Saturday, 7 June 2014

Hume on Miracles

Many people take David Hume's argument against miracles as discounting the possibility of miracles, or somehow loading the dice against theism. Consider Craig Keener in his 1248 page magnum opus Miracles:
[Hume] argues, based on “experience,” that miracles do not happen, yet dismisses credible eyewitness testimony for miracles (i.e., others’ experience) on his assumption that miracles do not happen. (Kindle Locations 4325-4326)
I will argue that to succeed logically [Hume's] approach must presuppose atheism or deism. (Kindle Location 4320)
Thus, on the usual reading of Hume, he manages to define away any possibility of a miracle occurring, by defining “miracle” as a violation of natural law, yet defining “natural law” as principles that cannot be violated. (Kindle Locations 4680-4682)
I cannot see that Hume assumes miracles do not happen, nor that he requires a presupposition of atheism or deism, nor that he defines natural law as principles that cannot be violated; the conclusion of his argument is simply that a rational person would not believe a miracle claim. This, though, is based on his view of knowledge acquisition; the argument itself looks valid to me (see below), if I'm reading him right, so to attack Hume one would have to attack the premises of his argument and show it to be unsound. The argument does not, in fact, rule out miracles a priori, but if someone agrees with his account of knowledge acquisition (which account does not exclude theists, though they might not consider it exhaustive) he shows that they would not believe miracle claims, if they were being rational.

Has a miracle occurred? Just to consider this question is to allow that miracles are not ruled out a priori. If, for example, a miracle is defined as a supernatural event, as it commonly is, and a person thinks the supernatural is impossible, they would conclude that miracles are impossible. This ‘hard naturalist’ argument goes something like:
Premise 1  Miracles are violations of natural law
Premise 2 Natural law is eternally inviolable
Conclusion Miracles never occur
Natural law has developed over the centuries and seems likely to continue to develop, so the problem is: we don’t know what the natural laws are precisely, so when an event violates the natural laws as we understand them, we do not know if this is because we have the natural laws wrong or because the event is genuinely supernatural. The hard naturalist above would counter that this is just an epistemological problem; in principle, there are inviolable natural laws, so in principle supernatural events are impossible. But a posteriori we have not confirmed their inviolability, so this is just an assertion. A believer in miracles could simply assert in response that natural law is just very rarely violated.

Hume takes a more subtle approach. Only our experience guides us in ‘matters of fact’, he writes in ‘Of Miracles’. Bear in mind his famous 'fork':
All the objects of human reason or enquiry may naturally be divided into two kinds, to wit, Relations of Ideas, and Matters of fact. Of the first kind are the sciences of Geometry, Algebra, and Arithmetic ... [which are] discoverable by the mere operation of thought ... Matters of fact, which are the second object of human reason, are not ascertained in the same manner; nor is our evidence of their truth, however great, of a like nature with the foregoing.
The distinction between 'relations of ideas' and 'matters of fact' is similar to the a priori/a posteriori distinction, and is commonly made by philosophers; Descartes drew a similar distinction in his method of doubt, for example. So we take into account all our (sometimes competing) experiences and draw conclusions about the world from them in proportion to the evidence they supply. If our conclusions are based on ‘an infallible experience’ we consider this a full proof for the future; if not, we proceed as cautiously as our past experience dictates. Hume is describing here how people generally behave, and perhaps should behave; by weighing the available evidence; that weighing is surely the source of the word rational, with its ‘ratio’ root.

Hume discusses human testimony, and agrees that there is a ‘useful conformity of facts to the reports of witnesses’. Experience tells us that memory is ‘tenacious’, people are usually truthful and being caught lying is shameful. Testimony can be false, but very well attested reports are a ‘proof or a probability’ that an event has occurred. If that most unusual event, a miracle, occurs, and the testimony is so good that it would normally be considered a proof, we have a ‘proof against proof’.

