Has David Hume shown that it can never be reasonable to believe in miracles? Part 2

In the previous article, we examined David Hume’s argument that no testimony could ever justify belief in a miracle, given the improbability of miraculous events compared to the apparent uniformity of natural laws. However, many philosophers have raised significant objections to Hume’s argument. In this article, we’ll explore three key critiques: concerns about circular reasoning, debates over how we assess the intrinsic likelihood of events, and the implications of Hume’s probabilistic framework for science itself. 

Circularity worries  

One worry that’s been raised about Hume’s argument is that it involves a certain kind of circularity. What Hume is doing is basically pitting two bodies of evidence against each other: on the one side, the enormous body of observations of the world behaving in a certain way — say, of dead people staying dead — and on the other side, a few isolated pieces of testimony claiming that in one instance, the world behaved in a radically different way — claiming, for instance, that a certain dead person came back to life.

It’s crucial for Hume’s argument that the laws of nature have indeed been established by “a firm and unalterable experience,” as he puts it — or in other words, it’s crucial for Hume that previous experience of the world really has been completely uniform. But the difficulty is that Hume only gets to portray previous human experience as uniform by excluding miracle reports from the pool of human experience before the argument has even begun, and that’s the sense in which the argument is circular. As one of Hume’s contemporaries, George Campbell, put it:  

I leave it to [Hume] to explain with what consistency he can assert that the laws of nature are established by a uniform experience (which experience is chiefly the result of testimony), and at the same time allow that all human histories are full of the reports of miracles and prodigies, which are violations of those laws.[1] 

Which factors determine the intrinsic likelihood of an event?  

I’ve said that when we’re considering whether to believe someone’s testimony that an event occurred, we’re essentially weighing two things against one another: the intrinsic likelihood of that event, (how likely or unlikely it is, in general, that such a thing might happen) and the strength of the direct evidence for the alleged event (for example, the testimony of someone claiming that that event happened or a photograph that seems to depict it happening, etc). The direct evidence needs to be strong enough to overcome the intrinsic unlikelihood of the event.

If it’s a fairly mundane event — say, that my wife drank tea with breakfast this morning — since that’s intrinsically not unlikely at all, then only a small amount of direct evidence is needed to tip the scales in favour of believing that it happened. But if it’s an event that’s intrinsically very unlikely, then some very strong evidence is going to be needed to make it reasonable to believe that it really happened. Again, this is in essence the principle that “extraordinary claims require extraordinary evidence.” But a crucial question is this: what factors determine how intrinsically likely or unlikely a certain event is?

Hume’s answer is very simple: it’s a matter of how often an event of that kind has been observed in the past. If the answer is “never,” then we should think that an event like that is extremely intrinsically unlikely. The intrinsic likelihood of a resurrection, for Hume, is simply a matter of how often dead people have come back to life. But a number of philosophers have argued that in fact this isn’t the only factor that is relevant to determining how intrinsically likely or unlikely an alleged event is. This is how the philosopher of science Stephen Wykstra puts it:   

[T]he Achilles’ Heel of Hume’s argument is that it estimates the intrinsic improbability [of an event] entirely on the basis of considerations ‘from below’, by essentially statistical considerations of how often the event (when classified under a certain type) has been observed to occur. Such considerations are not irrelevant, but equally relevant…are considerations ‘from above’, where we consider how expectable the event might be in the light of larger theoretical and metaphysical considerations, bringing these to bear on the full historical context of the event.[2] 

In other words, Wykstra is saying that the intrinsic likelihood of an alleged event isn’t just a matter of how often an event of that kind has happened in the past. It’s also a matter of bigger picture considerations about what sort of universe we’re living in, and not least, whether there’s a God who might occasionally intervene to bring about something that wouldn’t happen if nature was left to itself. When you think about it, if there is a God, one of the best ways for God to decisively communicate with humans would be to bring about an event that strikingly deviates from the way nature normally works — an event that stands out in stark contrast to an otherwise unbroken regularity. As the eighteenth-century theologian William Adams put it:  

