600, there was a total darkness over the whole earth for eight days: Suppose that the tradition of this extraordinary event is still strong and lively among the people: That all travellers, who return from foreign countries, bring us accounts of the same tradition, without the least variation or contradiction”. This admittedly sets a high bar, but that’s what we would expect when testimony is going up against a law of nature and it at least shows that testimony might be sufficient on some occasions.
3. Limitations of Hume’s Argument
Hume’s argument has many limitations. My belief is that the general principle established by the first lemma is fairly robust: it is difficult to see how else testimony could override a law of nature. The arguments adduced in support of the second lemma are much less robust. If you read any religious apologetics, you will know that arguments (A) and (C) are highly contested. Apologists introduce all sorts of arguments for thinking that the testimony we have is more reliable than Hume claims. For instance, in relation to the resurrection of Jesus, they will argue that we do have multiple witnesses (perhaps as many as 500), that some of them were well-educated and disinterested, and that some of them did suffer greatly for what they believed. Now, I tend to think that Hume is still, broadly speaking, correct and that the apologetical arguments ultimately fail, but clearly a lot more work would be needed to address each and every one of these contentions and thereby shore up the support for the second lemma.
I also think that arguments (B) and (D) are pretty contentious. Hume is certainly on to something. There are lots of putative miracle claims out there, and systems of religious thought sometimes do come with benefits for believers which may cause them to be more credulous than they ought to be. Furthermore, psychological evidence suggests that we do have a propensity to over-ascribe agency to events in the natural world, and to misinterpet what our senses have shown us. These errors probably lie at the foundation of many miracle claims. But to dismiss all religion as anathema to commonsense and rationality seems to go too far to me. I think, along with Oscar Wilde, that ‘commonsense’ is not that common and that it is a mistake to assume that secular/non-religious people have some innate epistemic superiority over their religious brethren. It’s more complicated than that.
Despite this, I would argue that Hume provides a pretty good framework for evaluating religious miracle claims and that while he may be wrong (or too superficial and glib) on some of the critical details, anyone who cares about this issue in the modern day plays on the terrain that he defined nearly three centuries ago.