Hume's argument, then, does not rely on a premise that testimony is unreliable per se, although this is often mentioned when he is brought up; consider theologian Randal Rauser here, in response to a miracle sceptic:
Mike is apparently invoking an epistemic principle like this:
Testimony skepticism principle (TSP): Carefully documented testimonial evidence has negligible evidential value because testimony has been shown to be unreliable.
While testimony can be unreliable (and that is an important part of Hume's argument - see P4 below) this does not mean it has 'negligible evidential value'; in fact, Hume specifically says well attested reports can be a 'proof' that an event has occurred:
And as the evidence, derived from witnesses and human testimony, is founded on past experience, so it varies with the experience, and is regarded either as a proof or a probability, according as the conjunction between any particular kind of report and any kind of object has been found to be constant or variable.
But since Hume defines a miracle as ‘a violation of the laws of nature’, which have been established by our ‘firm and unalterable experience’, there can be no better argument from experience than the one that supports the laws of nature . Hume invites us to weigh two arguments from experience; this is not an a priori matter, but an empirical one. A rational person must agree, then, that the chances of the natural laws being wrong are less than the chances that the best imaginable testimony could be wrong. So Hume’s argument is more like this:
Premise 1 The evidence for matters of fact are established by empirical enquiry
Premise 2 Both natural law and testimony are matters of fact
Conclusion 1 The evidence for natural law and testimony is established by empirical enquiry
Premise 3 Empirical enquiry records those things that occur reliably, without violation, as natural law
Premise 4 Empirical enquiry shows that testimony is often reliable, but not without violation
Premise 5 A rational person believes matters of fact in proportion to the evidence gathered by empirical enquiry.
Conclusion 2 A rational person believes a violation in testimony is more probable than a violation of natural law
Premise 6 Miracles are a violation of natural law
Conclusion A rational person always believes a violation in testimony before she believes a miracle has occurred
A note on P6; other definitions are available. Some say that divine intervention, while supernatural, does not necessarily violate natural law. But without the unusual stamp of a violation, any other divine event would appear to fall more properly under the notion of divine providence, covering the orderly conduct of the cosmos. Although in a footnote Hume says:
Sometimes an event may not, in itself, seem to be contrary to the laws of nature, and yet, if it were real, it might, by reason of some circumstances, be denominated a miracle; because, in fact, it is contrary to these laws. Thus if a person, claiming a divine authority, should command a sick person to be well, a healthful man to fall down dead, the clouds to pour rain, the winds to blow, in short, should order many natural events, which immediately follow upon his command; these might justly be esteemed miracles, because they are really, in this case, contrary to the laws of nature.
...which suggests that any divine intervention is miraculous. Even if an event does not seem contrary to natural law, it could still be. P3 derives from this famous sentence:
A miracle is a violation of the laws of nature; and as a firm and unalterable experience has established these laws, the proof against a miracle, from the very nature of the fact, is as entire as any argument from experience can possibly be imagined.
Keener surely agrees with P3 when he says:
Natural law is, after all, merely our construct of how nature functions. (Kindle Locations 4675-4676 - my emphasis)
...but then says:
If one chooses to define natural law in such a way as to make variation from it impossible, one has simply redefined words about reality rather than made an argument, and someone else could counter by redefining “miracle” as part of that reality. (Kindle Locations 4676-4677) 
But Hume clearly doesn't define natural law to make variation impossible; he defines it as Keener does. It should hopefully be clear by now that Hume is not defining natural law as uniform and inviolable; he is saying what we experience as uniform and inviolable we call natural law (it is 'our construct', to use Keener's words) - an important difference because it means natural laws can be violated - we can be wrong about the uniform and inviolable.

Keener cites many critics objecting to Hume's definition of miracle in P6. That is fine, but theists seem to be stuck between a rock and a hard place; either miracles are part of the natural world, in which case their effects can be monitored like every other natural event, and evidence accumulated in favour or against, or they're not, and Hume's definition appears closer to anything they might offer. A miracle needs to be an exception, not prosaic, and this definition communicates this well, so I will use it for the purposes of this discussion.

