Sunday, February 6, 2011

Arguments to and from Miracles (Part 2) - Two Arguments against Violation Miracles

This is the second part of my series on Chapter 6 of Nicholas Everitt’s book The Non-Existence of God. It also forms part of my series on Miracles.

Last time out, I covered Everitt’s interpretation of Hume’s argument against the rational acceptability of belief in violation miracles. A violation miracle is said to occur whenever some law of nature has been violated or transgressed.

Everitt would seem to summarise Hume’s argument as follows:
There is a complementary relationship between the evidence for the occurrence of a violation miracle and the evidence for the existence particular law of nature such that any evidence that was sufficiently strong to convince you that a violation had occurred would also be sufficiently strong to cast doubt over the existence of the relevant law of nature.
I won’t comment any further on this argument for the time being.

In addition to offering this interpretation of Hume, Everitt presents two further arguments against the possibility of violation miracles. These arguments count against the a priori possibility of violation miracles, not just against their rational acceptability. We will cover each of these arguments in this post.

To understand the two arguments, we first need to agree that any putative law of nature must possess the following feature:

  • Universality: For X to count as a law of nature, it must say something about all or every member of a class, or about what always happens when something else happens; or, alternatively, it must say something about no members of a class or about what never happens.

Examples of laws of nature that seem to satisfy this criterion could include something like “Water always boils at 100 ° C at normal pressure” or “No iron can be turned into gold.”

1. The Incompatibility Argument
The first argument against the possibility of a violation argument rests on what appears to be an innocent, unassuming and uncontroversial logical principle, namely:

  • What is incompatible with a truth must itself be false.

To see why this is uncontroversial, consider the following abstract example. Suppose there is statement “All As are Bs” and this statement is true. This statement implies a second statement “no As are not Bs”. Jointly these statements imply that if someone comes along claiming to have found an A that is not a B, what they say must be false.

This creates a problem for the proponent of violation miracles. Suppose it is alleged that the statement “no dead man can live again” is a law of nature. If this is true, then it implies that no living man could have been dead. Thus, anyone who claims that a dead man is now living must be saying something that is false.

2. Assessment of the Incompatibility Argument
On the face of it, this is an extremely simple objection to the violation sense of miracles. Indeed, it is so simple that some of the most sophisticated thinkers have dismissed it. Everitt considers some examples.

The first comes from the writings of J.L. Mackie who seems to accept that laws of nature must satisfy the criterion of universality but who has also said:
“An occasional violation [of a law of nature] does not in itself necessarily overthrow the independently established conclusion that this is a [working] law [of nature].”
This might be an accurate reflection of our psychological reaction to anomalies in our worldviews, but as a matter of logic Everitt thinks it is plain wrong. If laws of nature have the universalised form “All As are Bs” then it simply cannot be the case that there is a true statement of the form “An occasional A is not a B”.

A second example of the mistake comes from the work of Richard Swinburne. Swinburne employs what we might call the “exception-clause”-strategy in response to the incompatibility argument. In other words, he adds an exception to the law of nature that obviates the problem of logical incompatibility.

Specifically, Swinburne says that there can be non-repeatable exceptions to otherwise universal laws. This means that he thinks laws of nature look something like this:

  • Swinburnian Law of Nature = All As are Bs, unless there is a non-repeatable exception.

As Everitt points out, this does nothing to solve the problem. Why not? Because this is simply a new generalisation which embraces two types of case (i) All As bar one being Bs and (ii) One A being not-B. This implies that should we come across one A which is not B, we have no grounds for saying it is a violation of a law of nature - the reputed violation is already included in the definition of the law.

Look at it like this. Suppose the statement “All swans are white, except for the black ones in Australia” is taken to be a law of nature. Now ask yourself whether this implies that the presence of black swans in Australia is a violation of a law of nature. Obviously, it implies nothing of the sort.

3. The Conceptual Confusion Argument
No doubt many readers will find that previous argument to be suspect, so here’s a second one that calls attention to a potential confusion in the concept of a law of nature.

