One story about what distinguishes science from some other human pursuits, is that scientific theories must in principle be falsifiable. That is, there must be some way that it could be possible to show via observation that they are false. This story is very popular among scientists, and also popular among the kind of people who claim to very much love science.

The noted mathematical physicist Michiu Kaku put it this way:

Sometimes it is wielded as a weapon against theism, as Christopher Hitchens put it specifically in the case of Islamic belief, and elsewhere has applied it to theists more generally:

If we wanted to use this as an argument against theism, it could go something like this:

Unfortunately for our arguer, everything apart from ‘C1’ is either plausibly or probably false, and it only takes one weak link for the chain to snap. The argument has, you could say, been falsified.

Is Science Falsifiable? (P1)

The claim that a theory’s capacity to be falsified is what distinguishes science from non-science was made by Karl Popper, and has since become very popular among scientists – at least as a theoretical account of what science is about, if less in the actual practice of science. This has broken out in recent years with a highly public dispute between leading scientists over the nature of string theory, with its attendant multiverses. Leading cosmologists George Ellis and Joe Silk caution against dropping the requirement that scientific theories be falsifiable; in doing so, they critique multiverse theories, ‘many worlds’ interpretations of quantum mechanics, and speculations about what preceded the Big Bang. On the other side, Sean Carroll, Leonard Susskind and others argue that getting at the truth about the world may require going beyond falsifiability. This is a fascinating dispute from a few angles, including the fact that Ellis is a theist and adopting a view formerly beloved by sceptics (with the support, e.g. of arch-sceptic Jerry Coyne), while the others are atheists and adopting a position more typically associated with theists. It’s clearly a messy debate with more than two sides.

Later in life Popper nuanced his own account, seeing that some genuinely scientific theories are to some extent shielded from falsification, and that there is a legitimate place for verification in science. The first point remains one of the major critiques of falsificationism – that theories don’t come as single entities, but rather they’re clustered together, with each hypothesis typically associated with a few auxiliary hypotheses. What looks like a falsification for a hypothesis might actually be an indication that one of the auxiliary hypotheses was wrong, and maybe only a slight change is needed to that to bring the whole set in line with the data again. Any evolutionary hypothesis, for instance, is typically embedded within a complex web of other claims about fossils, biogeography, and the similarities between species in terms of morphology, proteins, and DNA. Aside from falsification being difficult or perhaps impossible in most cases, science does include clear cases of verification, or evidence for a theory counting in its favour. For instance the confirmation that the planet Mercury’s orbit fit the expectations of Einstein’s theory of general relativity was evidence in favour of that theory, and the X-ray diffraction patterns from Rosalind Franklin were evidence in favour of the double helix model of DNA’s structure.

If falsifiability is to still be considered a pre-requisite for science, it will only work if we acknowledge that actually falsifying a theory is seldom a simple matter of a single experiment. An example from molecular genetics is the ongoing dispute between neutral theory and adaptationism. The general consensus among biologists seems to be that neutral theory and its successor nearly-neutral theory demonstrated the falsehood of adaptationism, and evolutionary theory overall was happily updated. But, in reality amongst experts the fight continues to this day, with an influential opinion piece in one of the leading journals in the field recently arguing that neutral theory completely failed in its aims. Responses and counter-responses are expected to continue into the foreseeable future, and to me it seems that falsification won’t be of much relevance in the debate. Instead, we’ll want to see evidence for the action of natural selection or neutral evolutionary processes, and measures of how much of the genome is affected, in different species under different conditions.

If it’s Not Science, it’s Not Rational? (P3)

In response to this claim, I’ll just list a few kinds of beliefs which you probably have, and believe to be rational, and which are not scientific – as good a summary as any is given by William Lane Craig, in the most famous part of his classic debate with Peter Atkins. First, a couple which are not scientific: Historical beliefs, for instance that Alexander the Great conquered parts of India. Beliefs based on observation and/or memory, such as that you had breakfast this morning.  And a few which are neither scientific nor, as far as I can see, falsifiable: Ethical beliefs, for instance that killing children for fun is wrong. Mathematical beliefs, for instance that 1+2=3. Metaphysical beliefs, e.g. about causation. And finally, the belief that science is concerned with truth rather than just with building empirically adequate models of the world. None of these are scientific, but all are able to be held rationally.