The argument looks perfectly valid to me, with no hidden assumption of atheism; it simply states what Hume takes to be how humans acquire knowledge, be they theist or atheist. There is no premise that states that miracles do not happen. They may happen, in fact, but, the argument says, a rational person could not accept any report of them happening.

P1 Hume draws from his ‘fork’, which, as we've seen, distinguishes between abstract reasoning, like mathematics, and matters of fact – the a priori and a posteriori. Hume is firmly setting miracles in the domain of a posteriori arguments, and would reject the hard naturalist argument presented further up, since it is not an a posteriori argument. The premise can be attacked on the grounds that there are ‘other ways of knowing’; in particular, revelation, intuition and religious experience. If there is justification other than empirical enquiry for belief in matters of fact, the argument would be unsound, and to some there is a certainty to some matters of fact that does not submit to empirical proof. Consider these quotes from William James’s The Varieties of Religious Experience from theists:
I don't think I ever doubted the existence of God, or had him drop out of my consciousness.
What I felt on these occasions was a temporary loss of my own identity, accompanied by an illumination which revealed to me a deeper significance than I had been wont to attach to life. It is in this that I find my justification for saying that I have enjoyed communication with God.  
The suggestion is that there are supernatural things we know a priori and not a posteriori, so the fork should be a trident. It’s conceivable too that we know natural things a priori, but that is not important in this scenario; what is important is the claim that supernatural facts are not discovered empirically. This objection can be addressed by tightening the argument; we replace ‘matters of fact’ in P1, P2 and P5 with ‘matters of natural fact’. This is stretching Hume’s words a little, but it is still a pretty close reflection of Hume’s argument. This still rules out believing miracle testimony since, if the premises are correct, the communication of any miracle claim is natural (testimony is communicated naturally), so its reliability is tested empirically, and empirically it must be less reliable than natural law. The only defeater here would be to establish non-natural testimony, which leads us to the second objection.

The original argument might be attacked on the grounds that empirical enquiry reveals supernatural facts among the natural facts. Hume asks us to reject the greater miracle, and the ironic title of J. L. Mackie’s The Miracle of Theism hints at something. Mackie writes:
...I hope to show that [religion’s] continuing hold on the minds of many reasonable people is surprising enough to count as a miracle in at least the original sense. (p.12)
The ‘original sense’ here is ‘something surprising or marvellous’ (p.11), but the very survival of churches to the present day, the faith professed by the ‘many reasonable people’ down the centuries, and the wisdom and miracles recorded in holy books could amount to a testimony which is a miracle in itself. So, the argument goes, a reasonable person should consider it a greater miracle that this ‘testimony’ be wrong than that the laws of nature are violated. Is it not a more parsimonious explanation of this long history that some supernatural agency does intervene in unusual ways occasionally, and testifies through the Bible or the Quran, for example, and delivers revelations and experiences to people whose testimony survives, supernaturally?

This is possible, and the history does call for an explanation. The objection attacks P3, suggesting that empirical enquiry does not just record regularities as natural law, but also records irregularities from the natural law as, at least, marvels, and at best, miracles. Keener raises a similar objection, saying Hume's argument is:
...a circular argument that excludes the evidence of the claim supposedly under consideration and other claims like it. (Kindle Locations 4387-4388)
The point is, I think, that P3 excludes the violations which are under debate, automatically disallowing miracles. But, as discussed, Hume's argument simply states what humans do: we call regularities 'natural laws'. There is no circularity, just a description of the processes involved, and an observation of the respective evidences. In any case, it's easy to adjust Hume's premise to account for some irregularities and completely sidestep the objection:
Empirical enquiry records those things that occur reliably, almost without violation, as natural law
After all, even well-established natural laws such as Newton's continue to be used as natural laws even though we know that the movement of Mercury, for example, does not conform to them (and with Einstein we obviously have a new more accurate understanding of natural laws). Science accepts we have an incomplete and tentative understanding of the natural laws, and this is very much in keeping with Hume's epistemology, which is a sceptical approach to enquiry that falls short of Descartes' extreme approach to knowledge - the method of doubt. Nevertheless there is still a large asymmetry between the evidence for the regularities (the physical sciences) and the evidence for the irregularities (historical enquiry and theology), and that asymmetry is increasing, so the objection is unconvincing. There is less and less space left for miracles.