Everitt notes that there are two types of laws: prescriptive and descriptive. Prescriptive laws tell us what ought to be the case or what ought to be done; descriptive laws tell us how things are or how they will be. There is no obvious contradiction involved in asserting that a prescriptive laws has been violated; there is an obvious contradiction involved in asserting that a descriptive law has been violated. This is because an accurate descriptive law is supposed to provide an accurate description of how the world actually is.

The contrasts between the prescriptive and descriptive senses of the term law are summarised in the following picture.

This brings us to the second argument against violation miracles. The argument states that the claim that there can exist violations of laws of nature confuses the prescriptive and descriptive senses of the word “law”. It is simply not possible to say that a true description of the world can be violated, and since that is what laws of nature are, it is not possible for there to be violation miracles.

4. Assessment
Everitt thinks that this second argument is successful and, when combined with the other argument, offers a fairly decisive rebuttal to the proponent of violation miracles. I'm not entirely convinced, but in any event, as we saw when discussing Hume, the believer in miracles has a fairly simple response to these arguments: abandon the concept of violation miracles.

We’ll consider some of the alternatives in the next part.


  1. An internet gnome ate my response, so this is the second try:

    I adopt a Mackie (and, surprisingly, Swinburne) like position. I see the religious as claiming the world behaves in a two-tiered way: it usually follows certain regularities which we call, in practice, "laws of nature", but occasionally there are exceptions to these laws which we call "miracles". These exceptions may be rather common - for example, perhaps every Mormon is much healthier than he should be naturally, but that's still a rather rare exception in Nature considering its vast size and variety.

    The arguments Everitt raises are irrelevant to this conception of a "law of nature". Saying "[Usually] dead people stay dead" doesn't rule out analyzing when they don't and drawing conclusions from that. Saying "The laws of nature are violated" does not imply nature does not behave the way it does, it merely says it does not behave in the way it usually does.

    The two traditional interpretations of Hume you presented - that Hume says miracles are impossible or that accepting one implies changing our understanding of what the laws of nature are - don't work for this understanding of "law of nature" either. However, a Humean argument can still be marshaled that an established law of nature, by definition, has lots of evidence going for it and, by the Principle of Uniformity underlying Humean (and empirical) epistemology, the evidence that needs to be brought up against its universal validity should thus be equally enormous. This is, I think, the gist of Hume's in-principle argument against miracles, and I think it holds.

    I'd further note that the laws of physics we have in practice are uniform universal laws, applying at level well below the human person. If exceptions to these rules truly exist on the human person level (such as laws starting with "All Mormons..."), it would be eminently useful and efficient to phrase the description of nature in terms of the universal uniform laws and their exceptions. That all phenomena rigorously examined can be better explained by appealing to the uniform laws of nature is a testament to the strength of Hume's argument in my (not so humble?) opinion.


  2. Hi Yair,

    First off, I don't think Everitt's discussion is particularly great. But I included since I am doing a series on his book and thought it would be efficient to overlap it with my series on miracles.

    Second, on the Swinburnian definition of a law of nature. I'll be going into this later, but may as well say it now. Swinburne defines a miracle as a non-repeatable counter-instance to a law of nature. In some of his writings, he actually seems to agree with Everitt here in saying that introducing an exception-clause would mean that you didn't have a violation. But he counters this by saying that sometimes introducing the exception-clause is ad-hoc and sacrifices explanatory simplicity (something which Swinburne loves). He then argues that it is in these cases -- i.e. where an exception-clause would undermine simplicity -- that we are justified in calling the event a miracle.

    Morgan Luck discusses this in his article on Swinburne's arguments and goes on to say that this definition sits well with David Lewis's systematic theory of laws of nature. I'll cover this when I get round to Luck's article (will be a few weeks though).

    Third, on the interpretation of Hume's argument. I don't think I said, and certainly hope I didn't convey the impression that I said, that Hume says miracles are impossible. They are just difficult to believe in. As for your other points, the Bayesian interpretation of Hume's argument does help to clarify some of the issues concerning the evidential burden that must be met in these cases. I think the consensus position -- and the one with which I currently agree -- is that the burden is high (although this depends to some extent on your assumptions) but not insurmountable.