There must be an ordinary regular course of nature, before there can be anything extraordinary. A river must flow, before its stream can be interrupted.[3]  

If there was some reason to think there might be a God, then that in turn would give us some reason for expecting that the world might contain at least a few events that depart strikingly from the way nature normally works. To help unpack this idea here’s an analogy. Suppose that you work on a cupcake factory production line, and that your job is to inspect cupcakes for quality control as they come down the conveyor belt to you. Suppose you’ve been working in this factory for years now, and every single cupcake that has ever come down the conveyor belt has had plain white icing and no writing on. What’s the intrinsic likelihood that a cupcake with purple icing and “happy birthday” writing on it will come down the conveyor belt to you today? Is it simply a matter of how often that’s happened in the past (i.e., never)? No.

It seems like in principle there are other factors that could be relevant to the question of how unlikely such an event would be. Suppose, for example, that you’ve learned that one of your co-workers further up the production line, the one responsible for putting the icing on the fairy cakes, has a bit of a crush on you and that they know that it’s your birthday today and that your favourite colour is purple. That sort of information makes it a bit more likely that something quite out of the ordinary might happen today — something like a purple cupcake with “happy birthday” icing coming down the conveyor belt to you. The intrinsic likelihood of an event like that happening isn’t just a matter of how often it’s happened in the past. 

Given that other factors are also relevant to the intrinsic probability of miracles — it isn’t just a matter of how often such events have happened in the past — then the fact that we have a huge body of past experience of the world working in regular and orderly ways doesn’t, by itself, show that miracles are overwhelmingly improbable. Indeed, a miracle-working God would need there to be a backdrop of order and regularity in nature so that a miracle would stand out and be noticeable against that backdrop.

The medieval philosopher Thomas Aquinas defined a miracle as an event that nature wouldn’t have the power to bring about if left to itself, and this has seemed to many philosophers to be a more adequate definition of a miracle than Hume’s definition of a miracle as “a violation of the laws of nature.”[4] If God exists then there is the possibility that nature is sometimes not left to itself. So, in short, how intrinsically likely miracles are depends at least in part on whether there’s evidence for the existence of God. 

Richard Swinburne develops this point at length in his book The Resurrection of God Incarnate where he makes a philosophical and historical case for the resurrection of Jesus of Nazareth. A key part of his argument is the idea that if there’s a God (something which he’s argued at length across various books), then God might well have reason to communicate with humanity by becoming incarnate in the form of a human being and putting a “signature” on the life of that human being in a way that only God could do.[5]

Swinburne argues that Jesus is the candidate who best fits the bill, based on various details about Jesus’ life that are widely accepted among ancient historians. And so, Swinburne’s claim is that, whilst there’s no reason to expect a resurrection to happen in the case of most other human beings who’ve lived, there is some reason in the specific case of Jesus. Or in other words, we’re not asking how likely it is that God would raise some random Joe from the dead; we’re asking how likely it is that God would raise from the dead a man whose life bore the specific characteristics that Jesus of Nazareth’s life did – including the claims he made about himself, the kind of teaching he gave and the personal qualities he embodied. Swinburne’s point is that if there’s good reason to suppose that God would raise someone like that from the dead, then the amount of direct historical evidence that would be needed to establish Jesus’ resurrection isn’t enormous. 

Now, an atheist philosopher might try to argue that the fact that we observe so much regularity in the world is very much to be expected if atheism is true, because atheism predicts a universe with unbroken natural regularities, but not if theism is true, given its expectation of potentially lots of miracles. If that’s correct, the fact of natural regularities supports atheism over theism, and drives down the intrinsic probability of miracles. But there are a couple of things to say about this.

The first is that there’s a case to be made for thinking that natural regularities aren’t at all likely given naturalism — this is something Swinburne has argued.[6] The second thing is that even if we suppose that natural regularities are likely given atheism, they support atheism over theism only if natural regularities are unlikely given the hypothesis of a miracle-working God. But in fact, the hypothesis of a miracle-working God strongly predicts virtually unbroken regularities, because after all, the very point of a miracle is that it should stand out strikingly against a backdrop of otherwise unbroken regularities. 