The ‘miracle of theism’ claim also falls foul of Hume’s further objections to miracles. To employ these in Hume’s order: firstly, despite the claims of supernatural intervention there does seem to be an available natural explanation for even the most successful religion. Great intelligence is no barrier to deception nor indeed to self-deception, and ardency can be a sign of an ulterior motive or that reason is no longer being applied to a person’s beliefs. To prefer any natural explanation does imply that natural explanations must be more likely than miracles, but given miracles’ status as exceptional events evoking wonder and awe, compared to providence, this is not an unreasonable assumption.

Secondly, humans derive a great sense of ‘surprise and wonder’ from miraculous reports and this feeling pre-disposes us towards believing them and repeating them.

Thirdly, many miracle reports date from pre-industrial times, reported and written down by people who were ignorant of many natural facts about the world. Their credulity is excusable, and perfectly natural.

Fourthly, there is such a diversity of religious belief in time and space that any miracles used in the service of any religion are outweighed by the miracles used in the service of all the others. The maths seems inescapable; the resurrection of Jesus testifies that Christ is the son of God, but many believe that the prophet Mohammed mounting a flying horse testifies against Jesus’ divinity. Miracles are also used in the service of Judaism, Hinduism, Buddhism and more. So even if the supernatural is allowed, the evidence for any subset of miracles is outweighed by the evidence for the remainder. If miracles are supposed to testify to something vaguer than a religion’s specific claims, the less power it has to persuade. The less a miracle is attributable to a particular religion, the less pertinent it is to that particular religion’s claims.


There are, no doubt, other objections to Hume’s logic, but simply detailing hundreds of miracle claims does not cut the mustard, unless the weight of all that evidence outweighed our 'firm and unalterable experience' of the natural law. Needless to say even a book of 1248 pages does not come close to that; the theist would have to surmount all the evidence of all the physical sciences. This is really unlikely, so theists should save their breath and attack the premises of Hume's actual argument instead. Stories of 'miraculous' recoveries from illness, for example, while heart warming, will simply never outweigh the evidence for the natural laws.

One further twist: what if one experienced an event that looked miraculous? If our senses are natural, our own ‘testimony’ is subject to the same objection Hume presents – a rational person would believe the weight of evidence that our senses are more likely to be wrong than the natural laws, as currently understood. I confess I wonder if I could dismiss such an experience, if it is particularly convincing! After all, many have been  hoodwinked by charlatans and I'm just as susceptible to a vivid experience as the next man. Ultimately, though, unless there is some unambiguous way to distinguish natural from supernatural events, one should dismiss miracle claims and even miracle experiences.

This has obvious ramifications for many theistic claims. Consider the resurrection; many theists have made great efforts to establish the historicity of the resurrection. At this remove, Hume's argument would suggest that this is impossible on empirical grounds, and it's hard not to disagree. Note that theists are not obliged then to not believe in the resurrection; if they think they have non-empirical grounds, or a method to distinguish natural from supernatural experience, this is open to them. But unless theists can remove testimony from the realm of the a posteriori, miracle testimony should never be used as the justification for miracle belief, even by them.


Chappell, T. (2011) The Philosophy of Religion , Milton Keynes, The Open University.

Cottingham, J. (ed.) (2008) Western Philosophy: An Anthology, Oxford, Blackwell Pub..

Keener, Craig S.. Miracles: The Credibility of the New Testament Accounts. Baker Publishing Group. Kindle Edition. 

Mackie, J.L. (1982) The Miracle of Theism, Oxford, Oxford University Press.


Post a Comment