Time to reconsider the relationship between science and the supernatural. A number of colleagues in both science and philosophy argue that the supernatural is nothing special, that god-related hypotheses can be tested by ordinary scientific methods, and that — given the repeated failure of such tests — the only rational conclusion is that science has pretty much shown that there is no such thing as the supernatural.

I am a bit skeptical of this sort of sweeping statement, on two grounds: I think the power of scientific investigation is significantly more limited than the above mentioned colleagues seem to admit; and I think the ideas about god(s) and the supernatural are so confused and borderline incoherent that to raise them to the level of a scientifically testable hypothesis grants them far too much.

Before proceeding, let me state clearly my position: I agree that specific claims made by supernaturalists can and have been shown to be false on empirical grounds.

The obvious example is the idea that the earth is only a few thousand years old. But what I maintain is that even in such obvious cases this does not amount to a scientific rejection of the supernatural, for the simple reason that there is no coherent and precise enough relationship between the specific claim (the earth is young) and the general idea (there is a god) — contra, of course, the situation for actual scientific hypotheses (say, the relationship between the general theory of relativity and its prediction that gravitational fields bend light by a certain, precise, degree).

For instance, a number of young earth creationists reject the empirical disproof of their claims on the grounds that god made it appear as if the earth is old, but that this is really a test of our faith. Crazy, I know, but that is exactly where such ideas belong — to the dustbin of completely ludicrous notions — quite regardless of whatever science may have to say about them.

Think of the creationist’s strategy as a comically inflated version of the Duhem-Quine thesis against falsification of ordinary scientific hypotheses: Pierre Duhem pre-empted Popper’s famous analysis of scientific progress in terms of falsifiability of theories on the grounds that scientists usually do not, in fact, discard a theory as soon as its predictions do not match the data. Rather, they look first to the ancillary hypotheses that go into the test itself, such as whether the instrumentation was working correctly, whether the data were analyzed properly, and so on.

W.V.O. Quine made the same point more broadly, suggesting that our knowledge of the world depends on a complex web of notions, any one of which may need to be re-evaluated and possibly discarded, if there are sufficient reasons to do so. Quine included logic itself in the number of notions that we may be forced to modify, though not many philosophers, I think, would go that far (especially not many logicians!). The point is that the supernaturalist’s web of belief and set of ancillary “hypotheses” is much, much wider, and much, much fuzzier, than the web scientists deploy when they evaluate their ideas about the world.

So let’s examine a couple of specific examples that I hope will clearly highlight the limits of scientific investigations of the supernatural: the (alleged) virgin birth of Jesus and the Catholic doctrine of transubstantiation. I will argue that while the first seems to be within the realm of empirical investigation, such investigation is not only not possible in practice, but likely not even in principle. As for the second case, I simply don’t see how any scientific analysis could possibly be performed that would settle the matter. Again, I don’t think of these as limitations of science, but rather as indicative of the fuzziness and possible incoherence of the idea of the supernatural.

Richard Dawkins (pp. 82-83 of The God Delusion) asks: “Did Jesus have a human father, or was his mother a virgin at the time of his birth? Whether or not there is enough surviving evidence to decide it, this is a strictly scientific question with a definite answer in principle: yes or no.” Let us set aside the quasi sophistic use of the locution “whether or not there is enough surviving evidence” (since Dawkins damn well knows that there is no evidence at all), it would seem prima facie that he is right: either Jesus was born of a virgin mother via supernatural means or he had a much more earthly fatherly origin.

How would anyone test such a claim, assuming for the sake of argument that we could go back in time and collect all the evidence we want? Easy: just sample Jesus’, Mary’s and Joseph’s DNA and compare their profiles, right? Not even close. You see, that sort of approach works excellently well to establish natural paternity, but we don’t know what a supernatural paternity entails in terms of biological traces (i.e., the concept itself is hopelessly vague). The point will be even more, painfully, I would say, clear in a moment, when we turn to transubstantiation, but it should be obvious even now. A DNA test (or whatever else Dawkins would like to do in this case) assumes a number of notions about biology and physics, which are the very notions that — by definition — are being violated when there is a miracle [1].

David Hume famously pointed out that rational-empirical investigations of any sort, including scientific ones, are possible only under a certain number of assumptions, the major one being that nature does not behave capriciously. If it did, all bets would be off and we wouldn’t even know where or what to look for. But all bets are off when we are talking about miracles, since that’s what miracles are: violations of the continuity of nature’s operations.

Indeed, in his famous essay Of Miracles, Hume advanced the argument that the reason we shouldn’t believe in them is because no testimony could ever be sufficient to establish a violation of the laws of nature against the alternative hypotheses that there has been fraud or a mistake (the argument can and has been made rigorous within a Bayesian framework):
When anyone tells me, that he saw a dead man restored to life, I immediately consider with myself, whether it be more probable, that this person should either deceive or be deceived, or that the fact, which he relates, should have really happened. I weigh the one miracle against the other; and according to the superiority, which I discover, I pronounce my decision, and always reject the greater miracle. If the falsehood of the testimony would be more miraculous, than the event which he relates; then, and not till then, can he pretend to command my belief or opinion.
So Dawkins is right in dismissing the possibility of a supernatural conception of Jesus, but not because the hypothesis is scientifically testable — regardless of the actual availability of evidence. It is because we have no good reason, and plenty of contrary ones, to subscribe to the very notion of miracles itself. This, needless to say, is a philosophical, not a scientific, argument [2].