Is Theism Falsifiable? (P2)

One of the main proponents of the view that theism is not falsifiable, and its rationality is therefore placed in doubt, was Antony Flew (the leading 20th Century atheist who infamously converted to deism in the early 2000s). Incidentally the “it’s not falsifiable” charge is one of many ways in which the ‘new atheists’ parallel Flew’s thought and why the reasons for his conversion should be of significant interest to them. The short essay “Theology and Falsification”, which is part of a discussion well worth reading, gives a taste of his view – talk of God is, in his views at that time, infinitely flexible or open to ambiguities.

There are concepts of God for which this objection is probably fair, but I think it does not apply to the Christian God. What we’re really talking about in the falsifiability debate is whether there are empirical findings which make it more probable or less probable that God exists. Is there some empirical handle with which we could begin to answer the God question? Any proponent of standard atheistic arguments such as the evidential problem of evil, or the existence of divine hiddenness or non-resistant non-belief, will say yes. To accept any such arguments as legitimate responses to theism or Christian theism is to hold that it is in principle falsifiable. Similarly, anyone who holds any of the standard arguments for God’s existence, e.g. from the existence or orderliness of the universe, believes that empirical investigation is relevant to the God question. And, anyone who believes that historical evidence is relevant to our belief that Jesus died and/or rose again also believes that there could be evidence against Christian theism.

In Conclusion: Theism, Falsifiability, and Rationality

Some theists (including me) are very open to empirical evidence being relevant to the rationality of theism. An early leader in the field of criteria of theory choice was Richard Swinburne, particularly recognized as an expert on the criterion of simplicity. Amongst many other works in this area he has edited a leading overview of Bayes Theorem for OUP. He puts both the criterion of simplicity and Bayes theorem to good use in developing a cumulative argument for the existence of God (hopefully both the friar William of Ockham and the Rev. Thomas Bayes would approve). As such, he holds that the God question is open to empirical lines of evidence, and therefore to a kind of falsification in principle, but not of a simplistic “one experiment is enough” version. Perhaps a more helpful way to put it is that the merits of competing theories in both science and religion can be contrasted based on the evidence.

Some, I think most, atheists are also very open to empirical evidence being relevant. The late physicist Victor Stenger, for instance, held that God is a falsifiable hypothesis, and has in fact been falsified, and his work in responding to theistic arguments from cosmology was very popular. Other leading atheists such as Sean Carroll think that falsifiability is not even important in science. It would be good to see how they think this carries over into the realm of theistic claims. As Carroll defends confirmation theory as an alternative to falsificationism, I think he should be open to this approach being used in the theism-atheism discussion.

Other theists are more sceptical of the relevance of empirical evidence to the God question. Theistic commitment could also be a rational decision on pragmatic grounds (“Pascal’s Wager”), or theism could be held as as a properly basic belief (let’s call that “Plantinga’s Warrant”). Some recent advocates of a view that empirical evidence is not very pertinent to the ‘does God exist?’ question are the authors of an important new book on science and faith, ‘It Keeps me Seeking’. I disagree with their approach, but it is also worth considering; and in the process they end up discussing a lot of empirical evidence from cosmology, biology, and history anyway.

Whichever of these camps you fall into, on reflection you should agree that the claimed unfalsifiability of theism is not the problem that the popular atheist slogans make it out to be – either because theism is in some sense falsifiable, or because its rationality is grounded in another way as with many other claims we rationally believe. It seems to me however that at least Christian theism does make empirical claims, which are open to empirical evidence. As generic theism is of relatively little interest, most of us should especially be able to agree on the crucial relevance of the empirical questions concerning Jesus’s life, death, and resurrection.