Hume’s understanding of probability spells trouble for science 

A third worry I’d like to highlight with Hume’s argument against miracles has to do with the way he understands probability. Remember that Hume thinks that the more an alleged event would involve a deviation or a departure from what’s gone before, the more intrinsically unlikely it is that such an event would really happen. To put it another way, for Hume, the intrinsic likelihood of an event happening is a matter of how much it resembles what has happened previously. The less it would resemble what’s happened before, the more unlikely it is that the alleged event would happen. So, if in the past, every time we’ve observed an object being released in mid-air it’s fallen towards the ground, then an event involving an object being released in mid-air and just hovering would be a huge deviation from what’s gone before; it wouldn’t resemble the past. And so for Hume, it’s extremely improbable that such an event would happen.

The worry with this is that if we were to accept Hume’s understanding of probability then we would have to reject all sorts of striking scientific discoveries, on the grounds that those discoveries involve occurrences that seem to go profoundly against how we previously thought the world worked. Here’s how the philosopher of science Alan Hajek puts it: 

Science tells us that there are events that happen so rarely that they have not yet fallen under our experience—certain astronomical phenomena, for example. And other new phenomena may be quite commonplace, and yet have gone unnoticed…and indeed it is one of science’s tasks to discover them. Put simply, if miracles get vanishingly small probability because they are so disanalogous to our past experience, then so should these events.[7] 

The point of all this is that there’s a serious difficulty in Hume’s idea that events that deviate significantly from what we’ve previously experienced should be thought of as extremely improbable and hence rejected. If we adopted that attitude, then we couldn’t do science. And that’s not even to mention the way that Hume’s attitude towards probabilities in his discussion of miracles seems to be sharply at odds with things he says elsewhere about how we can’t know that the future will resemble the past.[8] 

Concluding reflections 

A careful examination of Hume’s argument reveals some serious flaws with it. It seems that Hume has failed to show that, without even looking at the details of any particular miracle claim, we should reject all miracle claims in principle. Even many atheist philosophers today would contend that Hume’s argument doesn’t work. The atheist philosopher of science John Earman, for example, wrote an important study of Hume’s in-principle argument, which concludes that the argument is an “abject failure.”[9] This doesn’t mean it’s open season for miracle claims, though.

Put another way, the failure of Hume’s argument doesn’t mean that anything goes. What it does mean is that we can take things on a case-by-case basis, exercising our critical faculties whilst keeping an open mind. The central miracle claim of Christianity is of course that on the third day after being executed by Roman crucifixion, Jesus of Nazareth was bodily raised from the dead. Given the failure of Hume’s argument, we are free to approach this claim ready to look at all of the specifics of the historical evidence, and not only that, but also the evidence for thinking that we live in a universe created by a God who might bring about such an event. 

References

1. Quoted in Earman, Hume’s Abject Failure (Oxford: Oxford University Press, 2000), p. 13 
2. Stephen J. Wykstra, “Not done in a corner: How to be a sensible evidentialist”, Philosophical Books 43 (2002), pp. 105-6. 
3. Quoted in Timothy J. McGrew, “Miracles,” The Stanford Encyclopedia of Philosophy, (2019) Edward N. Zalta (ed.), URL = https://plato.stanford.edu/entries/miracles/ 
4. See Section 1 of McGrew, “Miracles” 
5. Richard Swinburne, The Resurrection of God Incarnate (Oxford: Oxford University Press, 2003) 
6. Richard Swinburne, The Existence of God, 2nd ed. (Oxford: Oxford University Press, 2004), chapter 8. 
7. Alan Hajek, “Are miracles chimerical?”, Oxford Studies in Philosophy of Religion, Volume 1. (Oxford: Oxford University Press, 2008), pp. 82-104; p. 103 
8. For more on this, see Hajek, “Are miracles chimaerical?”[Chimerical], p. 101-02 
9. Earman, Hume’s Abject Failure