Let us now turn to the idea of transubstantiation, which is an “allied dogma” to the central dogma referred to by Catholics as the Fact of the Real Presence (whatever). This is, of course, the idea that during the sacrament of the Eucharist the substance of both wine and bread are not just metaphorically, but literally, the flesh and blood of Christ, even though — and this is the clincher — all that our senses can actually approach is the appearance of things, i.e. bread and wine.

The doctrine has been around at least since the 11th century, but here is how the fourth Lateran Council put the matter in 1215: “His body and blood are truly contained in the sacrament of the altar under the forms of bread and wine, the bread and wine having been transubstantiated, by God’s power, into his body and blood.” Not clear enough? Well, then, how about the definition given by the Council of Trent in 1551: “that wonderful and singular conversion of the whole substance of the bread into the Body, and of the whole substance of the wine into the Blood — the species only of the bread and wine remaining — which conversion indeed the Catholic Church most aptly calls Transubstantiation.”

What are we to make of that? The idea of transubstantiation is, I suggest, simply incoherent. But incoherent with what? Well, at the very least with the laws of physics (and biology) as we know them. One can simply not make sense of the claim that something is at the same time both blood and wine, or both flesh and bread and still talk about blood, wine, flesh and bread as made of molecules and the corresponding subatomic particles. More fundamentally, the idea seems to violate the law of identity in basic logic: A is A, and therefore A is not ~A.

Logically incoherent concepts do not need to be investigated empirically: we know that they must be false, if we wish to rely on logic at all (pace Quine). But let’s say that Dawkins wanted to prove on scientific grounds that transubstantiation does not actually happen. How on earth would he do that? He cannot simply get a piece of bread and a sample of wine during a Eucharist ceremony, analyze them chemically and then triumphantly say to the world: “See? There is no blood or flesh here! So there.” That would move precisely no fervent Catholic at all, nor should it. Indeed, the scientist who insisted in making such a move would make a fool of himself in a way very much parallel to the famous episode of Samuel Johnson “refuting” George Berkeley’s idealist doctrine: the latter maintained that matter does not exist, it only appears to exist (sounds familiar?). To which Johnson (in a conversation with James Boswell) replied by kicking a nearby stone and smugly concluding “I refute it thus!” The joke, to this day, is of course on Johnson, who simply did not understand Berkeley’s idealism, a doctrine that — much like transubstantiation — is simply immune from any conceivable empirical disproof. In neither case, however, should we conclude that this built-in resistance to falsification is a virtue: Berkeley’s idealism is conceptually possible, but not that interesting; transubstantiation is conceptually incoherent and therefore not even wrong.

Examples such as the ones above could easily be multiplied ad nauseam. Take “research” on the effectiveness of intercessory prayer, for instance. Not surprisingly, the results have been negative. But this says nothing about the existence of a god who answers prayers, for at least two obvious reasons: first, if you were that god and you saw that a bunch of earthlings had the audacity to try to “test” you in a controlled experiment, wouldn’t you simply refuse to play along to teach those mortals some respect? I mean, the hubris of putting god under the microscope! Second, as my Catholic friends often remind me: god answers all prayers, it’s just that some time (most of the times?) the answer is “no.” Talk about the ultimate unfalsifiable hypothesis!

Ah, but what if we did get positive results, say from the intercessory prayers experiments? Wouldn’t that be proof positive that the supernaturalists are right? I’m not so sure. As scientists, we would want to know how such a thing is possible within the Humean conceptual framework for science: naturalism. Accordingly, we would first check the reliability and repeatability of the results and methods deployed in the experiment; then we would double check the data analyses; then we would attempt to eliminate any possibility of fraud. And then? Well, at one point we may have to admit either that there is a strange natural phenomenon of unknown origin, or that there is some intelligence at play. But even at that point, nothing would compel us to admit to the supernatural: remember Arthur C. Clarke’s famous third law: “Any sufficiently advanced technology is indistinguishable from magic.” Or watch again this beautifully on target episode of Star Trek — The Next Generation, and watch Captain Picard, in perfect Humean fashion, unmask the devil herself [3].


[1] The savvy reader could reasonably ask, at this point: what about paranormal phenomena? Should we dismiss them also as being outside of the scientific purview? The answer there is: it depends. If the paranormalist claims that we are dealing with a lawful, if mysterious, natural phenomenon (say, the existence of Nessie, or the ability to read other people’s minds), then we can certainly carry out meaningful scientific studies of it. But the more said phenomenon begins to acquire supernatural connotations (ghosts come to mind), the less we can (or need to, really) say about it on scientific grounds. And yes, to make things more interesting, I have just outlined a continuum, not a sharp natural/supernatural dichotomy.

[2] When successful, philosophical arguments simply preempt — as in make unnecessary — scientific ones. Just as showing something to be logically impossible makes it superfluous to show that it is also empirically untenable.

[3] Of course, at some point Picard would have to admit that it becomes increasingly unreasonable to deny the supernatural, if the devil keeps defying his attempts at naturalistic explanations. But even at that point, there is no science that the Enterprise could deploy in order to further understand the phenomenon. Because science assumes the continuity of nature and the understandability of its laws. Hume docet, as usual.

Reprinted from Rationally Speaking, July 19th 2013